UNIVERSITI PUTRA MALAYSIA UPMpsasir.upm.edu.my/id/eprint/71397/1/FK 2018 91 IR.pdf · 2019. 9. 3. · dengan jelas dan tepat, teknik pemodelan yang betul harus digunakan. Kajian ini

HT UPM

UNIVERSITI PUTRA MALAYSIA

MODELING OF ROAD GEOMETRY AND TRAFFIC ACCIDENTS BY HIERARCHICAL OBJECT-BASED AND DEEP LEARNING METHODS

USING LASER SCANNING DATA

MAHER IBRAHIM SAMEEN

FK 2018 91

HT UPM

MODELING OF ROAD GEOMETRY AND TRAFFIC ACCIDENTS BY

HIERARCHICAL OBJECT-BASED AND DEEP LEARNING METHODS

USING LASER SCANNING DATA

MAHER IBRAHIM SAMEEN

Thesis Submitted to the School of Graduate Studies, Universiti Putra Malaysia,

in Fulfillment of the Requirements for the Degree of Doctor of Philosophy

March 2018

HT UPM

All material contained within the thesis, including without limitation text, logos, icons,

photographs, and all other artwork, is copyright material of Universiti Putra Malaysia

unless otherwise stated. Use may be made of any material contained within the thesis

for non-commercial purposes from the copyright holder. Commercial use of material

may only be made with the express, prior, written permission of Universiti Putra

Malaysia.

HT UPM

Abstract of thesis presented to the Senate of Universiti Putra Malaysia in fulfilment

of the requirement for the degree of Doctor of Philosophy

MODELING OF ROAD GEOMETRY AND TRAFFIC ACCIDENTS BY

HIERARCHICAL OBJECT-BASED AND DEEP LEARNING METHODS

USING LASER SCANNING DATA

MAHER IBRAHIM SAMEEN

March 2018

Chairman : Professor Biswajeet Pradhan, PhD

Faculty : Engineering

Road traffic accidents are global concerns since they affect human life, economy, and

road transportation systems. Rapid information acquisition and insight discovery are

key tasks in transportation management. Specifically, extraction of geometric road

features such as slopes and superelevation are essential information to understand the

effects of road geometry on road traffic accidents. However, to understand these

effects clearly and accurately, proper modeling techniques should be used. This study

aims to develop methods to extract geometric road features (e.g., vertical gradients,

superelevation, width, design speed) and establish associations between those features

and road traffic accidents including frequency and accident severity. There was a need

for efficient segmentation algorithm, optimization strategy, feature extraction and

classification, and robust statistical and computational intelligence models to

accomplish the set aims. Experimental results regarding road geometry extraction

indicated that the proposed methods could achieve relatively high accuracy (~ 85% -

User’s Accuracy) of road detection from airborne laser scanning data. Our method

improved the overall accuracy of classification by 7% outperforming the supervised

𝑘 nearest neighbour method. In addition, the results also showed that the proposed

hierarchical classification method could extract geometric road elements with an

average error rate of 6.25% for slope parameter and 6.65% for superelevation

parameter, and it is transferable to other regions of similar environments. On the other

hand, the geometric regression model predicted the number of accidents in the North-

South Expressway with a reasonable accuracy (R2 = 0.64). This model also could

identify the most influential factors contributing to the number of accidents.

Experiments on deep learning models showed that the recurrent neural network

performs better than the feedforward neural networks, statistical bayesian logistic

regression, and convolutional neural networks. This study also suggests that transfer

learning could improve the forecasting accuracy of the injury severity by nearly 10%.

HT UPM

Abstrak tesis yang dikemukakan kepada Senat Universiti Putra Malaysia sebagai

memenuhi keperluan untuk ijazah Doktor Falsafah

PERMODELAN GEOMETRI JALAN DAN KEMALANGAN LALULINTAS

DENGAN OBJEK HIRAEKI DAN KAEDAH DEEP LEARNING

MENGGUNAKAN IMBASAN DATA LASER

Oleh

MAHER IBRAHIM SAMEEN

Mac 2018

Pengerusi : Profesor Biswajeet Pradhan, PhD

Fakulti : Kejuruteraan

Kemalangan jalan raya adalah kebimbangan global kerana ia mempengaruhi

kehidupan manusia, ekonomi, dan sistem pengangkutan jalan raya. Pemerolehan

maklumat yang cepat dan penemuan wawasan adalah tugas utama dalam pengurusan

pengangkutan. Khususnya, pengekstrakan ciri-ciri jalan geometri seperti cerun dan

penyempurnaan adalah maklumat penting untuk memahami kesan geometri jalan raya

pada kemalangan jalan raya. Walau bagaimanapun, untuk memahami kesan-kesan ini

dengan jelas dan tepat, teknik pemodelan yang betul harus digunakan. Kajian ini

bertujuan untuk membangunkan kaedah untuk mengekstrak ciri-ciri jalan geometri

(cth., Kecerunan menegak, superelevasi, lebar, kelajuan reka bentuk) dan

mewujudkan persatuan antara ciri-ciri dan kemalangan jalan raya termasuk kekerapan

dan keterukan kemalangan. Terdapat keperluan untuk algoritma segmentasi yang

berkesan, strategi pengoptimuman, pengekstrakan dan klasifikasi ciri, dan model

perisikan statistik dan komputasi yang mantap untuk mencapai matlamat yang

ditetapkan. Hasil eksperimen mengenai pengekstrakan geometri jalan menunjukkan

bahawa kaedah yang dicadangkan dapat mencapai ketepatan yang agak tinggi (~ 85%

- Akurasi Pengguna) pengesanan jalan dari data pengimbasan laser udara. Kaedah

kami meningkatkan ketepatan keseluruhan klasifikasi sebanyak 7% yang melebihi

kaedah jiran terdekat yang diselia. Di samping itu, keputusan juga menunjukkan

bahawa kaedah klasifikasi hierarki yang dicadangkan dapat mengekstrak unsur jalan

geometri dengan kadar kesilapan purata 6.25% untuk parameter cerun dan 6.65%

untuk parameter superelevasi, dan ia boleh dipindahkan ke kawasan lain yang serupa

persekitaran. Sebaliknya, model regresi geometri meramalkan bilangan kemalangan

di Lebuhraya Utara-Selatan dengan ketepatan yang munasabah (R2 = 0.64). Model ini

juga dapat mengenal pasti faktor-faktor yang paling berpengaruh yang menyumbang

kepada bilangan kemalangan. Eksperimen dalam model pembelajaran mendalam

menunjukkan bahawa rangkaian neural berulang lebih baik daripada rangkaian saraf

HT UPM

iii

feedforward, regresi logistik bayesian statistik, dan rangkaian saraf convolutional.

Kajian ini juga menunjukkan bahawa pemindahan pembelajaran dapat meningkatkan

ketepatan ramalan keterukan cedera oleh hampir 10%.

HT UPM

ACKNOWLEDGEMENTS

I praise ALLAH for his great loving generosity, that has brought all of us to encourage

and tell each other and who has pulled us from the darkness to the light. All respect

for our Holy Prophet Muhammad (Peace be upon him), who guided us to identify our

creator.

I am very grateful for the support of my family and relatives – for their constant

inspiration and encouragement. First, my father Ibrahim Sameen - for his moral and

financial support, and my sisters - for their curious enthusiasm. Second, my heartfelt

thanks to my fiancee for her moral support, continues help and for standing by my

side. Third, I also thank my aunts Sarah and Sabiha – for their support and their role

in getting the work of this thesis done, my uncle, Abdullah, and all my other relatives

– for their help and support in diverse issues regarding life and study.

Finally, I also want to say thanks to all my dear friends - in Iraq and Malaysia, for their

understanding and interest, for helping me to enjoy my life besides work and study.

I also take this occasion to express my deep acknowledgment and profound regards to

my guide Prof Dr. Biswajeet Pradhan for his ideal guidance, monitoring and

continuous motivation during this thesis. The help, blessing, and guidance offered by

him from time to time will support me a long way in the life journey on which I am

about to embark. He formed an atmosphere that motivated innovation and shared his

remarkable experiences throughout the work. Without his constant encouragement, it

would have been impossible for me to finish this research.

I acknowledge my committee Assoc. Dr. Helmi Shafri and Assoc. Dr. Hussain Bin

Hamid, for the valuable information provided by them in their respective fields. I am

grateful for their cooperation.

HT UPM

This thesis was submitted to the Senate of the Universiti Putra Malaysia and has been

accepted as fulfilment of the requirement for the degree of Doctor of Philosophy. The

members of the Supervisory Committee were as follows:

Biswajeet Pradhan, PhD

Professor

Faculty of Engineering

Universiti Putra Malaysia

(Chairman)

Helmi Shafri, PhD

Associate Professor

Faculty of Engineering

Universiti Putra Malaysia

(Member)

Hussain Bin Hamid, PhD

Associate Professor

Faculty of Engineering

Universiti Putra Malaysia

(Member)

ROBIAH BINTI YUNUS, PhD

Professor and Dean

School of Graduate Studies

Universiti Putra Malaysia

Date:

HT UPM

vii

Declaration by graduate student

I hereby confirm that:

this thesis is my original work;

quotations, illustrations and citations have been duly referenced;

this thesis has not been submitted previously or concurrently for any other degree

at any institutions;

intellectual property from the thesis and copyright of thesis are fully-owned by

Universiti Putra Malaysia, as according to the Universiti Putra Malaysia

(Research) Rules 2012;

written permission must be obtained from supervisor and the office of Deputy

Vice-Chancellor (Research and innovation) before thesis is published (in the form

of written, printed or in electronic form) including books, journals, modules,

proceedings, popular writings, seminar papers, manuscripts, posters, reports,

lecture notes, learning modules or any other materials as stated in the Universit i

Putra Malaysia (Research) Rules 2012;

there is no plagiarism or data falsification/fabrication in the thesis, and scholarly

integrity is upheld as according to the Universiti Putra Malaysia (Graduate

Studies) Rules 2003 (Revision 2012-2013) and the Universiti Putra Malaysia

(Research) Rules 2012. The thesis has undergone plagiarism detection software

Signature: _______________________ Date: __________________

Name and Matric No.: Maher Ibrahim Sameen, GS44022

HT UPM

viii

Declaration by Members of Supervisory Committee

This is to confirm that:

the research conducted and the writing of this thesis was under our supervision;

supervision responsibilities as stated in the Universiti Putra Malaysia (Graduate

Studies) Rules 2003 (Revision 2012-2013) were adhered to.

Signature:

Name of Chairman

of Supervisory

Committee:

Professor Dr. Biswajeet Pradhan

Signature:

Name of Member

of Supervisory

Committee:

Associate Professor Dr. Helmi Shafri

Signature:

Name of Member

of Supervisory

Committee:

Associate Professor Dr. Hussain Bin Hamid

HT UPM

TABLE OF CONTENTS

Page

ABSTRACT i

ABSTRAK ii

ACKNOWLEDGEMENTS iv

APPROVAL v

DECLARATION vii

LIST OF TABLES xiii

LIST OF FIGURES xvi

LIST OF ABBREVIATIONS xx

CHAPTER

1 INTRODUCTION 1

1.1 Background of Study 1 1.2 The Statement of Problem 3

1.3 Research Objectives 4 1.4 Research Questions 5

1.5 Thesis Contributions 5 1.6 Scope of Study 6

1.7 Thesis Organization 7

2 LITERATURE REVIEW 8

2.1 Introduction 8 2.2 Introduction to Laser Scanning Systems (LiDAR) 10

2.2.1 General 10 2.2.2 LiDAR 10

2.2.2.1 ALS 11 2.2.2.2 MLS 14

2.2.3 Comparison of AS, MLS for Road Extraction and

Modeling 17

2.3 Road Geometric Modeling 17 2.3.1 General 17

2.3.2 Road Geometric Model 18 2.3.3 Delineation of Road Geometric from LiDAR Data 20

2.3.3.1 Road Extraction Based on Types of

Sensors (LiDAR) 21

2.3.3.2 Classification on Road Extraction According

to the Preset Objective 22

2.3.3.3 Road Extraction Techniques 23 2.4 Road Traffic Accident Modeling 30

2.4.1 General 30 2.4.2 Road Accident Setting 30

HT UPM

2.4.3 Model Factors for Predicting Road Accident Frequency 31 2.4.4 Model Factors for Predicting Road Accident Severity 32

2.4.5 Frequency Modeling 33 2.4.6 Injury Severity Modeling 34

2.4.7 Comparison of NN and Statistical Models for Traffic

Accident Modeling 36

2.4.8 Use of NN in Road Accidents Prediction 38 2.4.9 Rationale of Using NN 40

2.4.10 Predictive Performance of Different Types of NN 41 2.5 Chapter Summary 43

2.6 Research Gaps 44

3 MATERIALS AND METHODOLOGY 46

3.1 Introduction 46 3.2 Overall Methodology 46

3.3 Experimental Test Sites 47 3.3.1 Test Area 1 - Universiti Putra Malaysia (UPM) 47

3.3.2 Test Area 2 – North-South Expressway (NSE) 49 3.4 Data 50

3.4.1 LiDAR 51 3.4.1.1 ALS Data 51

3.4.1.2 MLS Data 51 3.4.2 Traffic Accident Data 53

3.4.2.1 Accident Frequency Data 53 3.4.2.2 Accident Severity Data 54

3.4.3 Field Data 57 3.5 Data Pre-processing 58

3.5.1 LiDAR Data 58 3.5.1.1 Geometric Calibration and Outlier Removing 58

3.5.1.2 Generating DSM 58 3.5.1.3 Generating DEM 59

3.5.2 Traffic Accident Data 60 3.5.2.1 Data Transformation 60

3.5.2.2 Removing Missing Data and

Multicollinearity Assessment 60

3.6 Road Geometric Modeling Using a Hybrid ACO-OBIA Method 62 3.6.1 Overview 62

3.6.2 Segmentation 62 3.6.2.1 Overall Workflow 64

3.6.2.2 Identifying Segmentation Levels 65 3.6.2.3 Identifying Best Segmentation Parameters

by Taguchi Method 66 3.6.2.4 Segmentation Assessment 68

3.6.3 Feature Extraction and Classification 69 3.6.3.1 Background of ACO 69

3.6.3.2 ACO for Attribute Selection 70 3.6.3.3 Selection of Attribute Subsets 71

HT UPM

3.6.3.4 Ruleset Development with DT Algorithm 72 3.7 Road Geometric Modeling Using Hierarchical Feature

Extraction Approach 73 3.7.1 Overview 73

3.7.2 MS Segmentation 74 3.7.3 PSO Optimization 75

3.7.4 Classification of Segments Using SVM 76 3.7.5 Extraction of Road Geometric Parameters Using PCA 76

3.7.6 Extraction of Road Geometric Design Parameters 78 3.8 Modeling Traffic Accident Frequency Using GR 80

3.8.1 Geometric Regression (GR) 80 3.8.2 Road Segmentation 81

3.8.3 Modeling Accident Frequency Using GR 82 3.9 Modeling Accident Severity Using RNN 83

3.9.1 Network Architecture 83 3.9.2 Training Methodology 86

3.9.3 Mitigating Overfitting 87 3.9.4 Hyperparameter Tuning 87

3.10 Modeling Accident Severity Using TL Model 88 3.10.1 Transfer Learning (TL) 88

3.10.2 Network Architecture 89 3.10.3 Network Parameters 90

3.10.4 Hyperparameter Optimization 91 3.11 Summary 92

4 RESULTS AND DISCUSSION 93 4.1 Introduction 93

4.2 Implementation and Software 93 4.3 Results of Road Geometric Modeling in 2D 94

4.3.1 Results of Segmentation 94 4.3.2 Results of Feature Selection and Classification 98

4.3.2.1 Results of ACO 98 4.3.2.2 Comparison of ACO with Other Techniques 100

4.3.3 Results of Image Classification 102 4.3.4 Validation 103

4.3.5 Evaluation of ACO-based Rules Transferability 106 4.3.6 Discussion 107

4.4 Results of Road Geometric Modeling in 3D 110 4.4.1 Results of Hierarchical Classification 110

4.4.2 Accuracy and Model Transferability Assessments 113 4.4.3 Sensitivity Analysis 114

4.4.4 Results of Calculating NSE Design Parameters 115 4.5 Results Traffic Accident Frequency Modeling Using GR 117

4.5.1 Modeling Results 117 4.5.2 Discussion 122

4.6 Results of Accident Severity Modeling Using RNN 123 4.6.1 Results of RNN Model Performance 123

HT UPM

xii

4.6.2 Sensitivity Analysis of Optimization Algorithm 124 4.6.3 Sensitivity Analysis of Learning Rate and RNN

Sequence Length 125 4.6.4 Network Depth Analysis 126

4.6.5 Extraction Factor Contribution in the RNN Model 127 4.6.6 Comparative Experiment with Statistical Methods 128

4.6.7 Comparative Experiment with Other DL Models 130 4.6.7.1 Performance Evaluation 130

4.6.7.2 Optimization and Sensitivity Analysis 131 4.6.7.3 Discussion 133

4.6.8 Computational Complexity of the Model 135 4.6.9 Applicability and Limitations of the Proposed Method 135

4.7 Results of TL 136 4.7.1 Training from Scratch vs. TL 136

4.7.2 Effects of Batch Size 138 4.7.3 Effects of Data Transformation 139

4.7.4 Effects of Network Hyperparameters 139 4.7.5 Discussion 142

4.7.6 Time Complexity Analysis 143 4.7.7 Model Comparisons 143

4.7.8 Importance of Accident-Related Factors 144 4.8 Summary of Results 146

5 CONCLUSION AND FUTURE WORK RECOMMENDATIONS 149 5.1 General 149

5.2 Conclusion 149 5.3 Recommendation for Future Work 151

REFERENCES 153 APPENDICES 173

BIODATA OF STUDENT 181 LIST OF PUBLICATIONS 182

HT UPM

xiii

LIST OF TABLES

Table Page

2.1 Performance specifications of common ALS systems used for

road extraction and geometric road modeling in the earlier

studies

2.2 Performance comparison between ALS, MLS, and TLS for road

extraction and modeling [modified after Williams et al. (2013)]

2.3 The proposed classification of road extraction methods based on

three criteria identified by Mena (2003)

2.4 Advantages and disadvantages of published methods for

delineation of geometric road information from ALS data

2.5 Basic NN terminology and statistical equivalent (adapted from

Sarle (1994))

2.6 Summary of studies used NN for traffic accident modeling 39

3.1 Sets of data and parameters used in the study 50

3.2 Typical technical specification of Riegl VZ-2000 mobile scanner

(Source: Datasheet VZ-2000, 2015)

3.3 Sample of accident frequency data acquired from PLUS; the

detailed version of this table is in the appendix

3.4 Driver injury severity distribution according to accident-related

factors

3.5 Multicollinearity assessment for the accident frequency model

factors

3.6 Multicollinearity assessment among the accident related factors 61

3.7 Segmentation levels and their corresponding selected values of

user-defined parameters used to optimize image segmentation

process by Taguchi method

3.8 The L16 orthogonal array for segmentation level 1, created by

Taguchi design to conduct experiments for analyzing effects of

segmentation parameters on classification accuracy

HT UPM

xiv

3.9 The L16 orthogonal array for segmentation level 2, created by

Taguchi design to conduct experiments for analyzing effects of

segmentation parameters on classification accuracy

3.10 The L9 orthogonal array for segmentation level 3, created by

Taguchi design to conduct experiments for analyzing effects of

segmentation parameters on classification accuracy

3.11 Selected attributes from the available attributes for LiDAR data

classification

3.12 SVM hyperparameters that are tuned 75

3.13 Summary of characteristics of hom*ogeneous section variables

considered for model development

3.14 The optimized hyperparameters of the proposed RNN model 87

3.15 Hyperparameters and their descriptions of the proposed DL

model

3.16 Search space and the optimal values of the model

hyperparameters

4.1 Geometric quality assessment of segmentation process 98

4.2 Attributes selected by ACO and classification accuracy produced

by different percent subsets

100

4.3 Rulesets generated by DT algorithm using the best attribute

subset selected by ACO proposed for LiDAR data classification

102

4.4 Accuracy assessment of the proposed classification method for

LiDAR data and its comparison with a rule-based method with

full attribute sets-and supervised KNN approach

105

4.5 Accuracy assessment of the proposed classification method for

the second dataset

107

4.6 Results compared with field measured values based on MLS

point clouds for slope and superelevation parameters

113

4.7 Results of testing area compared with field measured values

based on MLS point clouds for slope and superelevation

parameters

114

4.8 Accuracy of our method against other methods in the literature 114

HT UPM

4.9 NSE design parameters calculated in AutoCAD based on

extracted road geometric model

116

4.10 Vertical curves with their characteristics of NSE study area 117

4.11 The GR model and the estimated coefficients 118

4.12 The analysis of variance and the estimated elasticities 118

4.13 Coefficients and elasticity estimates for the explanatory variables

based on fixed-length segments

122

4.14 The performance of different optimization methods evaluated in

this study

125

4.15 The training and validation accuracy of the proposed RNN model

with a different number of dense layers

126

4.16 The training and validation accuracy of the proposed RNN model

with a different number of LSTM layers

127

4.17 The calculated weights of accident-related factors 128

4.18 Performance comparison of the proposed RNN model with MLP

and BLR models

130

4.19 Calculated ranks of the accident-related factors in the RNN,

MLP, and BLR models

130

4.20 The average cross-validation accuracy of the proposed models 131

4.21 Average training and testing time per iteration of the proposed

model

135

4.22 Average training and testing time per iteration/ prediction of the

proposed model

143

4.23 The training and testing accuracies estimated for the proposed

model and its comparison with other ML methods

144

HT UPM

xvi

LIST OF FIGURES

Figure Page

2.1 General and the specific topics reviewed in this thesis 9

2.2 The main components of a typical ALS system 12

2.3 An example of ALS data, (a) LiDAR point clouds colored by

elevation attributes, low elevations are shown in blue colour

while the high elevations are indicated by red colour, (b) LiDAR

intensity data, low intensities in dark colour, and bright colour

shows high intensities, (c) true colour aerial orthophoto acquired

by the same airborne LiDAR system in the same data acquisition

mission

2.4 MLS system components (e.g., Topcon IP-S3 HD 3D Mobile

Mapping System)

2.5 An example of MLS point clouds acquired by a Riegl VZ-2000

mobile scanner and Nikon camera D800 (30 megapixels) for a

highway section

2.6 Basic geometric elements of road model, (a) a horizontal curve,

(b) superelevation, (c) vertical curves (g refers to road gradient

often in percent)

2.7 A simple sketch shows the typical cross-section of

roads/highways

3.1 The flowchart of the overall methodology 48

3.2 Location of UPM area 49

3.3 Location of NSE area 50

3.4 Operational elements of Riegl VZ-2000 and its georeferencing

concept, (a) the concept of direct georeferencing, (b) operational

elements and connectors of the device

3.5 Photos were taken at NSE site during MLS data collection 52

3.6 Sample from MLS data captured by Riegl VZ-2000 covers urban

road point clouds; points are color-coded by natural color of

digital photos

HT UPM

xvii

3.7 Typical vehicle types that involved in traffic accidents along

NSE, (a) motorcycle, (b) different vehicle types such as a car,

bus, van, and a heavy car

3.8 The overall method proposed for LiDAR data classification with

ACO and OBIA

3.9 The workflow of the proposed two-stage optimization strategy

for OBIA

3.10 ACO-based attribute selection workflow 70

3.11 Workflow of geospatial model based on a hierarchical strategy 74

3.12 Correspondence between PCA-based eigenvalues and

eigenvectors with vertical geometric parameters of the road

3.13 Segmentation in road cross-sections and parameter estimation

steps

3.14 Estimation of road geometry features, (a) the process of curve

fitting, (b) an example of attribute information about the road

segments and their design parameters, (3) a basic diagram of

vertical curves

3.15 An example of the factor “distance to the nearest access point”

used in the accident count modeling

3.16 Example for segregation of a road section into four hom*ogenous

segments based on values of the observed explanatory variables

3.17 The high-level architecture of the proposed RNN model 84

3.18 The proposed NN model for injury severity prediction of traffic

accidents

3.19 The proposed CNN model for injury severity prediction of traffic

accidents

3.20 An illustration of the principal goal of TL with DL (e.g., NN)

models

3.21 The architecture of the proposed model based on RNN and TL 90

4.1 Class separability distance estimated using BD algorithm used to

optimize the scale parameters

HT UPM

xviii

4.2 SNRs used to select the optimal scale, shape, and compactness

parameters for OBIA. The SNR measures how the response

varies relative to the target value under different noise

conditions. The negative values are due to the logarithmic

transformation used in SNR formula. The negative values

indicate not favourable parameter values for the given task

4.3 Land cover classes of the study area used for method

development (with their assigned segmentation levels)

4.4 Results of image segmentation obtained by MRS algorithm with

optimized parameters, (a) an example from NSE area, (b) UPM

area

4.5 Evaluation of the attribute subset selected by ACO using OA and

the kappa coefficient

4.6 Comparison of classification accuracies with selected attributes

by different methods, overall accuracies (left), kappa indices

(right)

101

4.7 Results of image classification using the best attribute subset and

rule-based method

103

4.8 Results of image classification with a rule-based method with

full attribute sets (a) and supervised KNN approach (b)

106

4.9 Results of image classification with ACO-optimized rule-based

method for NSE area

107

4.10 Example showing the extracted road surface (a) interpolated

DSM from MLS point clouds using the TIN-based method, (b)

extracted road surface, (c) an example on accurate road edges,

which were identified using the proposed method

111

4.11 Superelevation graphical diagram 112

4.12 Slope graphical diagram 112

4.13 Effects of sample size on SVM accuracy 115

4.14 Scatter plot of observed accidents versus those predicted by the

Bayesian logistic model

119

4.15 Accuracy performance and loss of the RNN model calculated for

100 epochs

124

HT UPM

xix

4.16 The sensitivity of the RNN model for different learning rate and

sequence length configurations

126

4.17 The accuracy of the models tested by a 10-fold cross-validation

method

131

4.18 Effects of optimization algorithm on the accuracy of the

proposed models

132

4.19 Effects of the batch size on the accuracy of the proposed models 133

4.20 Effects of the dropout rate on the CNN and RNN models 133

4.21 Variations of model loss (left) and model validation accuracy

(right) without TL

137

4.22 Variations of model loss (left) and model validation accuracy

(right) with TL

138

4.23 Effects of batch size on the validation accuracy of the model 139

4.24 Effects of different optimization methods on the validation

accuracy of the model

140

4.25 Effects of some hidden units in the LSTM and dense layers on

the validation accuracy of the model

141

4.26 Effects of learning rate and weight decay on the validation

accuracy of the model

142

4.27 A diagram shows the estimated coefficients for the accident-

related factors based on the proposed model with and without TL

146

HT UPM

LIST OF ABBREVIATIONS

LiDAR Light Detection And Ranging

ALS Airborne Laser Scanning

MLS Mobile Laser Scanning

TLS Terrestrial Laser Scanning

3D Three Dimension (s)

OBIA Object-Based Image Analysis

DEM Digital Elevation Model

DL Deep Learning

2D Two Dimension (s)

FOV Field of View

RNN Recurrent Neural Network (s)

TL Transfer Learning

IMU Inertial Measurement Unit

DSM Digital Surface Model

NN Neural Network (s)

UAV Unmanned Aerial Vehicle (s)

ACO Ant Colony Optimization

SVM Support Vector Machine (s)

GNSS Global Navigation Satellite System

GIS Geographic Information System

GR Geometric Regression

GPS Global Positioning System

PCA Principal Component Analysis

TOF Time of Flight

UPM Universiti Putra Malaysia

ML Machine Learning

HT UPM

xxi

LR Logistic Regression

CNN Convolutional Neural Network (s)

HD High Definition

NSE North-South Expressway

PSO Particle Swarm Optimization

LULC Land Use and Land Cover

EA Evolutionary Algorithm (s)

RF Random Forest

GLONASS Global Orbiting Navigation Satellite System

USGS United States Geological Survey

DTM Digital Terrain Model

MCC Multiscale Curvature Classification

EDM Electronic Distance Measurement

RMS Root Mean Square

PCD Phase Coded Disk

DT Decision Tree

SPOT Système Pour l'Observation de la Terre

MRS Multi-Resolution Segmentation

NDVI Normalized Difference Vegetation Index

GVF Gradient Vector Flow

RANSAC RANdom SAmple Consensus

CRF Conditional Random Field

CAD Computer-Aided Design

ETC Electronic Toll Collection

NB Negative Binomial

MA Moving Average

AR Autoregressive

HT UPM

xxii

KNN K-Nearest Neighbors

MLPNN Multilayer Perceptron Neural Networks

RBFNN Radial Basis Function Neural Networks

SAP Smeed Accident Prediction

SLPNN Single-Layer Perceptron Neural Network

GRNN General Regression Neural Network

MLR Multilayer Regression

GDP Gross Domestic Product

MSE Mean Square Error

ReLU Rectified Linear Unit

GA Genetic Algorithms

LSTM Long Short-Term Memory

GCPs Ground Control Points

PLUS Projek Lebuhraya Usaha Sama

VIF Variance Inflated Factor

DGPS Differential Global Positioning System

AADT Average Annual Daily Traffic

BD Bhattacharyya Distance

POF Plateau Objective Function

SNR Signal-To-Noise Ratio

MS Mean Shift

UA User’s Accuracy

ED Euclidean Distance

SGD Stochastic Gradient Descent

BPTT Backpropagation Through Time

OA Overall Accuracy

AIC Akaike Information Criterion

HT UPM

xxiii

APM Accident Prediction Model

MCMC Markov Chain Monte Carlo

BGR Brooks-Gelman-Rubin

DIC Deviance Information Criterion

PC Personal Computer

GPUs Graphics Processing Units

KIA Kappa Index of Agreement

BLR Bayesian Logistic Regression

HT UPM

CHAPTER 1

1 INTRODUCTION

1.1 Background of Study

Transportation systems specifically road transport are highly essential to economic

activities. They play a vital role in marketing products and providing citizens with ease

of access to points of contact for business activities, health, education, and agriculture.

Road transportation systems must be safe and efficient to keep the above services

active. However, almost every road section has a significant risk of traffic accidents—

a primary global concern due to many fatalities and economic losses every year. For

example in Malaysia, the recent statistics show that deaths per 100,000 people are

nearly 24 for all road users (Global Status Report on Road Safety 2015). Mainly,

expressways and highways in urban areas are potential sites of fatal traffic accidents.

On an average, 18 people are killed daily nationwide in road accidents as per a survey

conducted in 2015. In 2016, the number of fatalities jumped to 7,152 from 6,706

deaths in the earlier year (Malaysian Road Transport Department, 2017). In addition,

Malaysia had the highest fatality risk (per 100,000 population) among the Asian

countries. The majority of road accident fatalities involve motorcyclists, making up

50% of the total number of accidents (Manan and Várhelyi, 2012). Unless action is

taken, the number of deaths is expected to increase in the coming years.

Given all above, recent developments in laser scanning technology have improved

spatial data acquisition in road environments. These developments have helped to

build geometric road models that can help assessing road safety and make accurate

and valid predictions on road traffic accidents. Laser scanning systems or LiDAR

(Light Detection And Ranging) are the latest technology to capture the 3D geometry

of various objects such as 3D point clouds on large surfaces accurately and densely.

Furthermore, they offer rapid and cost-effective data acquisition about road corridors

and surrounding environments. Most of the road inventory information can be

extracted from the original design of the roads. However, road geometric elements

such as pavement roughness, vertical gradients, and horizontal curves change over

time due to construction, degradation in road conditions, and vegetation growth

(Shamayleh and Khattak, 2003). Apart from that, collecting data by conventional

measurement methods are not safe, and the surveyor may not find safe sight distances

to get readings necessary to calculate vertical grades and side slopes of the road

(Uddin, 2008). The conventional measurement methods are relatively expensive to

produce detailed topographic models and time-consuming. On the other hand,

photogrammetry and optical satellite imagery have the disadvantages of poor

visibility/ cloudy daytime, the relatively low accuracy of required topographic models,

and computationally are not efficient to generate topographic maps due to the

HT UPM

requirements of collecting dense points distributed uniformly in the focus area (Uddin,

2008).

Therefore, LiDAR technology is a promising method for acquiring data about road

corridors. In general, LiDAR sensors have higher spatial resolution than satellite-

based sensors, wider Field of View (FOV), and lower cost than traditional aerial

photogrammetric and surveying methods for right projects. By evaluating specific

height information in LiDAR data, along with high-resolution orthophotos, objects

such as road can be distinguished efficiently from other objects. Overall, LiDAR data

is practical for extracting road geometry and roadside features that many applications

including traffic accident modeling can benefit (Lee and Mannering, 2002). Thus,

developing methods that can extract accurately and rapidly road information from

LiDAR data to improve road safety is highly crucial.

One way to improve road safety is by developing accurate prediction models of traffic

accidents using road geometry and sophisticated Machine Learning (ML) and

statistical techniques. However, development of such models is not straightforward

and needs careful analysis and optimization to be practical for transportation agencies.

The research of geometric road modeling from LiDAR data has progressed a lot since

the development of advanced LiDAR systems. Various algorithms, data processing

workflows, and practical guidelines have been proposed by different researchers and

showed a significant performance compared to the traditional methods (Poullis and

You, 2010; Fix et al., 2016). The latest methods combine LiDAR data with other

ancillary information (e.g., road centerline, road attributes, knowledge about road

shape and geometry) in hierarchical forms to create accurate three-dimensional (3D)

models of road sections (Holgado‐ Barco et al., 2015). A hierarchical model (or a

model with multiple levels of processing) is defined as a data processing model that

combines sub-models (usually different algorithms) at several processing stages to

solve specific problems. For instance, in a study by Holgado‐ Barco et al. (2015), a

hierarchical model was developed by combining segmentation algorithms and

Principal Component Analysis (PCA) to model road geometry from mobile LiDAR

data. These methods have shown promising advantages over traditional methods.

However, their main limitation is the dependency on the accuracy and the

completeness of the ancillary information used to obtain the road features. Thus,

progress should be made to overcome such limitations and advance the road extraction

algorithms, and such developments will open up new areas for transportation

applications.

The literature shows the momentous development of modeling approaches for

predicting road traffic accidents including statistical and soft computing algorithms

(Hosseinpour et al., 2014; Karlaftis and Vlahogianni, 2011; Pei et al., 2011). The

traditional methods significantly focused on hand-crafted features identified by

experts in the transportation field and used to model accident frequency and injury

severity using statistical techniques such as Logistic Regression (LR) and Support

HT UPM

Vector Machine (SVM). However, as generating handcrafted features requires experts

and time, researchers thought about alternatives such as Neural Networks (NN),

Evolutionary Algorithms (EA) and developing powerful statistical models that can

better handle accident data without expensive feature engineering (Yu et al ., 2014).

A part of these developments, the availability of data and having powerful computing

resources such as processing on GPUs (Graphical Processing Units) have led to

rethinking about NN methods. Deep Learning (DL), a recent groundbreaking

development in ML community, has shed light on using NN models more efficiently

than before. DL allows computational models to learn hierarchal representations of

data with multiple levels of abstraction at different processing layers. In addition, in

limited data situations, Transfer Learning (TL) can be used in DL models to overcome

over-fitting problems. Overall, combining DL and TL with careful fine-tuning in a

single workflow is expected to provide sufficient prediction power for models

designed to simulate traffic accidents based on historical records.

Since the above shows the importance of having robust tools to model road geometry

for road safety assessment, the primary goal of this thesis is to investigate issues

surrounding the development of methods that can accurately and efficiently model

road geometry and make predictions on road traffic accidents at high-speed

expressways. This first chapter serves as an overview of the entire thesis.

Besides the state of the general topic and background of the study, it provides the

statement of the problem, the gap of knowledge. The importance of the proposed

research, the research questions, aims and research objectives, the thesis contribution,

the scope of the topic, and outlines the order of information in the rest of this thesis.

1.2 The Statement of Problem

Everyone in this world wants to have safe transportation systems to travel from a place

to another easily and securely. However, today there are many issues and challenges

making transportation systems less safe than they should be. Among these issues, rapid

urbanization over various landscape forms, population growth and migration of people

from rural to urban areas. Other challenges include lack of technical tools that can

support road safety managers to simulate future scenarios and make better plans to

solve problems related to road safety efficiently. If these problems continue, failure of

transportation systems would significantly affect the stability and development of

modern cities as transportation systems are the heart of the cities.

Specifically, there are a need to construct 3D models for highways and establish

relationships among road geometry features (e.g., vertical and horizontal curves, side

slopes) and road traffic accidents to improve safety assessments for road

transportation. Research shows that using object-based methods (or OBIA—Object-

Based Image Analysis) which take into consideration not only the spectral

information, but also other geometric, textural, and contextual information for

HT UPM

extracting features from data is more powerful than pixel-based methods (Gudex-

Cross et al., 2017). However, OBIA methods need careful optimization of the

segmentation process and the choice of relevant features as well as rule sets that are

transferable to other areas without significant loss of accuracy (Robb et al., 2015).

Although many optimization methods have been proposed for OBIA, the proposed

methods lack optimization of the two main steps of OBIA, segmentation, and

classification at the same time. Choosing segmentation parameters that can produce

the best the possible segmentation quality and classification accuracy will help

producing more accurate and complete road features.

On the other hand, technical tools are essential to make predictions for future scenarios

of road safety. There are two main groups of predictive models namely, statistical

(e.g., LR) and computational intelligence (e.g., NN). The former requires extensive

engineering works and significant efforts to extract relevant features for accident

frequency and injury severity predictions. The latter requires relatively large datasets

for training and careful optimization of model’s hyper-parameters. NN due to limited

data suffers from over-fitting, lack of generalization and computing the importance of

accident predictors and modeling the temporal/contextual structures inherent in the

accident data. The traditional feed-forward NN does not allow compositionality with

the adequate flexibility to improve the generalization and predictive ability of the

model. However, the recent DL methods allow compositionality and using accident

data as sequential data allowing modeling their inherent temporal and contextual

structures. With additional information such as spatial-temporal relationships among

accident events, it is expected to improve the accuracy and generalization of the

models. Additionally, the volume of traffic accident data usually plays a significant

role in deciding a proper prediction modeling approach. With limited data (i.e., <500

records), simple models (i.e., statistical models with fixed parameters) are often

preferred. However, some sophisticated modeling approaches such as DL models have

higher prediction capabilities and attract many attentions in recent years. Their

implementation with limited data requires the development of sophisticated models

with TL methods, which the traffic accident literature lacks.

1.3 Research Objectives

The master goal of this thesis is to contribute to the efficient road geometric modeling

from LiDAR data and to the prediction of traffic accidents on highways. For being

practical, this objective has to be split up into smaller and more specific goals, which

may be organized into methodological aims as follows:

1. To delineate road geometry in two dimensions (2D) using an integrated ACO

and OBIA methods with a novel two-stage optimization strategy for the

segmentation step.

2. To extract road geometry (i.e., 3D) from mobile LiDAR data using a hierarchal

classification technique that combines Mean Shift (MS) segmentation, SVM,

and PCA.

HT UPM

3. To develop and validate a Geometric Regression (GR) model for predicting

the frequency of traffic accidents using road geometry information.

4. And, to develop and validate a Deep Learning approach using TL for

forecasting the injury severity of road traffic accidents using historical accident

records.

1.4 Research Questions

This thesis comprehensively addresses the following research questions:

1. OBIA based requires optimizing segmentation parameters (e.g., scale, shape,

compactness). In this step, an objective function is designed to judge a sub-

optimal combination of parameters that can achieve an accurate segmentation

process as possible.

However, it is not known if optimization of segmentation and classification

processes at the same time will improve the performance of feature extraction or

not?

2. To what extent the integration of meta-heuristic optimization methods such as

ACO and OBIA can improve modeling of road geometry.

3. Can hierarchical classification (combination of several algorithms) process

mobile LiDAR data efficiently for delineating 3D road geometric features such

as vertical gradients and superelevation?

4. GR is one of the modeling methods that can be used to model the frequency of

traffic accidents. How efficient is this model for processing traffic accident

data for the Malaysian context?

5. DL due to data availability and improvements in computing power is getting

popular for many applications. How much is the prediction power of DL for

modeling road traffic accidents? In addition, can new NN architectures such

as RNN and Convolutional Neural Networks (CNN) outperform the traditional

feed-forward NN?

6. Is TL an efficient method for modeling injury severity of traffic accidents of a

limited volume data in the context of DL architectures or not?

1.5 Thesis Contributions

Significant efforts have been made in the literature on pursuing solutions for geometric

modeling of roads from LiDAR data and advancing prediction models of road traffic

accidents using both statistical and soft computing techniques. However, the existing

studies still have some major concerns as discussed in Section 1.1 and Section 1.2. As

a result, there is a need to develop new techniques or making improvements on the

HT UPM

existing techniques to provide better solutions for road safety assessment based on

road geometry.

This thesis mainly contributes to the development of new methods for image

segmentation, extracting road features in 2D and 3D from Airborne and Mobile

LiDAR data, modeling accident frequency and injury severity using DL-based

methods. First, it develops a new optimization strategy for OBIA classification based

on a two-stage optimization approach. The method optimizes the two basic steps of

OBIA, namely, segmentation and classification, to realize accurate road extraction

from LiDAR data. This is achieved by selecting an optimal scale parameter first to

maximize class separability and optimal shape and compactness parameters to

optimize the final image segments.

The second contribution of this thesis is the development of a hybrid approach that

combines ACO and OBIA-based feature extraction for LiDAR data classification and

road geometry extraction. In this approach, ACO is used to find the best combination

of features to use in OBIA for road extraction. In addition, since this approach only

extracts 2D information of a road, it was necessary to develop a semi-automated

approach for delineating 3D road geometry from mobile LiDAR data without

information about vehicle trajectory, which the state-of-the-art methods lack.

Third, this thesis then goes beyond just extracting information about roads from

LiDAR data but further uses that information with some additional data about road

traffics and environment to make predictions on traffic accidents. In particular, it

develops a model based on GR for predicting traffic accident frequency. Furthermore,

it also designs and implements models based on DL such as RNN and CNN to simulate

the injury severity of traffic accidents utilizing the temporal structure of accident data.

Finally, when data on traffic accidents are scarce, this thesis provides a model that can

work based on TL concept to overcome the need for DL models for large datasets and

to avoid overfitting problems.

1.6 Scope of Study

This study has three main scopes as follows:

1. Only two types of LiDAR systems, airborne and mobile-based were studied

for road geometry modeling, and other systems (e.g., terrestrial) were not

investigated. The latter systems are efficient for detailed assessments of

transportation assists (e.g., bridge, culvert, and tunnel) which is not the case of

the current study.

2. The validations of these models in this thesis were based on an area in

Malaysia. No transferability to other countries has been investigated due to the

non-availability of data and permission to access to police reports on traffic

accidents elsewhere. However, various experiments and evaluations were

HT UPM

conducted, analyzed and discussed on a comprehensive data that have more

than a thousand of historical records of traffic accidents.

3. The data duration was from 2009 to 2015 due to the availability of the relevant

data.

1.7 Thesis Organization

The thesis is split into five chapters.

The first chapter introduces the research topic and gives a brief background of the

study, the statement of the problem, research questions and objectives, thesis

contribution and significance of the study.

The second chapter provides an overview of the available models and discusses

several important and relevant studies. It provides a cohesive review on several topics

such as LiDAR, geometric road modeling, and traffic accident modeling.

The third chapter explains the various steps of data processing and analysis, which

makes up the newly proposed models for geometric road modeling and analyzing road

traffic accident data.

The fourth chapter discusses the results of the experiments and simulations

conducted in the current research and presents evaluations of the proposed models on

real datasets.

The last chapter summarizes the main findings of the research and offers

recommendations for future work.

HT UPM

153

6 REFERENCES

Abadi, M., Agarwal, A., Barham, P., Brevdo, E., Chen, Z., Citro, C., ... & Ghemawat,

S. (2016). Tensorflow: Large-scale machine learning on heterogeneous

distributed systems. arXiv preprint arXiv:1603.04467.

Abdelwahab, H., & Abdel-Aty, M. (2001). Development of artificial neural network

models to predict driver injury severity in traffic accidents at signalized

intersections. Transportation Research Record: Journal of the Transportation

Research Board, 1746, 6-13.

Abdelwahab, H., & Abdel-Aty, M. (2002). Artificial neural networks and logit models

for traffic safety analysis of toll plazas. Transportation Research Record:

Journal of the Transportation Research Board, 1784, 115-125.

Aghdam, M. H., Ghasem-Aghaee, N., & Basiri, M. E. (2009). Text feature selection

using ant colony optimization. Expert systems with applications, 36(3), 6843-

6853.

Akar, Ö., & Güngör, O. (2013). Classification of multispectral images using Random

Forest algorithm. Journal of Geodesy and Geoinformation, 1(2).

Akel, N. A., Kremeike, K., Filin, S., Sester, M., & Doytsher, Y. (2005). Dense DTM

generalization aided by roads extracted from LiDAR data. ISPRS WG III/3,

III, 4, 54-59.

Akgüngör, A. P., & Dogan, E. (2008). Estimating road accidents of Turkey based on

regression analysis and artificial neural network approach. Advances in

Transportation studies, 2008(16), 11-22.

Al-Ghamdi, A. S. (2002). Using logistic regression to estimate the influence of

accident factors on accident severity. Accident Analysis & Prevention, 34(6),

729-741.

Alharthy, A., & Bethel, J. (2003). Automated road extraction from LIDAR data. Paper

presented at the Proceedings of the ASPRS Annual Conference.

Alkheder, S., Taamneh, M., & Taamneh, S. (2017). Severity Prediction of Traffic

Accident Using an Artificial Neural Network. Journal of Forecasting, 36(1),

100-108.

Alwan, H. B., & Ku-Mahamud, K. R. (2012, December). Optimizing support vector

machine parameters using continuous ant colony optimization. In Computing

and Convergence Technology (ICCCT), 2012 7th International Conference on

(pp. 164-169). IEEE.

Amiridis, K., & Psarianos, B. (2015). Three Dimensional Road Design By Applying

Differential Geometry And Conventional Design Approach Criteria. Paper

presented at the Transportation Research Board 94th Annual Meeting.

HT UPM

154

Arastounia, M. (2012). Automatic classification of lidar point clouds in a railway

environment. University of Twente, Netherlands, 83.

Baatz, M. (2000). Multiresolution segmentation: an optimization approach for high

quality multi-scale image segmentation. Angewandte Geographische

Informations verarbeitung XII, 12-23.

Baczyński, D., & Parol, M. (2004). Influence of artificial neural network structure on

quality of short-term electric energy consumption forecast. IEE Proceedings-

Generation, Transmission and Distribution, 151(2), 241-245.

Baltsavias, E. P. (1999). Airborne laser scanning: basic relations and formulas. ISPRS

Journal of Photogrammetry and Remote Sensing, 54(2), 199-214.

Baltsavias, E. P. (1999b). A comparison between photogrammetry and laser scanning.

ISPRS Journal of Photogrammetry and Remote Sensing, 54(2), 83-94.

Bayer, J., Osendorfer, C., Korhammer, D., Chen, N., Urban, S., & van der Smagt, P.

(2013). On fast dropout and its applicability to recurrent networks. arXiv

preprint arXiv:1311.0701.

Bazi, Y., & Melgani, F. (2006). Toward an optimal SVM classification system for

hyperspectral remote sensing images. Geoscience and Remote Sensing, IEEE

Transactions on, 44(11), 3374-3385.

Belton, D., & Lichti, D. D. (2006). Classification and segmentation of terrestrial laser

scanner point clouds using local variance information. International Archives

of Photogrammetry, Remote Sensing and Spatial Information Sciences, 36(5),

44-49.

Blake, A., & Isard, M. (1998). Active Contours: The Application of Techniques from

Graphics, Vision, Control Theory and Statistics to Visual Tracking of Shapes

in Motion. Secaucus. Springer0Verlag, London.

Blaschke, T. (2010). Object based image analysis for remote sensing. ISPRS Journal

of Photogrammetry and Remote Sensing, 65(1), 2-16.

Blaschke, T., & Strobl, J. (2001). What’s wrong with pixels? Some recent

developments interfacing remote sensing and GIS. GeoBIT/GIS, 6(1), 12-17.

Bousquet, O., & Bottou, L. (2008). The tradeoffs of large scale learning. In Advances

in neural information processing systems (pp. 161-168).

Boyko, A., & Funkhouser, T. (2011). Extracting roads from dense point clouds in large

scale urban environment. ISPRS Journal of Photogrammetry and Remote

Sensing, 66(6), S2-S12.

Brennan, R., & Webster, T. (2006). Object-oriented land cover classification of lidar-

derived surfaces. Canadian Journal of Remote Sensing, 32(2), 162-172.

HT UPM

155

Cai, H., & Rasdorf, W. (2008). Modeling Road Centerlines and Predicting Lengths in

3‐ D Using LIDAR Point Cloud and Planimetric Road Centerline Data.

Computer‐ Aided Civil and Infrastructure Engineering, 23(3), 157-173.

Cánovas-García, F., & Alonso-Sarría, F. (2015). A local approach to optimize the

scale parameter in multiresolution segmentation for multispectral imagery.

Geocarto International, 30(8), 937-961.

Cao, C., & Sun, Y. (2014). Automatic Road Centerline Extraction from Imagery Using

Road GPS Data. Remote Sensing, 6(9), 9014-9033. doi: 10.3390/rs6099014

Cao, L. Y., & Xia, L. Z. (2007, November). An ant colony optimization approach for

SAR image segmentation. In Wavelet Analysis and Pattern Recognition, 2007.

ICWAPR'07. International Conference on (Vol. 1, pp. 296-300). IEEE.

Castelli, M., Gonçalves, I., Trujillo, L., & Popovič, A. (2017). An evolutionary system

for ozone concentration forecasting. Information Systems Frontiers, 19(5),

1123-1132.

Charaniya, A. P., Manduchi, R., & Lodha, S. K. (2004). Supervised parametric

classification of aerial lidar data. Paper presented at the Computer Vision and

Pattern Recognition Workshop, 2004. CVPRW'04. Conference on.

Chen, D., & He, X. (2015). Fast automatic three-dimensional road model

reconstruction based on mobile laser scanning system. Optik-International

Journal for Light and Electron Optics, 126(7), 725-730.

Chen, L.-C., & Lo, C.-Y. (2009). 3D road modeling via the integration of large-scale

topomaps and airborne LIDAR data. Journal of the Chinese Institute of

Engineers, 32(6), 811-823. doi: 10.1080/02533839.2009.9671565

Chen, Q., Zhao, Y., Pan, S., & Wang, Y. (2016). Survey of the influence of the width

of urban branch roads on the meeting of two-way vehicle flows. PLoS one,

11(2), e0149188.

Choi, E., & Lee, C. (2003). Feature extraction based on the Bhattacharyya distance.

Pattern Recognition, 36(8), 1703-1709.

Choi, Y.-W., Jang, Y.-W., Lee, H.-J., & Cho, G.-S. (2008). Three-dimensional LiDAR

data classifying to extract road point in urban area. Geoscience and Remote

Sensing Letters, IEEE, 5(4), 725-729.

Chollet, F. (2015). Keras.

Chung, J., Gulcehre, C., Cho, K., & Bengio, Y. (2014). Empirical evaluation of gated

recurrent neural networks on sequence modeling. arXiv preprint

arXiv:1412.3555.

Clode, S., Kootsookos, P. J., & Rottensteiner, F. (2004). The automatic extraction of

roads from LIDAR data. Paper presented at the ISPRS 2004.

HT UPM

156

Clode, S., Rottensteiner, F., Kootsookos, P., & Zelniker, E. (2007). Detection and

vectorization of roads from lidar data. Photogrammetric Engineering &

Remote Sensing, 73(5), 517-535.

Cohen, L. D. (1991). On active contour models and balloons. CVGIP: Image

understanding, 53(2), 211-218.

Collobert, R., & Weston, J. (2008, July). A unified architecture for natural language

processing: Deep neural networks with multitask learning. In Proceedings of

the 25th international conference on Machine learning (pp. 160-167). ACM.

Comaniciu, D., & Meer, P. (2002). Mean shift: A robust approach toward feature

space analysis. IEEE Transactions on pattern analysis and machine

intelligence, 24(5), 603-619.

Congalton, R. G. (1981). The use of discrete multivariate analysis for the assessment

of Landsat classification accuracy. Mater thesis, Virginia Polytechnic Institute

and State University, Blacksburg, VA.

Dean, J., Corrado, G., Monga, R., Chen, K., Devin, M., Mao, M., ... & Ng, A. Y.

(2012). Large scale distributed deep networks. In Advances in neural

information processing systems (pp. 1223-1231).

Deka, L., & Quddus, M. (2014). Network-level accident-mapping: distance based

pattern matching using artificial neural network. Accident Analysis &

Prevention, 65, 105-113.

Delen, D., Sharda, R., & Bessonov, M. (2006). Identifying significant predictors of

injury severity in traffic accidents using a series of artificial neural networks.

Accident Analysis & Prevention, 38(3), 434-444.

Delen, Dursun, Ramesh Sharda, and Max Bessonov. "Identifying significant

predictors of injury severity in traffic accidents using a series of artificial

neural networks." Accident Analysis & Prevention 38.3 (2006): 434-444.

Dial, G., Gibson, L., & Poulsen, R. (2001). IKONOS satellite imagery and its use in

automated road extraction. Automatic Extraction of Man-Made Objects from

Aerial and Space Images, 3, 357-367.

Ding, S., Xu, X., & Nie, R. (2013). Extreme learning machine and its applications.

Neural Computing and Applications, 25(3-4), 549-556. doi: 10.1007/s00521-

013-1522-8

Donahue, J., Anne Hendricks, L., Guadarrama, S., Rohrbach, M., Venugopalan, S.,

Saenko, K., & Darrell, T. (2015). Long-term recurrent convolutional networks

for visual recognition and description. In Proceedings of the IEEE conference

on computer vision and pattern recognition (pp. 2625-2634).

Dong, N., Huang, H., & Zheng, L. (2015). Support vector machine in crash prediction

at the level of traffic analysis zones: Assessing the spatial proximity effects.

Accid Anal Prev, 82, 192-198. doi: 10.1016/j.aap.2015.05.018

HT UPM

157

Dorigo, M., & Stützle, T. (2003). The ant colony optimization metaheuristic:

Algorithms, applications, and advances. In Handbook of metaheuristics (pp.

250-285). Springer US.

Dorigo, M., Birattari, M., & Stützle, T. (2006). Ant colony optimization.

Computational Intelligence Magazine, IEEE, 1(4), 28-39.

Duan, Y., Lv, Y., Liu, Y. L., & Wang, F. Y. (2016). An efficient realization of deep

learning for traffic data imputation. Transportation Research Part C: Emerging

Technologies, 72, 168-181.

Duchi, J., Hazan, E., & Singer, Y. (2011). Adaptive subgradient methods for online

learning and stochastic optimization. Journal of Machine Learning Research,

12(Jul), 2121-2159.

Eberhart, R., & Kennedy, J. (1995, October). A new optimizer using particle swarm

theory. In Micro Machine and Human Science, 1995. MHS'95., Proceedings

of the Sixth International Symposium on (pp. 39-43). IEEE.

El-Basyouny, K. & T. Sayed (2010) Application of generalized link functions in

developing accident prediction models. Safety science, 48, 410-416.

El-Basyouny, K., & Sayed, T. (2006). Comparison of two negative binomial

regression techniques in developing accident prediction models.

Transportation Research Record: Journal of the Transportation Research

Board, (1950), 9-16.

Elberink, S. O., & Vosselman, G. (2006). Adding the third dimension to a topographic

database using airborne laser scanner data. Int. Arch. Photogramm. Remote

Sens. Spat. Inf. Sci, 36, 92-97.

Elvik, R. (2013). International transferability of accident modification functions for

horizontal curves. Accid Anal Prev, 59, 487-496. doi:

10.1016/j.aap.2013.07.010

Engelbrecht, A. P. (2007). Computational intelligence: an introduction. John Wiley &

Sons.

ESRI. Locate Outliers. 2016. http://pro.arcgis.com/en/pro-app/tool-reference/3d-

analyst/locate-outliers.htm. Accessed Jan 19, 2016.

Evans, J. S., & Hudak, A. T. (2007). A multiscale curvature algorithm for classifying

discrete return LiDAR in forested environments. Geoscience and Remote

Sensing, IEEE Transactions on, 45(4), 1029-1038.

Ferraz, A., Mallet, C., & Chehata, N. (2016). Large-scale road detection in forested

mountainous areas using airborne topographic lidar data. ISPRS Journal of

Photogrammetry and Remote Sensing, 112, 23-36.

HT UPM

158

Fix, A., Amediek, A., Ehret, G., Groß, S. M., Kiemle, C., Reitebuch, O., & Wirth, M.

(2016). On the benefit of airborne demonstrators for space borne lidar

missions.

Friedl, M. A., & Brodley, C. E. (1997). Decision tree classification of land cover from

remotely sensed data. Remote Sensing of Environment, 61(3), 399-409.

Gardner, M. W., & Dorling, S. R. (1998). Artificial neural networks (the multilayer

perceptron)—a review of applications in the atmospheric sciences.

Atmospheric environment, 32(14), 2627-2636.

Ghasemloo, N., Reza Mobasheri, M., Madanchi Zare, A., & Memar Eftekhari, M.

(2013). Road and Tunnel Extraction from SPOT Satellite Images Using Neural

Networks. Journal of Geographic Information System, 05(01), 69-74. doi:

10.4236/jgis.2013.51007

Ghasemlou, K., Aydi, M. M., & Yildirim, M. S. (2015). Prediction of pedal cyclists

and pedestrian fatalities from total monthly accidents and registered private car

numbers. Archives of Transport, 34(2), 29-35.

Gikas, V. (2012). Three-dimensional laser scanning for geometry documentation and

construction management of highway tunnels during excavation. Sensors,

12(8), 11249-11270.

Gikas, V., & Stratakos, J. (2012). A novel geodetic engineering method for accurate

and automated road/railway centerline geometry extraction based on the

bearing diagram and fractal behavior. Intelligent Transportation Systems,

IEEE Transactions on, 13(1), 115-126.

Glymour, C., Madigan, D., Pregibon, D., & Smyth, P. (1997). Statistical themes and

lessons for data mining. Data mining and knowledge discovery, 1(1), 11-28.

Goepfert, J., & Rottensteiner, F. (2009). Adaptation of roads to ALS data by means of

network snakes: IntArchPhRS 38-3/W8.

Gong, J., Zhou, H., Gordon, C., & Jalayer, M. (2012). Mobile terrestrial laser scanning

for highway inventory data collection. In Computing in Civil Engineering

(2012) (pp. 545-552).

Gong, L., Zhang, Y., Li, Z., & Bao, Q. (2010). Automated road extraction from

LiDAR data based on intensity and aerial photo. Paper presented at the Image

and Signal Processing (CISP), 2010 3rd International Congress on.

Göpfert, J., Rottensteiner, F., & Heipke, C. (2011). Using snakes for the registration

of topographic road database objects to ALS features. ISPRS Journal of

Photogrammetry and Remote Sensing, 66(6), 858-871.

Goulette, F., Nashashibi, F., Abuhadrous, I., Ammoun, S., & Laurgeau, C. (2006). An

integrated on-board laser range sensing system for on-the-way city and road

modelling. International Archives of the Photogrammetry, Remote Sensing

and Spatial Information Sciences, 34(A).

HT UPM

159

Graves, A., Mohamed, A. R., & Hinton, G. (2013, May). Speech recognition with deep

recurrent neural networks. In Acoustics, speech and signal processing (icassp),

2013 ieee international conference on (pp. 6645-6649). IEEE.

Grote, A., Heipke, C., & Rottensteiner, F. (2012). Road network extraction in

suburban areas. The Photogrammetric Record, 27(137), 8-28.

Guan, H., Li, J., Chapman, M., Deng, F., Ji, Z., & Yang, X. (2013). Integration of

orthoimagery and lidar data for object-based urban thematic mapping using

random forests. International journal of remote sensing, 34(14), 5166-5186.

Guan, H., Li, J., Yu, Y., Ji, Z., & Wang, C. (2015). Using mobile LiDAR data for

rapidly updating road markings. IEEE Transactions on Intelligent

Transportation Systems, 16(5), 2457-2466.

Guan, H., Li, J., Yu, Y., Wang, C., Chapman, M., & Yang, B. (2014). Using mobile

laser scanning data for automated extraction of road markings. ISPRS Journal

of Photogrammetry and Remote Sensing, 87, 93-107.

Guan, J., Wang, Z., & Yao, X. (2010). A new approach for road centerlines extraction

and width estimation. Paper presented at the Signal Processing (ICSP), 2010

IEEE 10th International Conference on.

Gudex-Cross, D., Pontius, J., & Adams, A. (2017). Enhanced forest cover mapping

using spectral unmixing and object-based classification of multi-temporal

Landsat imagery. Remote Sensing of Environment, 196, 193-204.

Hamedianfar, A., & Shafri, H. Z. (2014). Development of fuzzy rule-based parameters

for urban object-oriented classification using very high resolution imagery.

Geocarto International, 29(3), 268-292.

Hamedianfar, A., & Shafri, H. Z. M. (2016). Integrated approach using data mining-

based decision tree and object-based image analysis for high-resolution urban

mapping of WorldView-2 satellite sensor data. Journal of Applied Remote

Sensing, 10(2), 025001-025001

Hamedianfar, A., Shafri, H. Z. M., Mansor, S., & Ahmad, N. (2014). Improving

detailed rule-based feature extraction of urban areas from WorldView-2 image

and lidar data. International Journal of Remote Sensing, 35(5), 1876-1899.

Han, L., Luo, S., Yu, J., Pan, L., & Chen, S. (2015). Rule extraction from support

vector machines using ensemble learning approach: an application for

diagnosis of diabetes. IEEE journal of biomedical and health informatics,

19(2), 728-734.

Hand, D. J. (2000). Data mining: new challenges for statisticians. Social Science

Computer Review, 18(4), 442-449.

Hanspal, N.S. Allison, B.A., Deka, L., & Das, D.B. (2013). Artificial neural network

(ANN) modelling of dynamic effects on two-phase flow in hom*ogeneous

porous media. Journal of Hydroinformatics, 15(2), 540–554.

HT UPM

160

Hartfield, K. A., Landau, K. I., & Van Leeuwen, W. J. (2011). Fusion of high

resolution aerial multispectral and LiDAR data: Land cover in the context of

urban mosquito habitat. Remote Sensing, 3(11), 2364-2383.

Hashmienejad, S. H. A., & Hasheminejad, S. M. H. (2017). Traffic accident severity

prediction using a novel multi-objective genetic algorithm. International

Journal of Crashworthiness, 1-16.

Hatger, C., & Brenner, C. (2003). Extraction of road geometry parameters from laser

scanning and existing databases. International Archives of Photogrammetry,

Remote Sensing and Spatial Information Sciences, 34(3/W13), 225-230.

He, Y., Song, Z., & Liu, Z. (2017). Updating highway asset inventory using airborne

LiDAR. Measurement, 104, 132-141.

Hernández, J., & Marcotegui, B. (2009, September). Filtering of artifacts and

pavement segmentation from mobile lidar data. In ISPRS Workshop

Laserscanning 2009.

Herumurti, D., Uchimura, K., Koutaki, G., & Uemura, T. (2013). Urban road

extraction based on hough transform and region growing. Paper presented at

the Frontiers of Computer Vision,(FCV), 2013 19th Korea-Japan Joint

Workshop on.

Hinton, G. E., Rumelhart, D. E., & Williams, R. J. (1985). Learning internal

representations by back-propagating errors. Parallel Distributed Processing:

Explorations in the Microstructure of Cognition, 1.

Hinton, G. E., Srivastava, N., Krizhevsky, A., Sutskever, I., & Salakhutdinov, R. R.

(2012). Improving neural networks by preventing co-adaptation of feature

detectors. arXiv preprint arXiv:1207.0580.

Hirpa, D., Hare, W., Lucet, Y., Pushak, Y., & Tesfamariam, S. (2016). A bi-objective

optimization framework for three-dimensional road alignment design.

Transportation Research Part C: Emerging Technologies, 65, 61-78.

Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural

computation, 9(8), 1735-1780.

Hofmann, P. (2014). Defining Robustness Measures for OBIA Framework: A Case

Study for Detecting Informal Settlements. In Q. Weng (Ed.), Global Urban

Monitoring and Assessment through Earth Observation (pp. 303-324). CRC

Press.

Holgado-Barco, A., Gonzalez-Aguilera, D., Arias-Sanchez, P., & Martinez-Sanchez,

J. (2014). An automated approach to vertical road characterisation using

mobile LiDAR systems: Longitudinal profiles and cross-sections. ISPRS

Journal of Photogrammetry and Remote Sensing, 96, 28-37.

HT UPM

161

Holgado‐ Barco, A., Riveiro, B., González‐ Aguilera, D., & Arias, P. (2017).

Automatic Inventory of Road Cross‐ Sections from Mobile Laser Scanning

System. Computer‐ Aided Civil and Infrastructure Engineering, 32(1), 3-17.

Hong, H., Pradhan, B., Jebur, M. N., Bui, D. T., Xu, C., & Akgun, A. (2016). Spatial

prediction of landslide hazard at the Luxi area (China) using support vector

machines. Environmental Earth Sciences, 75(1), 40.

Hu, X., Li, Y., Shan, J., Zhang, J., & Zhang, Y. (2014). Road Centerline Extraction in

Complex Urban Scenes From LiDAR Data Based on Multiple Features.

Geoscience and Remote Sensing, IEEE Transactions on, 52(11), 7448-7456.

Hu, X., Tao, C. V., & Hu, Y. (2004). Automatic road extraction from dense urban area

by integrated processing of high resolution imagery and lidar data.

International Archives of Photogrammetry, Remote Sensing and Spatial

Information Sciences. Istanbul, Turkey, 35, B3.

Huang, C.-L., & Dun, J.-F. (2008). A distributed PSO–SVM hybrid system with

feature selection and parameter optimization. Applied Soft Computing, 8(4),

1381-1391.

Huang, H., Zeng, Q., Pei, X., Wong, S. C., & Xu, P. (2016). Predicting crash frequency

using an optimised radial basis function neural network model.

Transportmetrica: A Transport Science, 12(4), 330-345.

Huang, W., Song, G., Hong, H., & Xie, K. (2014). Deep architecture for traffic flow

prediction: Deep belief networks with multitask learning. IEEE Transactions

on Intelligent Transportation Systems, 15(5), 2191-2201.

Hui, Z., Hu, Y., Jin, S., & Yevenyo, Y. Z. (2016). Road centerline extraction from

airborne LiDAR point cloud based on hierarchical fusion and optimization.

ISPRS Journal of Photogrammetry and Remote Sensing, 118, 22-36.

Ibrahim, S., & Lichti, D. (2012). Curb-based street floor extraction from mobile

terrestrial LiDAR point cloud. ISPRS-International Archives of the

Photogrammetry, Remote Sensing and Spatial Information Sciences, 39, B5.

Im, J., Jensen, J. R., & Hodgson, M. E. (2008). Object-based land cover classification

using high-posting-density LiDAR data. GIScience & Remote Sensing, 45(2),

209-228.

Immitzer, M., Atzberger, C., & Koukal, T. (2012). Tree species classification with

random forest using very high spatial resolution 8-band WorldView-2 satellite

data. Remote Sensing, 4(9), 2661-2693.

Jaakkola, A., Hyyppä, J., Hyyppä, H., & Kukko, A. (2008). Retrieval algorithms for

road surface modelling using laser-based mobile mapping. Sensors, 8(9),

5238-5249.

Jabari, S., & Zhang, Y. (2013). Very high resolution satellite image classification

using Fuzzy rule-based systems. Algorithms, 6(4), 762-781.

HT UPM

162

Jadaan, K. S., Al-Fayyad, M., & Gammoh, H. F. (2014). Prediction of road traffic

accidents in jordan using artificial neural network (ANN). Journal of Traffic

and Logistics Engineering, 2(2), 92-94.

Jalayer, M., Gong, J., Zhou, H., & Grinter, M. (2015). Evaluation of remote sensing

technologies for collecting roadside feature data to support highway safety

manual implementation. Journal of Transportation Safety & Security, 7(4),

345-357.

Kang, M. W., Jha, M. K., & Schonfeld, P. (2012). Applicability of highway alignment

optimization models. Transportation Research Part C: Emerging

Technologies, 21(1), 257-286.

Kang, M.-W., Shariat, S., & Jha, M. K. (2013). New highway geometric design

methods for minimizing vehicular fuel consumption and improving safety.

Transportation Research Part C: Emerging Technologies, 31, 99-111. doi:

10.1016/j.trc.2013.03.002

Karlaftis, M. G., & Vlahogianni, E. I. (2011). Statistical methods versus neural

networks in transportation research: Differences, similarities and some

insights. Transportation Research Part C: Emerging Technologies, 19(3), 387-

399.

Kass, M., Witkin, A., & Terzopoulos, D. (1988). Snakes: Active contour models.

International Journal of Computer Vision, 1(4), 321-331.

Kennedy, J. (2010). Particle swarm optimization Encyclopedia of Machine Learning

(pp. 760-766): Springer.

Kennedy, J., & Optimization, R. E. P. S. (1995). IEEE Int. Paper presented at the Conf.

on Neural Networks.

Kingma, D., & Ba, J. (2014). Adam: A method for stochastic optimization. arXiv

preprint arXiv:1412.6980.

Klang, D. (1998). Automatic detection of changes in road data bases using satellite

imagery. International Archives of Photogrammetry and Remote Sensing, 32,

293-298.

Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). Imagenet classification with

deep convolutional neural networks. In Advances in neural information

processing systems (pp. 1097-1105).

Kumar, P., Lewis, P., & McCarthy, T. (2017). The Potential of Active Contour Models

in Extracting Roads from Mobile Laser Scanning Data.

Kumar, P., McElhinney, C. P., Lewis, P., & McCarthy, T. (2013). An automated

algorithm for extracting road edges from terrestrial mobile LiDAR data. ISPRS

Journal of Photogrammetry and Remote Sensing, 85, 44-55.

HT UPM

163

Kumar, P., McElhinney, C. P., Lewis, P., & McCarthy, T. (2014). Automated road

markings extraction from mobile laser scanning data. International Journal of

Applied Earth Observation and Geoinformation, 32, 125-137. doi:

10.1016/j.jag.2014.03.023

Kunt, M. M., Aghayan, I., & Noii, N. (2011). Prediction for traffic accident severity:

comparing the artificial neural network, genetic algorithm, combined genetic

algorithm and pattern search methods. Transport, 26(4), 353-366.

Lak, A. M., Zoej, M. J. V., & Mokhtarzade, M. (2016). A new method for road

detection in urban areas using high-resolution satellite images and Lidar data

based on fuzzy nearest-neighbor classification and optimal features. Arabian

Journal of Geosciences, 9(5), 1-11.

Lam, J., Kusevic, K., Mrstik, P., Harrap, R., & Greenspan, M. (2010, May). Urban

scene extraction from mobile ground based lidar data. In Proceedings of

3DPVT (pp. 1-8).

Lary, D. J., Alavi, A. H., Gandomi, A. H., & Walker, A. L. (2015). Machine learning

in geosciences and remote sensing. Geoscience Frontiers. doi:

10.1016/j.gsf.2015.07.003

LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Nature, 521(7553), 436-

444.

Lee, J., & Mannering, F. (2002). Impact of roadside features on the frequency and

severity of run-off-roadway accidents: an empirical analysis. Accident

Analysis & Prevention, 34(2), 149-161.

Lek, S., Belaud, A., Dimopoulos, I., Lauga, J., & Moreau, J. (1995). Improved

estimation, using neural networks, of the food consumption of fish

populations. Marine and Freshwater Research, 46(8), 1229-1236.

Li, C., Wang, J., Wang, L., Hu, L., & Gong, P. (2014). Comparison of classification

algorithms and training sample sizes in urban land classification with Landsat

thematic mapper imagery. Remote Sensing, 6(2), 964-983.

Li, L. L., & Wang, J. K. (2012, October). SAR image ship detection based on Ant

Colony Optimization. In Image and Signal Processing (CISP), 2012 5th

International Congress on (pp. 1100-1103). IEEE.

Li, M., Stein, A., Bijker, W., & Zhan, Q. (2016). Region-based urban road extraction

from VHR satellite images using Binary Partition Tree. International Journal

of Applied Earth Observation and Geoinformation, 44, 217-225.

Li, P., & He, J. (2016). Geometric design safety estimation based on tire–road side

friction. Transportation Research Part C: Emerging Technologies, 63, 114-

125.

HT UPM

164

Li, X., Cheng, X., Chen, W., Chen, G., & Liu, S. (2015). Identification of Forested

Landslides Using LiDar Data, Object-based Image Analysis, and Machine

Learning Algorithms. Remote Sensing, 7(8), 9705-9726.

Li, X., Lord, D., Zhang, Y., & Xie, Y. (2008). Predicting motor vehicle crashes using

support vector machine models. Accident Analysis & Prevention, 40(4), 1611-

1618.

Li, Z., Liu, P., Wang, W., & Xu, C. (2012). Using support vector machine models for

crash injury severity analysis. Accident Analysis & Prevention, 45, 478-486.

Lin, X., Zhang, J., Liu, Z., Shen, J., & Duan, M. (2011). Semi-automatic extraction of

road networks by least squares interlaced template matching in urban areas.

International Journal of Remote Sensing, 32(17), 4943-4959. doi:

10.1080/01431161.2010.493565

Linn, K. A., Gaonkar, B., Satterthwaite, T. D., Doshi, J., Davatzikos, C., & Shinohara,

R. T. (2016). Control-group feature normalization for multivariate pattern

analysis of structural MRI data using the support vector machine. NeuroImage,

132, 157-166.

Lipton, Z. C., Berkowitz, J., & Elkan, C. (2015). A critical review of recurrent neural

networks for sequence learning. arXiv preprint arXiv:1506.00019.

Liu, Y., Bian, L., Meng, Y., Wang, H., Zhang, S., Yang, Y., ... & Wang, B. (2012).

Discrepancy measures for selecting optimal combination of parameter values

in object-based image analysis. ISPRS journal of photogrammetry and remote

sensing, 68, 144-156.

Liu, Y., Monteiro, S. T., & Saber, E. (2016, July). Vehicle detection from aerial color

imagery and airborne LiDAR data. In Geoscience and Remote Sensing

Symposium (IGARSS), 2016 IEEE International (pp. 1384-1387). IEEE.

Long, M., Wang, J., & Jordan, M. I. (2016). Deep transfer learning with joint

adaptation networks. arXiv preprint arXiv:1605.06636.

Longley, P. A., Goodchild, M. F., Maguire, D. J., & Rhind, D. W. (2001). Geographic

information system and Science. England: John Wiley & Sons, Ltd.

Lv, Y., Duan, Y., Kang, W., Li, Z., & Wang, F. Y. (2015). Traffic flow prediction

with big data: a deep learning approach. IEEE Transactions on Intelligent

Transportation Systems, 16(2), 865-873.

Lv, Y., Tang, S., & Zhao, H. (2009, April). Real-time highway traffic accident

prediction based on the k-nearest neighbor method. In Measuring Technology

and Mechatronics Automation, 2009. ICMTMA'09. International Conference

on (Vol. 3, pp. 547-550). IEEE.

Ma, X., Yu, H., Wang, Y., & Wang, Y. (2015). Large-scale transportation network

congestion evolution prediction using deep learning theory. PloS one, 10(3),

e0119044.

HT UPM

165

Manan, M. M. A., & Várhelyi, A. (2012). Motorcycle fatalities in Malaysia. IATSS

research, 36(1), 30-39.

Manan, M. M. A., Jonsson, T., & Várhelyi, A. (2013). Development of a safety

performance function for motorcycle accident fatalities on Malaysian primary

roads. Safety science, 60, 13-20.

Matkan, A. A., Hajeb, M., & Sadeghian, S. (2014). Road extraction from lidar data

using support vector machine classification. Photogrammetric Engineering &

Remote Sensing, 80(5), 409-422.

McElhinney, C. P., Kumar, P., Cahalane, C., & McCarthy, T. (2010, June). Initial

results from European Road Safety Inspection (EURSI) mobile mapping

project. In ISPRS Commission V Technical Symposium (pp. 440-445).

Mena, J. B. (2003). State of the art on automatic road extraction for GIS update: a

novel classification. Pattern Recognition Letters, 24(16), 3037-3058. doi:

10.1016/s0167-8655(03)00164-8

Mia, M. M. A., Biswas, S. K., Urmi, M. C., & Siddique, A. (2015). An algorithm for

training multilayer perceptron (MLP) for Image reconstruction using neural

network without overfitting. Int J Sci Techno Res, 4(2), 271-275.

Ming, D., Li, J., Wang, J., & Zhang, M. (2015). Scale parameter selection by spatial

statistics for GeOBIA: Using mean-shift based multi-scale segmentation as an

example. ISPRS Journal of Photogrammetry and Remote Sensing, 106, 28-41.

Minh, N. Q., & Hien, L. P. (2011). Land cover classification using LiDAR intensity

data and neural network. Journal of the Korean Society of Surveying, Geodesy,

Photogrammetry and Cartography, 29(4), 429-438.

Moghaddam, F. R., Afandizadeh, S., & Ziyadi, M. (2011). Prediction of accident

severity using artificial neural networks. International Journal of Civil

Engineering, 9(1), 41.

Mokhtarzade, M., & Zoej, M. V. (2007). Road detection from high-resolution satellite

images using artificial neural networks. International journal of applied earth

observation and geoinformation, 9(1), 32-40.

Mountrakis, G., Im, J., & Ogole, C. (2011). Support vector machines in remote

sensing: A review. ISPRS Journal of Photogrammetry and Remote Sensing,

66(3), 247-259.

Munoz, D., Vandapel, N., & Hebert, M. (2008). Directional associative markov

network for 3-d point cloud classification.

Nicholls, D. (1999). Presidential Address: Statistics into the 21st century (with

discussion). Australian & New Zealand Journal of Statistics, 41(2), 127-139.

HT UPM

166

Ogwueleka, F. N., Misra, S., Ogwueleka, T. C., & Fernandez-Sanz, L. (2014). An

artificial neural network model for road accident prediction: a case study of a

developing country. Acta Polytechnica Hungarica, 11(5), 177-197.

Olden, J. D., & Jackson, D. A. (2002). Illuminating the “black box”: a randomization

approach for understanding variable contributions in artificial neural networks.

Ecological modelling, 154(1), 135-150.

Oord, A. V. D., Kalchbrenner, N., & Kavukcuoglu, K. (2016). Pixel recurrent neural

networks. arXiv preprint arXiv:1601.06759.

Oude Elberink, S. J., & Vosselman, G. (2009). 3D information extraction from laser

point clouds covering complex road junctions. The Photogrammetric Record,

24(125), 23-36.

Pagounis, V., Tsakiri, M., & Palaskas, S. (2007). Road safety analysis: terrestrial laser

scanning to improve road safety.

Parpinelli, R. S., Lopes, H. S., & Freitas, A. A. (2002). Data mining with an ant colony

optimization algorithm. Evolutionary Computation, IEEE Transactions on,

6(4), 321-332.

Pascanu, R., Mikolov, T., & Bengio, Y. (2013, February). On the difficulty of training

recurrent neural networks. In International Conference on Machine Learning

(pp. 1310-1318).

Pauly, M., Gross, M., & Kobbelt, L. P. (2002, October). Efficient simplification of

point-sampled surfaces. In Proceedings of the conference on Visualization'02

(pp. 163-170). IEEE Computer Society.

Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., ... &

Vanderplas, J. (2011). Scikit-learn: Machine learning in Python. Journal of

Machine Learning Research, 12(Oct), 2825-2830.

Pei, X., Wong, S. C., & Sze, N. N. (2011). A joint-probability approach to crash

prediction models. Accident Analysis & Prevention, 43(3), 1160-1166.

Pei, Y., Wang, W., & Zhang, S. (2012, March). Basic ant colony optimization. In

Computer Science and Electronics Engineering (ICCSEE), 2012 International

Conference on (Vol. 1, pp. 665-667). IEEE.

Poullis, C., & You, S. (2010). Delineation and geometric modeling of road networks.

ISPRS Journal of Photogrammetry and Remote Sensing, 65(2), 165-181.

Pradhan, B., Jebur, M. N., Shafri, H. Z. M., & Tehrany, M. S. Data Fusion Technique

Using Wavelet Transform and Taguchi Methods for Automatic Landslide

Detection From Airborne Laser Scanning Data and QuickBird Satellite

Imagery.

HT UPM

167

Priyanka, A., & Sathiyakumari, K. (2014). A comparative study of classification

algorithm using accident data. International Journal of Computer Science &

Engineering Technology (IJCSET), 5(10), 1018-1023.

Pu, S., Rutzinger, M., Vosselman, G., & Elberink, S. O. (2011). Recognizing basic

structures from mobile laser scanning data for road inventory studies. ISPRS

Journal of Photogrammetry and Remote Sensing, 66(6), S28-S39.

Raschka, S. (2015). Python machine learning. Packt Publishing Ltd.

Rees, W. G. (2000). The accuracy of digital elevation models interpolated to higher

resolutions. International journal of remote sensing, 21(1), 7-20.

Robb, C., Willis, I., Arnold, N., & Guðmundsson, S. (2015). A semi-automated

method for mapping glacial geomorphology tested at Breiðamerkurjökull,

Iceland. Remote Sensing of Environment, 163, 80-90.

Sadek, A., Spring, G., & Smith, B. (2003). Toward more effective transportation

applications of computational intelligence paradigms. Transportation

Research Record: Journal of the Transportation Research Board, (1836), 57-

63.

Samadzadegan, F., & Bigdeli, B. (2009). Combining Multiple Classifiers for

Automatic Road Extraction from Lidar Data.

Sarikaya, R., Hinton, G. E., & Deoras, A. (2014). Application of deep belief networks

for natural language understanding. IEEE/ACM Transactions on Audio,

Speech and Language Processing (TASLP), 22(4), 778-784.

Sarle, W. S. (1994). Neural networks and statistical models.

Schindler, K., & Van Gool, L. (2008, June). Action snippets: How many frames does

human action recognition require?. In Computer Vision and Pattern

Recognition, 2008. CVPR 2008. IEEE Conference on (pp. 1-8). IEEE.

Sebari, I., & He, D.-C. (2013). Automatic fuzzy object-based analysis of VHSR

images for urban objects extraction. ISPRS Journal of Photogrammetry and

Remote Sensing, 79, 171-184.

Semeida, A. M. (2015). Derivation of level of service by artificial neural networks at

horizontal curves: a case study in Egypt. European Transport Research

Review, 7(1). doi: 10.1007/s12544-014-0152-2

Shamayleh, H., & Khattak, A. (2003, August). Utilization of LiDAR technology for

highway inventory. In Proceedings of the 2003 Mid-Continent Transportation

Research Symposium, Ames, Iowa.

Shin, H. C., Roth, H. R., Gao, M., Lu, L., Xu, Z., Nogues, I., ... & Summers, R. M.

(2016). Deep convolutional neural networks for computer-aided detection:

CNN architectures, dataset characteristics and transfer learning. IEEE

transactions on medical imaging, 35(5), 1285-1298.

HT UPM

168

Sohn, S. Y., & Lee, S. H. (2003). Data fusion, ensemble and clustering to improve the

classification accuracy for the severity of road traffic accidents in Korea.

Safety Science, 41(1), 1-14.

Sohn, S. Y., & Shin, H. (2001). Pattern recognition for road traffic accident severity

in Korea. Ergonomics, 44(1), 107-117.

Song, J.-H., Han, S.-H., Yu, K., & Kim, Y.-I. (2002). Assessing the possibility of land-

cover classification using lidar intensity data. International Archives of

Photogrammetry Remote Sensing and Spatial Information Sciences, 34(3/B),

259-262.

Song, M., & Civco, D. (2004). Road extraction using SVM and image segmentation.

Photogrammetric Engineering & Remote Sensing, 70(12), 1365-1371.

Srivastava, N., Hinton, G. E., Krizhevsky, A., Sutskever, I., & Salakhutdinov, R.

(2014). Dropout: a simple way to prevent neural networks from overfitting.

Journal of Machine Learning Research, 15(1), 1929-1958.

Stoica, R., Descombes, X., & Zerubia, J. (2004). A Gibbs point process for road

extraction from remotely sensed images. International Journal of Computer

Vision, 57(2), 121-136.

Su, L. (2009). Optimizing support vector machine learning for semi-arid vegetation

mapping by using clustering analysis. ISPRS Journal of Photogrammetry and

Remote Sensing, 64(4), 407-413.

Sutskever, I., Martens, J., & Hinton, G. E. (2011). Generating text with recurrent

neural networks. In Proceedings of the 28th International Conference on

Machine Learning (ICML-11) (pp. 1017-1024).

Taamneh, M., Taamneh, S., & Alkheder, S. (2016). Clustering-based classification of

road traffic accidents using hierarchical clustering and artificial neural

networks. International Journal of Injury Control and Safety Promotion, 1-8.

Tao, C. V. (2000). Mobile mapping technology for road network data acquisition.

Journal of Geospatial Engineering, 2(2), 1-14.

Tehrany, M. S., Pradhan, B., & Jebur, M. N. (2014). Flood susceptibility mapping

using a novel ensemble weights-of-evidence and support vector machine

models in GIS. Journal of Hydrology, 512, 332-343.

Tieleman, T., & Hinton, G. (2012). Lecture 6.5-rmsprop: Divide the gradient by a

running average of its recent magnitude. COURSERA: Neural networks for

machine learning, 4(2).

Ton, J., Jain, A. K., Enslin, W. R., & Hudson, W. D. (1989). Automatic road

identification and labeling in Landsat 4 TM images. Photogrammetria, 43(5),

257-276.

HT UPM

169

Uddin, W. (2008). Airborne laser terrain mapping for expediting highway projects:

Evaluation of accuracy and cost. Journal of Construction Engineering and

Management, 134(6), 411-420.

Vosselman, G. (2009, September). Advanced point cloud processing. In

Photogrammetric week (Vol. 9, pp. 137-146).

Vosselman, G., & Zhou, L. (2009). Detection of curbstones in airborne laser scanning

data. The International Archives of the Photogrammetry, Remote Sensing and

Spatial Information Sciences, 38, 111-116.

Wang, C., M. A. Quddus & S. G. Ison (2011) Predicting accident frequency at their

severity levels and its application in site ranking using a two-stage mixed

multivariate model. Accident Analysis & Prevention, 43, 1979-1990.

Wang, G., Zhang, Y., Li, J., & Song, P. (2011). 3D road information extraction from

LIDAR data fused with aerial-images. Paper presented at the Spatial Data

Mining and Geographical Knowledge Services (ICSDM), 2011 IEEE

International Conference on.

Wang, J., González-Jorge, H., Lindenbergh, R., Arias-Sánchez, P., & Menenti, M.

(2013). Automatic estimation of excavation volume from laser mobile

mapping data for mountain road widening. Remote Sensing, 5(9), 4629-4651.

Wang, J., Zheng, Y., Li, X., Yu, C., Kodaka, K., & Li, K. (2015). Driving risk

assessment using near-crash database through data mining of tree-based

model. Accident Analysis & Prevention, 84, 54-64.

Wang, X., & Abdel-Aty, M. (2006). Temporal and spatial analyses of rear-end crashes

at signalized intersections. Accident Analysis & Prevention, 38(6), 1137-1150.

White, R. A., Dietterick, B. C., Mastin, T., & Strohman, R. (2010). Forest Roads

Mapped Using LiDAR in Steep Forested Terrain. Remote Sensing, 2(4), 1120-

1141. doi: 10.3390/rs2041120

Wild, C. J., & Pfannkuch, M. (1999). Statistical thinking in empirical enquiry.

International Statistical Review, 67(3), 223-248.

Williams, K., Olsen, M. J., Roe, G. V., & Glennie, C. (2013). Synthesis of

transportation applications of mobile LiDAR. Remote Sensing, 5(9), 4652-

4692.

Wulder, M. A., White, J. C., Nelson, R. F., Næsset, E., Ørka, H. O., Coops, N. C., . . .

Gobakken, T. (2012). Lidar sampling for large-area forest characterization: A

review. Remote Sensing of Environment, 121, 196-209.

Xiao, L., Wang, R., Dai, B., Fang, Y., Liu, D., & Wu, T. (2017). Hybrid Conditional

Random Field based Camera-LIDAR Fusion for Road Detection. Information

Sciences.

HT UPM

170

Xie, Z., & Yan, J. (2013). Detecting traffic accident clusters with network kernel

density estimation and local spatial statistics: an integrated approach. Journal

of Transport Geography, 31, 64-71.

Xu, C., & Prince, J. L. (1997). Gradient vector flow: A new external force for snakes.

Paper presented at the Computer Vision and Pattern Recognition, 1997.

Proceedings., 1997 IEEE Computer Society Conference on.

Xu, C., W. Wang, P. Liu, R. Guo & Z. Li (2014) Using the Bayesian updating

approach to improve the spatial and temporal transferability of real-time crash

risk prediction models. Transportation research part C: emerging technologies,

38, 167-176.

Xu, S., Wang, R., & Zheng, H. (2017). Road Curb Extraction from Mobile LiDAR

Point Clouds. IEEE Transactions on Geoscience and Remote Sensing, 55(2),

996-1009.

Xue, B., Zhang, M., & Browne, W. N. (2013). Particle swarm optimization for feature

selection in classification: A multi-objective approach. IEEE transactions on

cybernetics, 43(6), 1656-1671.

Yager, N., & Sowmya, A. (2003, January). Support vector machines for road

extraction from remotely sensed images. In CAIP (pp. 285-292).

Yan, G., Mas, J. F., Maathuis, B., Xiangmin, Z., & Van Dijk, P. (2006). Comparison

of pixel‐ based and object‐ oriented image classification approaches—a case

study in a coal fire area, Wuda, Inner Mongolia, China. International Journal

of Remote Sensing, 27(18), 4039-4055.

Yan, W. Y., Shaker, A., & El-Ashmawy, N. (2015). Urban land cover classification

using airborne LiDAR data: A review. Remote Sensing of Environment, 158,

295-310. doi: 10.1016/j.rse.2014.11.001

Yan, W. Y., Shaker, A., Habib, A., & Kersting, A. P. (2012). Improving classification

accuracy of airborne LiDAR intensity data by geometric calibration and

radiometric correction. ISPRS Journal of Photogrammetry and Remote

Sensing, 67, 35-44. doi: 10.1016/j.isprsjprs.2011.10.005

Yang, B., Fang, L., & Li, J. (2013). Semi-automated extraction and delineation of 3D

roads of street scene from mobile laser scanning point clouds. ISPRS Journal

of Photogrammetry and Remote Sensing, 79, 80-93.

Yang, G. Y. C. (1995). Geological mapping from multi-source data using neural

networks. Geomatics Engineering, University of Calgary.

Yen, K. S., Ravani, B., & Lasky, T. A. (2011). LiDAR for Data efficiency (No. WA-

RD 778.1).

Yilmaz, A., Aci, C., & Aydin, K. (2016). Traffic accident reconstruction and an

approach for prediction of fault rates using artificial neural networks: A case

study in Turkey. Traffic Injury Prevention, 17(6), 585-589.

HT UPM

171

Youn, J., Bethel, J. S., Mikhail, E. M., & Lee, C. (2008). Extracting urban road

networks from high-resolution true orthoimage and lidar. Photogrammetric

Engineering & Remote Sensing, 74(2), 227-237.

Yu, L., Hermann, K. M., Blunsom, P., & Pulman, S. (2014). Deep learning for answer

sentence selection. arXiv preprint arXiv:1412.1632.

Yu, Y., Li, J., Guan, H., Jia, F., & Wang, C. (2015). Learning hierarchical features for

automated extraction of road markings from 3-D mobile LiDAR point clouds.

IEEE Journal of Selected Topics in Applied Earth Observations and Remote

Sensing, 8(2), 709-726.

Yuan, X., Zhao, C. X., Cai, Y. F., Zhang, H., & Chen, D. B. (2008, December). Road-

surface abstraction using ladar sensing. In Control, Automation, Robotics and

Vision, 2008. ICARCV 2008. 10th International Conference on (pp. 1097-

1102). IEEE.

Zai, D., Chen, Y., Li, J., Yu, Y., Wang, C., & Nie, H. (2015, July). Inventory of 3D

street lighting poles using mobile laser scanning point clouds. In Geoscience

and Remote Sensing Symposium (IGARSS), 2015 IEEE International (pp.

573-576). IEEE.

Zampa, F., & Conforti, D. (2009). Mapping with mobile lidar. GIM international,

23(4), 35-37.

Zeng, Q., & Huang, H. (2014). A stable and optimized neural network model for crash

injury severity prediction. Accident Analysis & Prevention, 73, 351-358.

Zeng, Z., Wan, J., & Liu, H. (2016). An entropy-based filtering approach for airborne

laser scanning data. Infrared Physics & Technology.

Zhang, J., Duan, M., Yan, Q., & Lin, X. (2014). Automatic vehicle extraction from

airborne LiDAR data using an object-based point cloud analysis method.

Remote Sensing, 6(9), 8405-8423.

Zhang, Q., & Couloigner, I. (2007). Accurate centerline detection and line width

estimation of thick lines using the radon transform. Image Processing, IEEE

Transactions on, 16(2), 310-316.

Zhang, W. (2010, June). Lidar-based road and road-edge detection. In Intelligent

Vehicles Symposium (IV), 2010 IEEE (pp. 845-848). IEEE.

Zhang, Y., & Yan, L. (2007). Road surface modeling and representation from point

cloud based on fuzzy clustering. Geo-spatial Information Science, 10(4), 276-

281. doi: 10.1007/s11806-007-0106-0

Zhang, Y., Maxwell, T., Tong, H., & Dey, V. (2010). Development of a supervised

software tool for automated determination of optimal segmentation parameters

for ecognition (pp. 5-7). na.

HT UPM

172

Zhao, J., & You, S. (2012). Road network extraction from airborne LiDAR data using

scene context. Paper presented at the Computer Vision and Pattern

Recognition Workshops (CVPRW), 2012 IEEE Computer Society Conference

on.

Zhao, J., You, S., & Huang, J. (2011). Rapid extraction and updating of road network

from airborne LiDAR data. Paper presented at the Applied Imagery Pattern

Recognition Workshop (AIPR), 2011 IEEE.

Zhou, L., & Vosselman, G. (2012). Mapping curbstones in airborne and mobile laser

scanning data. International Journal of Applied Earth Observation and

Geoinformation, 18, 293-304.

Zhou, W. (2013). An object-based approach for urban land cover classification:

integrating LiDAR height and intensity data. Geoscience and Remote Sensing

Letters, IEEE, 10(4), 928-931.

Zur, R. M., Jiang, Y., Pesce, L. L., & Drukker, K. (2009). Noise injection for training

artificial neural networks: A comparison with weight decay and early stopping.

Medical physics, 36(10), 4810-4818.

UNIVERSITI PUTRA MALAYSIA UPMpsasir.upm.edu.my/id/eprint/71397/1/FK 2018 91 IR.pdf · 2019. 9. 3. · dengan jelas dan tepat, teknik pemodelan yang betul harus digunakan. Kajian ini - [PDF Document] (2024)