Computer Vision and Imaging in Intelligent Transportation Systems

Robert P. Loce, Raja Bala, Mohan Trivedi, Robert P. Loce, Raja Bala, Mohan Trivedi

• Publisher: Wiley-IEEE Press
• Year: 2017
• Print ISBN: 9781118971604
• eBook ISBN: 9781118971659
• Edition: 1
• Topic: Tecnología e ingeniería
• Subtopic: Ingeniería eléctrica y telecomunicaciones

Acts as single source reference providing readers with an overview of how computer vision can contribute to the different applications in the field of road transportation

This book presents a survey of computer vision techniques related to three key broad problems in the roadway transportation domain: safety, efficiency, and law enforcement. The individual chapters present significant applications within those problem domains, each presented in a tutorial manner, describing the motivation for and benefits of the application, and a description of the state of the art.

Key features:

• Surveys the applications of computer vision techniques to road transportation system for the purposes of improving safety and efficiency and to assist law enforcement.
• Offers a timely discussion as computer vision is reaching a point of being useful in the field of transportation systems.
• Available as an enhanced eBook with video demonstrations to further explain the concepts discussed in the book, as well as links to publically available software and data sets for testing and algorithm development.

The book will benefit the many researchers, engineers and practitioners of computer vision, digital imaging, automotive and civil engineering working in intelligent transportation systems. Given the breadth of topics covered, the text will present the reader with new and yet unconceived possibilities for application within their communities.

1
Introduction

Raja Bala¹ and Robert P. Loce²

¹ Samsung Research America, Richardson, TX, USA

² Conduent Labs, Webster, NY, USA

With rapid advances in driver assistance features leading ultimately to autonomous vehicle technology, the automobile of the future is increasingly relying on advances in computer vision for greater safety and convenience. At the same time, providers of transportation infrastructure and services are expanding their reliance on computer vision to improve safety and efficiency in transportation and addressing a range of problems, including traffic monitoring and control, incident detection and management, road use charging, and road condition monitoring. Computer vision is thus helping to simultaneously solve critical problems at both ends of the transportation spectrum—at the consumer level and at the level of the infrastructure provider. The book aims to provide a comprehensive survey of methods and systems that use both infrastructural and in‐vehicle computer vision technology to address key transportation applications in the following three broad problem domains: (i) law enforcement and security, (ii) efficiency, and (iii) driver safety and comfort. Table 1.1 lists the topics addressed in the text under each of these three domains.

Table 1.1 Taxonomy of problem domains and applications described in the book.

Problem domains	Applications and methods	Imaging system employed
Law enforcement and security	License plate recognition for violations	Infrastructure
	Vehicle classification	Infrastructure
	Passenger compartment violation detection	Infrastructure, in‐vehicle
	Moving violation detection	Infrastructure
	Intersection monitoring	Infrastructure
	Video anomaly detection	Infrastructure
Efficiency	Traffic flow analysis	Infrastructure
	Parking management	Infrastructure, in‐vehicle
	License plate recognition for tolling	Infrastructure
	Passenger compartment occupancy detection	Infrastructure, in‐vehicle
Driver safety and comfort	Lane departure warning	In‐vehicle
	Collision avoidance	In‐vehicle
	Pedestrian detection	In‐vehicle
	Driver monitoring	In‐vehicle
	Traffic sign recognition	In‐vehicle
	Road condition monitoring	In‐vehicle, infrastructure

This chapter introduces and motivates applications in the three problem domains and establishes a common computer vision framework for addressing problems in these domains.

1.1 Law Enforcement and Security

Law enforcement and security are critical elements to maintaining the well‐being of individuals and the protection of property. Societies rely on law enforcement agencies to provide these elements. Imaging systems and computer vision are means to sense and interpret situations in a manner that can amplify the effectiveness of officers within these agencies. There are several common elements shared by computer vision law enforcement and security applications, such as the detection and identification of events of interest. On the other hand, there are also distinctions that separate a security application from law enforcement. For instance, prediction and prevention are important for security applications, while accuracy and evidence are essential for law enforcement. In many cases, modules and components of a security system serve as a front end of a law enforcement system. For example, to enforce certain traffic violations, it is necessary to detect and identify the occurrence of that event.

Consider the impact of moving vehicle violations and examples of benefits enabled by computer vision law enforcement systems. There is a strong relationship between excessive speed and traffic accidents. In the United States in 2012, speeding was a contributing factor in 30% of all fatal crashes (10,219 lives) [1]. The economic cost of speeding‐related crashes was estimated to be $52 billion in 2010 [2]. In an extensive review of international studies, automated speed enforcement was estimated to reduce injury‐related crashes by 20–25% [3]. The most commonly monitored moving violations include speeding, running red lights or stop signs, wrong‐way driving, and illegal turns. Most traffic law enforcement applications in roadway computer vision systems involve analyzing well‐defined trajectories and speeds within those trajectories, which leads to clearly defined rules and detections. In some cases, the detections are binary, such as in red light enforcement (stopped or passed through). Other applications require increased accuracy and precision, such as detecting speed violations and applying a fine according to the estimated vehicle speed. There are other deployed applications where the violation involves less definitive criterion, such as reckless driving.

Several moving violations require observation into the passenger compartment of a vehicle. Failure to wear a seat belt and operating a handheld cell phone while driving are two common safety‐related passenger compartment violations. Seat belt use in motor vehicles is the single most effective traffic safety device for preventing death and injury to persons involved in motor vehicle accidents. Cell phone usage alone accounts for roughly 18% of car accidents caused by distracted drivers [4]. In addition, the National Highway Traffic Safety Administration (NHTSA) describes other behaviors resulting in distracted driving, including occupants in the vehicle eating, drinking, smoking, adjusting radio, adjusting environmental controls, and reaching for an object in the car. The conventional approach to enforcement of passenger compartment violations has been through traffic stops by law enforcement officers. This approach faces many challenges such as safety, traffic disruption, significant personnel cost, and the difficulty of determining cell phone usage or seat belt usage at high speed. Imaging technology and computer vision can provide automated or semiautomated enforcement of these violations.

Security of individuals and property is another factor in the monitoring of transportation networks. Video cameras have been widely used for this purpose due to their low cost, ease of installation and maintenance, and ability to provide rich and direct visual information to operators. The use of video cameras enables centralized operations, making it possible for an operator to “coexist” at multiple locations. It is also possible to go back in time and review events of interest. Many additional benefits can be gained by applying computer vision technologies within a camera network. Consider that, traditionally, the output of security cameras has either been viewed and analyzed in real‐time by human operators, or archived for later use if certain events have occurred. The former is error prone and costly, while the latter has lost some critical capabilities such as prediction and prevention. In a medium‐sized city with several thousand roadway cameras, computer vision and video analytics allow a community to fully reap the benefits of analyzing this massive amount of information and highlighting critical events in real‐time or in later forensic analysis.

In certain security and public safety applications, very rapid analysis of large video databases can aid a critical life or death situation. An Amber Alert or a Child Abduction Emergency is an emergency alert system to promptly inform the public when a child has been abducted. It has been successfully implemented in several countries throughout the world. When sufficient information is available about the incident (e.g., description of captor’s vehicle, plate number, and color), a search can be conducted across large databases of video that have been acquired from highway, local road, traffic light, and stop sign monitoring, to track and find the child. Similar to Amber Alert and much more common is Silver Alert, which is a notification issued by local authorities when a senior citizen or mentally impaired person is missing. Statistics indicates that it is highly desirable that an Amber‐/Silver Alert‐related search is conducted in a very fast and efficient manner, as 75% of the abducted are murdered within the first 3 h. Consider a statement from the US West Virginia code on Amber Alert 15‐3A‐7:

The use of traffic video recording and monitoring devices for the purpose of surveillance of a suspect vehicle adds yet another set of eyes to assist law enforcement and aid in the safe recovery of the child.

Human analysis of video from thousands of camera could take many days, while computer vision methods have the potential to rapidly extract critical information. The speed can be scaled by the available computational power, which is rapidly advancing due to high‐speed servers and cloud computing.

Whether it is safety of individuals or security of property, recognition of a vehicle is a key component of a roadway security system. Vehicles traveling on the public roadways in most countries are required by law to carry a c...

Table of contents

• Cover
• Title Page
• Table of Contents
• List of Contributors
• Preface
• Acknowledgments
• About the Companion Website
• 1 Introduction
• II: Imaging from the Roadway Infrastructure
• IIII: Imaging from and within the Vehicle
• Index
• End User License Agreement