Kunfeng Wang

Video processing techniques for traffic flow monitoring: A survey

Video-based traffic flow monitoring is a fast emerging field based on the continuous development of computer vision. A survey of the state-of-the-art video processing techniques in traffic flow monitoring is presented in this paper. Firstly, vehicle detection is the first step of video processing and detection methods are classified into background modeling based methods and non-background modeling based methods. In particular, nighttime detection is more challenging due to bad illumination and sensitivity to light. Then tracking techniques, including 3D model-based, region-based, active contour-based and feature-based tracking, are presented. A variety of algorithms including MeanShift algorithm, Kalman Filter and Particle Filter are applied in tracking process. In addition, shadow detection and vehicles occlusion bring much trouble into vehicle detection, tracking and so on. Based on the aforementioned video processing techniques, discussion on behavior understanding including traffic incident detection is carried out. Finally, key challenges in traffic flow monitoring are discussed.

Location estimation in ZigBee Network based on fingerprinting

Location-aware computing becomes an exciting research as recent advancements in RF circuits and wireless communication stacks. In this paper, we present a fingerprinting based location estimation technology in ZigBee network. The system uses the signal strength from several base stations rather than time or angle for determining the location of mobile station. Instead of modeling the complex attenuation of signal strength, the system models the probabilistic distribution in different geographical areas which we called fingerprinting. It combines the measured data and fingerprinting to determine the mobile stations location. The experiment results demonstrate the validity of location estimation in ZigBee network based on fingerprinting.

Vehicle License Plate Recognition Based on Extremal Regions and Restricted Boltzmann Machines

This paper presents a vehicle license plate recognition method based on character-specific extremal regions (ERs) and hybrid discriminative restricted Boltzmann machines (HDRBMs). First, coarse license plate detection (LPD) is performed by top-hat transformation, vertical edge detection, morphological operations, and various validations. Then, character-specific ERs are extracted as character regions in license plate candidates. Followed by suitable selection of ERs, the segmentation of characters and coarse-to-fine LPD are achieved simultaneously. Finally, an offline trained pattern classifier of HDRBM is applied to recognize the characters. The proposed method is robust to illumination changes and weather conditions during 24 h or one day. Experimental results on thorough data sets are reported to demonstrate the effectiveness of the proposed approach in complex traffic environments.

A Review on Vision-Based Pedestrian Detection for Intelligent Vehicles

Vision-based pedestrian detection techniques for smart vehicles have emerged as a hot research topic in the field of vehicular electronics and driving safety. A vision-based system can recognize pedestrians in front of the moving vehicle, then warns the driver of the dangerous situation loudly or slows the vehicle down automatically to protect both drivers and pedestrians. In general, the vision-based pedestrian detection process can be divided into three consecutive steps: pedestrian detection, pedestrian recognition, and pedestrian tracking. In this paper, a great variety of methods associated with these three steps is introduced and compared in detail. In addition, the implementation of vision-based pedestrian detection on vehicles is also presented. In the end, we analyze the difficulties and the research trend in the future.

A vehicle license plate recognition system based on analysis of maximally stable extremal regions

Vehicle License Plate Recognition (VLPR) system is a core module in Intelligent Transportation Systems (ITS). In this paper, a VLPR system is proposed. Considering that license plate localization is the most important and difficult part in VLPR system, we present an effective license plate localization method based on analysis of Maximally Stable Extremal Region (MSER) features. Firstly, MSER detector is utilized to extract candidate character regions. Secondly, the exact locations of license plates are inferred according to the arrangement of characters in standard license plates. The advantage of this license plate localization method is that less assumption of environmental illumination, weather and other conditions is made. After license plate localization, we continue to recognize the license plate characters and color to complete the whole VLPR system. Finally, the proposed VLPR system is tested on our own collected dataset. The experimental results show the availability and effectiveness of our VLPR system in locating and recognizing all the explicit license plates in an image.

Parallel Traffic Management System and Its Application to the 2010 Asian Games

Field data are important for convenient daily travel of urban residents, reducing traffic congestion and accidents, pursuing a low-carbon environment-friendly sustainable development strategy, and meeting the extra peak traffic demand of large sporting events or large business activities, etc. To meet the field data demand during the 2010 Asian (Para) Games held in Guangzhou, China, based on the novel Artificial systems, Computational experiments, and Parallel execution (ACP) approach, the Parallel Traffic Management System (PtMS) was developed. It successfully helps to achieve smoothness, safety, efficiency, and reliability of public transport management during the two games, supports public traffic management and decision making, and helps enhance the public traffic management level from experience-based policy formulation and manual implementation to scientific computing-based policy formulation and implementation. The PtMS represents another new milestone in solving the management difficulty of real-world complex systems.

A review on vision-based pedestrian detection in intelligent transportation systems

Pedestrians are key participants in transportation systems, so pedestrian detection in video surveillance systems is of great significance to the research and application of Intelligent Transportation Systems (ITS). We review some methods and models for vision-based pedestrian detection in recent years. In this paper, the pedestrian detection techniques are divided into macroscopic and microscopic according to different application in transportation systems. Macroscopic pedestrian detection aims to estimate crowd density without distinguishing each pedestrian, and microscopic pedestrian detection focuses on detection and recognition of individual pedestrians. The latter detection style is deeply studied, so it is presented in detail in this paper, especially for the feature-classifier-based detection method. Finally, the pedestrian detection algorithms are discussed and concluded from the viewpoint of video surveillance and ITS. Existing problems and future trends are presented in that section.

A survey of vision-based automatic incident detection technology

Automatic incident detection (AID) has become a necessity for the ever increasing traffic density for most major intersections and highways. Using vision-based AID systems, real-time incident information can be obtained automatically and precisely, and communicated to the Traffic Management Centre (TMC) for other posterior activities such as oncoming driver warning, incident processing and removal. Vision-based AID algorithms generally include three consecutive steps: object detection, vehicle tracking and activity understanding. In this paper, a great variety of vision-based AID methods are introduced and compared in detail. A method is suggested for evaluating performances of AID algorithms. In addition, this paper proposes several key technical difficulties and possible resolutions, which are often met in traffic incident detection process.

A Multi-view Learning Approach to Foreground Detection for Traffic Surveillance Applications

This paper proposes an effective multi-view learning approach to foreground detection for traffic surveillance applications. This approach involves three main steps. First, a reference background image is generated via temporal median filtering, and multiple heterogeneous features (including brightness variation, chromaticity variation, and texture variation, each of which represents a unique view) are extracted from the video sequence. Then, a multi-view learning strategy is devised to online estimate the conditional probability densities for both the foreground and the background. The probability densities of three features are approximately conditionally independent and are estimated with kernel density estimation. Pixel soft labeling is conducted by using Bayes rule, and the pixelwise foreground posteriors are computed. Finally, a Markov random field is constructed to incorporate the spatiotemporal context into the foreground/background decision model. The belief propagation algorithm is used to label each pixel of the current frame. Experimental results verify that the proposed approach is effective to detect foreground objects from challenging traffic environments and outperforms some state-of-the-art methods.

An automated vehicle counting system for traffic surveillance

The objective of this paper is to present a detailed description of using DSP board and image processing techniques to construct an automated vehicle counting system. Such a system has many potential applications, such as traffic signal control and district traffic abduction. We use TITMS320DM642 DSP as the computational unit to avoid heavy investment in industrial control computer while obtaining improved computational power and optimized system structure. The overall software is comprised of two parts: embedded DSP software and host PC software. The embedded DSP software acquires the video image from stationary cameras, detects and counts moving vehicles, and transmits the processing results and realtime images after compression to PC software through network. The host PC software works as a graphic user interface through which the end user can configure the DSP board parameters and access the video processing results. The vehicle detection and counting algorithm is carefully devised to keep robust and efficient in traffic scenes for longtime span and with changeful illumination. Experimental results show that the proposed system performs well in actual traffic scenes, and the processing speed and accuracy of the system can meet the requirement of practical applications.