Shuming Tang

DynaCAS: Computational Experiments and Decision Support for ITS

Accurate, reliable, and timely traffic information is critical for deployment and operation of intelligent transportation systems (ITSs). Traffic forecasting for travelers and traffic operators should become at least as useful and convenient as weather reports. In the US, the Federal Highway Administration (FHWA) has envisioned a real-time traffic estimation and prediction system (TrEPS) as an ITS support platform that resides at traffic management centers (TMCs) for dynamic route assignment (DRA) and other transportation operations.

Automatic Detection and Recognition of Circular Road Sign

Road sign detection and recognition is an important research issue of driver assistance systems. This paper proposes a two-stage automatic detection and recognition algorithm for the most common road signs in China. In the first phase, color segmentation and Hough transform are used for ROI detection. Considering the difference between the actual road signs and standard road signs, the pictograph on road signs is extracted for recognition. Adaptive Hausdorff distance based on similarity weighting is employed in the recognition phase. The experiment results show the effectiveness for recognition and the insensitivity to slight occlusion.

Pedestrian Detection Using Boosted HOG Features

This paper presents a novel approach in pedestrian detection in static images. The state-of-art feature named histograms of oriented gradients (HOG) is adopted as the basic feature which we modify and create a new feature using boosting algorithm. The detection is achieved by training a linear SVM with the boosted HOG feature. We experimentally demonstrate that our solution achieve comparable performance as the HOG algorithm on the INRIA pedestrian dataset yet considerably reduce storage requirement and simplify the computation in terms of elementary operations.

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 software architecture for artificial transportation systems - principles and framework

Artificial Transportation Systems(ATS) is an extension of the traffic micro-simulation, which integrates the transportation system with other urban systems, such as logistic systems, social and economic systems, etc., to behave as a coordinated tool for transportation analysis, evaluation, decision-making, and training. Research, especially implemental work, on ATS is far from enough. In this paper, a software architecture for ATS is proposed. Four principles of object-oriented software engineering, considerations for computational experiments and parallel systems, issues of expansibility and cooperative development for the software design and implementation, as well as a software framework, which explains the functional structure of ATS are addressed. Furthermore, the rationale behind the framework is explained, with emphasis on the discussion about transportation scenarios, the agent characteristics of travelers, and the spread and flow of information in ATS.

Modeling and analysis of artificial transportation system based on multi-agent technology

This paper presents the multi-agent architecture for artificial transportation system. In this architecture, Petri net is used as basic model to represent agents. At an intersection, the agents are divided into two groups: one for the traffic-signal and the other for the vehicle flow, which are integrated to represent the behavior of this intersection. In addition, those agents can be used as the modularity to represent urban network of more scale. To coordinate different intersection agents, game theory is used to design coordination strategy between agents. The iterated elimination of strictly dominated strategies algorithm is presented to find Nash equilibrium

A 2DLDA Based Algorithm for Real Time Vehicle Type Recognition

Vehicle type recognition generally refers to identifying the make and model of vehicles through collected images or other information. This paper presents a Two-Dimensional Linear Discriminant Analysis (2DLDA) based algorithm for real time vehicle type recognition. The algorithm is initiated by extracting a region of interest (ROI) relative to the located license plate, and then robust features obtained by performing 2DLDA on the gradients of ROI are used for recognition. Experimental results show that the algorithm can achieve high recognition accuracy (94.7%) and that it is not sensitive to variations of vehicle colors and light conditions. The computational complexity is low so that the algorithm can be implanted in real time.

Design of an OSEK/VDX and OSGi-based embedded software platform for vehicular applications

This paper addresses an OSEK/VDX and OSGi-based embedded software platform for vehicular applications. Basically, it is a wireless and networked distributed service system and consists of (1) an OSEK/VDX-based vehicular application specific embedded operating systems platform; (2) an OSGi-based automotive service system which is composed of automotive central service platform, remote vehicular service platform and vehicle/home interactive platform. The system is deployed by integrating the functions of communication, navigation, entertainment, diagnostics, monitoring and many others. Meanwhile, by using wireless network, it realizes the interconnection among vehicles, home service platform, remote vehicular service center and roadside systems.

An Implementation of Artificial Transportation Systems based on JXTA

An artificial transportation system is a new type of traffic simulation software that employs agent modeling, distributed computing and concept of artificial societies to support the research and development in transportation. In this paper, models of vehicles, transportation infrastructures and a shopping center are described first, and the design and implementation of the basic ideas of artificial transportation systems based on a framework using P2P computing platform JXTA are stated in steps. Future work and issues related to operational mechanisms and strategies for such implementation are also addressed.

Moving Object Refining in Traffic Monitoring Applications

Moving object segmentation is an important task in vision-based traffic monitoring applications. In traffic scenes, various outliers such as sudden illumination changes, moving cast shadows, camera jitter, etc., often cause serious errors in image analysis due to misclassiflcation of moving objects. An efficient moving object refining approach is thus expected. In this paper, we address the problem of moving object refining by processing the background subtraction results. In an analytical multi-stage procedure, we remove sudden illumination changes and local reflected regions employing photometric color invariants, remove moving cast shadows based on a single Gaussian shadow model, uniform-region classification, and spatial analysis, and further remove other types of outliers in a postprocessing stage of area filtering and area-to-perimeter test. Experimental results on actual video sequences representative of different traffic scenes and illuminations are presented. The results illustrate that our approach is efficient when handling widely different conditions that can occur.