Design and Implementation of a DSP-Based Embedded Intelligent Traffic Surveillance System

Abstract

Increasing congestion on urban streets and freeways boosts the demands of more powerful and economical traffic surveillance equipment. Rapid development of Digital Signal Processor greatly facilitates the implementation of faster, easier and more reliable video and image processing system. In this paper we present a novel embedded platform based on Texas Instruments Davinci^TM Digital Media Processor, dedicated particularly to the remote surveillance of urban intersections under harsh environmental conditions, featuring vehicle detection and counting methods, and various video and image compression algorithms as well as support for local systems and devices. The system is composed of Video Server Module and Video Analysis Module. Signal Integrity (SI) problem is carefully analyzed and simulated during hardware design process to maintain the system stability and performance. A flexible video and image analysis, focusing on vehicle detection using background extraction and blob analysis, as well as video compression and decompression algorithm, and Ethernet data communication software running in the embedded platform is presented and tested that the software can enhance the accuracy of vehicle detection and license plate recognition.