Design Of Automatical Red Light Running Detection System Based On DSP

With the rapid development of China's economy, the city high-speed traffic has been increased; the road traffic has becoming increasingly crowded. Realizing urban intelligent real-time traffic control is very important to traffic information collection, traffic management and reducing traffic accidents. Automatic detection red light running system is an important component of city intelligent transportation systems, hasing incomparably advantages compared to the old methods, is a new development direction of vehicle detection system.Traditional vehicle detection is mainly realized by magnetic induction coil buried in the ground; this method causes road damages, not facilitates for the construction and installation, and must keep maintenance constantly. With the rapid development of video technology and doubled computer speed, computer-based video detection system for traffic monitoring has become possible. Computer video traffic detection system has advantages than the old one, it is easy to get installation, causing no damage to the roads, the maintenance is also very simple, have no impact on road traffic conditions.,most important, it can get more informations.The digital image processing based video detection method is the development direction of the vehicle detection.This article focuses on the algorithm, hardware systems and software design of DSP-based red light running automatically detection system.First of all, this essay discussed some of the current popular vehicle detection algorithm, including the context differential based detection methods; optical flow field based detection methods, frame difference based detection methods, analyzing the advantages and disadvantages of these algorithms. Based on these analyses, the article made the video detection algorithm based on the virtual coil,neighborhood region compared three-step matching method. Vehicles detection algorithm based on the virtual coil is one of the methods based on the region sensor technology; it is a simulation of actual magnetic induction coil laying under the road , only process the image of the region by placing detection regions in the video image. This essay researched on the red light running detecting system reflected by three critic moments, which include image pre-processing algorithm, neighborhood region compared and three-step matching method based on it.Secondly, this essay introduced the main hardware modules of this system in details. This entire hardware system based on the TMS320DM642, composes the hardware platform together with external video input and output module, Ethernet module and serial port modules, fulfilling video data collection and outputing, video data processing and network transmission.Finally, this essay summarized the software architecture of this system, based on the DSP/BIOS development tools and TI's RF5 Framework, using NDK development tools provide by TI to design DSP server software, realizing parameters data transportion extracted by this vehicle detecting algorithm.The client software of this detection system is based on VC++6.0 as development platform, achieving monitoring interface design, receiving real-time data and video real-time playback. The author selected different road traffic video record detected by this system for playback, results showed that when the traffic situation is not very severe, the algorithm is able to accurately determine the vehicle information, and capture the pictures timely.