The Preliminary Study And Application Of Traffic Detection Algorithm

Video traffic detectors have obvious advantages, it can make up their own shortcomings and provide a comprehensive means of detection, which automatically extracted the vast majority of the traffic flow data valid, including traffic flow, speed measurement, Length of the queue of vehicles, vehicle classification, as well as the share of density, but also be able to achieve automatic detection of traffic accidents; video has detected a large area detection features, from different regions, can be integrated analysis, quickly and conveniently Traffic control centers for data traffic control or direct the output signal for automatic management and control. In addition, the video detection has its own characteristics that can provide visual surveillance capability, the realization of visual information and records preservation, with off-line processing. Video detector is installed without the use of physical contact with the road, do not undermine the road without interference thousands of normal traffic, easy to maintain. To sum up, video detection technology can be applied flexibly in different areas, from the complex analysis of the traffic flow to the traffic management and planning, are very effective.Of course, should also be noted that in the face of the complexity of the application of ITS, video detection is still deficient, which on the one hand, by a number of objective conditions and constraints such as poor environmental conditions caused by the importation of lower quality images, or due to the acquisition of equipment Dynamic and resolution is limited, may be caused by an error detection or identification of sexual ambiguity. In addition, image processing algorithms on the defects brought by deviate from the test results, such as shadow and block the view of the algorithm will be brought to the interference. These deficiencies are revealed its future development direction of research, a dedicated hardware image processing to speed up the development, research more adaptive algorithm, the establishment of a more flexible mathematical model is the technology research efforts.Adaptive Gaussian background to introduce hybrid model of the background initialization model, background model update, the prospects of regional issues such as extraction, and in accordance with the crowded traffic flow characteristics, based on the background information campaign model to improve the flow of traffic congestion detection effect. Prospects in the region when the extraction due to the impact of noise, the prospect of images obtained with salt and pepper noise, we use a simple method to determine the eight neighborhood, has greatly improved the image of the treatment.Now the existence of the shadow of applications for video detection much trouble, in order to improve the detection results in the fourth chapter describes in detail a variety of detection methods. This chapter in the current detection method based on the RGB model invariance, and to simplify the division of the region on the basis of several proposed by the shadow of the adaptive threshold detection model to improve the detection of a shadow effect, increased the flexibility of model Sex. Finally, we use a simplified method to fill the region to further improve the detection of the shadow.The video detection technique is an important research field of ITS. There many work include in the research such as vehicle detection, vehicle statistic, real time tracking, speed measurement, etc.Real time segmentation of moving objects in image sequence is a crucial step in traffic surveillance. Base on the foundation of Previous work we did a great deal of experiment, and improve the Adaptive background mixture models according to the characteristics of the video vehicle detection, putting forward the background model according to the sport information.Consider the complexity of the environment, especially the influence of object shadow to video vehicle detection, this text still inquiries into the existing shadow detection model, and combine the color characteristics of the shadow, we still design a RGB constant detection model, RGB constant detection model base on EM algorithm. This had improved the result of shadow detection. Still we put forward the simplification algorithm for Image segmentation and area-fill in the design process, guarantee the movement efficiency of the system.