Research And Design Of Interested Target Detection System In Traffic Surveillance Video

With the development of the science and technology and the continued progress of computer vision, traffic surveillance video system has developed quickly, but how to extract the interested static and moving targets from the actual scene in traffic surveillance video is still the hotspot in the current study. Therefore, this dissertation puts forward a detection method of interested regions and interested targets in traffic surveillance video, the traffic lines and moving targets are detected and color coded. The video compression ratio is improved by using different compression strategy for region of interest and other regions.In the transmission of video data in different scenes, research on the correct segmentation method of effective data and non-effective data, the clarification method of the region of interest and the degradation strategy of other regions, accord with the characteristics of nonlinear transmission of Internet of things and the discrete characteristics of large capacity data, In this dissertation, we study the algorithm of the target detection in traffic surveillance video, and the application system is designed and implemented.The main work of the dissertation is as follows:First of all, for traffic lines in the actual scene, the Canny operator is used to detect the pixel, color, brightness and direction of traffic lines, then traffic straight lines is obtained by Hough Transform and color coded.Secondly, moving target is detected by the background subtraction method. This dissertation presents the detection method of moving targets in detail, design the detection module of moving target.Finally, based on interested target detection system, design the traffic lines detection module, the moving target detection module and video compression module. The performance of these detection module are tested through experiments. The experiments results indicate these detect methods meet the requirements of real-time and accuracy.According to the analysis of system test results, this system can meet the real-time requirements, gets rid of the irrelevant targets, retain the semantic relevant traffic target; using the method of reducing the transmission rate and improving the compression rate, the compression is improved about ten percent. The test results show that the interested target detection system in traffic surveillance video designed in this dissertation meet the requirements of real-time and accuracy, and meet the requirements of practical application.