Research On Moving Target Detection And Tracking Methods Based On Video

With the rapid development of machine vision related hardware and the increasingly urgent current public security situation, the intelligent visual security systems are playing more and more significant role in almost all aspects of social life today. Being the most fundamental and pivotal research content in intelligent video monitoring system, the moving target detection and tracking technology, although after many years of research and development by numerous researchers at home and abroad who has achieved many breakthroughs, due to complicated actual situation,moving target's mutation status,occlusion and other factors,there is still no one method can be perfectly proposed to adapted to all complex scenes, so it also has extremely important research significance and practical value in future. To continuous explore and optimize the existing research results is just the only way to gradually improve the moving target detection and tracking technology, based on this point, this thesis mainly hold the idea that around the moving target detection and tracking method based on video to have a deep research, the main contents are as follows:1.Further demonstrates the video image sequence preprocessing technology. Mainly from color image gray processing, smoothing, image edge extraction, image morphology processing and other preprocessing techniques to explore and discuss.2.Deep study and compare the current mainstream moving target detection method. Mainly focus on the background difference method, frame difference method and optical flow method to have a deep analysis and verification, compare their respective advantages and disadvantages; Combined with edge detection technology to improve frame difference method, proposed fusion edge detection of three frame difference and four frame difference moving target detection algorithm. Under the existing theoretical framework, a variety of traditional methods are integrated to achieve a combination of their respective advantages to make up for their own shortcomings. Based on this thought, the author makes an innovative way to make the background subtraction and four differential methods be fused, obtain the fusion of four differential background subtraction moving object detection algorithms. The simulation results demonstrate the effectiveness of the proposed detection algorithm in this thesis.3.Deep research the method of removing the shadow region in moving target detection. In view of the series of problems that it is difficult to determine the threshold value in the traditional HSV space shadow removing model and complex calculation process as well as the low detection efficiency, using the first-order gradient information to targeted improve its shortcomings. Thereafter, it puts forward the joint shadow removal algorithm based on improved HSV space and reflection ratio invariants. The simulation results of multiple group experimental data show that the proposed algorithm has a certain practical value.4.Further research and analysis the current mainstream moving target tracking method. Classifying existing tracking methods and summarizes their respective advantages and disadvantages. Derivation and simulation discussed the Mean Shift tracking method, Kalman filter algorithm and particle filter algorithm. As for moving target state mutation and large proportion occlusion circumstances, innovatively put forward an fusion improved MS algorithm with Kalman filtering of double template matching double determine algorithm. The simulation results of multiple group experimental data shows that, compared with other similar algorithms, the proposed algorithm has higher movement target tracking accuracy and robustness.