Moving Target Detection And Tracking Algorithm Research On Embedded Platform

In recent years, with the rapid development of computer vision technology, embedded technology and communication technology, the intelligent video surveillance technology has been more and more widely used in many fields of society. Moving object detection and tracking is not only one of the core technology of intelligent video surveillance technology, but also the basis for intelligent video surveillance applications, which currently is moving forward embedded technology and communications and networks technology rapid development. This research have very good prospects in scientific research and practical application, so based on embedded intelligent video surveillance technology to attract the majority of researchers has become a prominent research focus.The thesis has finished a deep theoretic study and sufficient experimental works in moving object detection and tracking in static background, and design improved algorithms. The main work of this thesis is as follows:On research of the moving object detection, the first analyze several the currently main three algorithms of motion detection:optical flow method, background subtraction, inter-frame difference method, in which focus on the advantages and disadvantages of the inter-frame difference and background subtraction, improve both, and combine with the edge detecting operators, and evaluate the performance of three edge detection operators objectively, and then design a motion object detection algorithm based on background edge-characteristic difference combining five inter-frame difference. The experimental results show that the algorithm can eliminate the shadow and inanition and quickly extract a complete, accurate contour of the moving target, and make a good foundation for the moving object tracking.On research of moving object tracking, the first introduce different moving object tracking algorithms, focus on the Mean Shift algorithm, then by existed disadvantages of the traditional Mean Shift the problems which included serious occlusion and moving fast, to pinpoint object’s centroid and monitor Bhattachayy coefficient changes, and then a moving object tracking algorithm based on the Kalman filter combining Mean Shift to solve the disadvantages. Lastly the experimental results show the algorithm can track object accurately and stably, compared with the traditional Mean Shift and the improved algorithm, the improved algorithm has been significantly improved.On research of system functions, the first transplant the embedded operate system, the computer vision library OpenCV and GUI library Qt into embedded development board, then design modular software, write the programs, lastly achieve the system functions. Finally, the experiments show that the improved moving target detection and tracking algorithms to improve the adaptability of the algorithm in a complex environment, can achieve accurate tracking of moving object, and have wide application prospects.