Study Of Moving Target Tracking Technology In Clutter Background Based On SOPC

This research originates from the development of the tracking system for ground targetbased on multi-rotor type aircraft. While the aircraft is tracking the ground target, the aircraft,the camera and the target are all moving. Because of the inherent constraints of the multi-rotortype aircraft and the complexity of the environment, the existing target tracking technology isunable to meet the requirements. Therefore, it’s necessary to study on moving target trackingtechnology in the clutter background.The concrete research content is as follows:（1） In the existing complex environment, the thorough research and comparativeanalysis mainly focus on the existing situation of target detection tracking technology underthe dynamic background. This article summarizes the situation of study on target detectionand tracking and analyzes relevant technical difficulties. Suitable algorithm for multi-rotoraircraft target tracking system is chosen and modified, and the revamping process includestwo aspects, reforming on algorithm principle and algorithm implementation.（2）The TLD target tracking algorithm is studied and analyzed deeply. The correlativetechnique involved in the algorithm is analyzed and studied, including template matching andcorrelation coefficient method, forward-backward error, median flow tracker, windowscanning, EM method, self-learning, co-training and P-N learning method. Moreover, thesystem architecture, system flow and functional modules of the original TLD algorithm arestudied with the in-depth analysis and research as well.（3）The algorithm is modified. Based on the deep research of TLD algorithm, theoriginal TLD algorithm is modified. The original TLD algorithm is mainly modified on theintegrated module. The experiments prove that under this premise, the identification of targettracking is optimized and promoted to some extent without influencing real-time performance.（4）The implement method of the algorithm is modified. The program is transplantedinto System On Programmable Chip platform. The experimental testing and verification isimplemented on the established SOPC with using standard test video, which leads to analysisabout stability, accuracy, real-time, robustness, etc. The main work content coverstransplantation of Linux operating system, transplantation of OpenCV and cross-compiling.（5） Through hardware-software co-design on the SOPC platform, the algorithm isfurther modified and optimized.