Design Of Dynamic Target Tracking System Based On FPGA

Target tracking technology plays a key role in computer vision,and it is very important to improve the performance of target tracking system in practical application.Compared with CPU,FPGA can greatly improve the calculation speed of the algorithm,and the system performance can be greatly improved by using FPGA for system design.The most important performance index of target tracking system is real-time and reliability.in this paper,a dynamic target tracking system based on FPGA is designed from the two performance indexes of improving realtime and reliability.The traditional target tracking algorithm may be affected by external interference,which makes the tracking accuracy can not meet the requirements.To solve this problem,a Mean Shift tracking method based on gradient vector flow model is adopted.In this paper,the software and hardware platform is built,and then the mainstream target detection algorithm,traditional template matching algorithm and Mean Shift target tracking algorithm are analyzed experimentally.the corresponding efficiency can be clearly observed from the accuracy curve of each algorithm,and Mean Shift algorithm is selected as the basic algorithm for systematic implementation.The target detection algorithm uses the background difference method to extract the initial outline of the target as the initial iterative point of the target tracking algorithm.Because the original Mean Shift algorithm will cause the target tracking failure due to camera jitter and other reasons in this system,the gradient vector flow model is introduced to correct the target,so that the tracking can be carried out fluently.The correction process obtains the target profile by calculating the mean square error,discriminates to the lost target profile and determines the position of the lost target.Using this method,the efficiency of Mean Shift algorithm for target tracking can be improved.This can greatly improve the real-time performance of the system.Finally,the FPGA implementation and experimental analysis of the tracking algorithm are carried out.In this paper,the semi-ping-pong operation is used to solve the problem that the processing speed of the internal algorithm is not synchronized with the data transmission in the internal data transmission of the system.After many tests,it is proved that the algorithm studied in this paper improves the robustness of the system.