Research On Capture System Of Moving Objects Based On Davinci Platform

With the society’s development, target detection and tracking technology in the field of video surveillance has been a growing number of applications, however, the traditional target detection and tracking algorithm, the researchers tend to pay more attention to the effectiveness of the algorithm, while weakening the right algorithm for real-time considerations, which makes the algorithm implemented in the project depends on the high cost of processing equipment, and eventually the entire cause the system expensive. In order to solve the above problems, this paper provides algorithms and code optimization two-pronged approach which can construct a higher cost DAVINCI platform to acomplish the moving object capture system.In this paper, we study a real-time and high portability of moving target tracking algorithm. Specifically, this paper’s Surendra algorithm generate real-time background video image, and then extract the moving object using background difference, and ultimately the use of run-connectivity algorithm locking and tracking moving targets. During construction of the algorithm, this paper combines a large simulation, analysis of the important parameters of the target capture algorithm performance. For example:Background generation algorithm iteration factors on the impact of background update rate, threshold selection for image binarization effect, and the maximum number of iterations of the algorithm of real-time impact.Secondly, we will examine the algorithms to versatility, cost-effective DAVINCI platform up, and according to characteristics of the algorithm and platform characteristics of the code has been optimized to obtain a real-time with a good, easy to implement projects moving objects capture system. Specifically, the code in the loop for large time overhead image processing loop structure and more features, we use loop unrolling to reduce the code running time; DAVINCI platform for fixed-point arithmetic efficient and moving target detection algorithm on the data accuracy requirements the actual situation is not high, this article will use data Q format convert floating point to fixed-point arithmetic, which makes real-time code is enhanced.The final test results show that the proposed method can achieve capture of moving objects and the validity and timeliness can meet the engineering use.