Implementation Of Moving Target Detection And Tracking System Based On Embedded Platform

In recent years, the moving target detection and tracking is one of the hot topics of concern in the field of computer vision, it is to detect, recognize and track the moving targets from image sequences containing moving objects, and then to understand and describe their behaviors, and which is widely used in intelligent video surveillance. With the wisdom of life, the concept of smart home advocate, which is now facing the combination of networking technology and embedded technology continues to progress, and it both has good prospects in scientific research and practical application, thus the study, design, development and implement of moving target detection and tracking system based on embedded platform, has become a hot research topic of the researchers.In this paper, the moving target detection algorithm and tracking algorithm are studied and simulated in the static background, and the detection algorithm and tracking algorithm are improved and optimized in view of the common algorithms. Finally, the detection and tracking of moving objects are realized on the ARM embedded development board.In the aspect of moving object detection, this paper mainly studies three methods of continuous frame difference, background subtraction and optical flow method, and the simulation experiment is carried out in Linux and Qt development environment. Based on the experimental results and embedded system hardware resources, discuss the advantages and disadvantages of each algorithm. On top of this, improve continuous frame difference method for three frame difference method, and the edge detection operator is introduced and evaluated objectively. In the end, this paper designs a moving target detection algorithm which is suitable for embedded system based on the combination of three frame difference and background edge detection.In the aspect of moving target tracking, this paper mainly studies the moving target tracking algorithm based on MeanShift and CamShift, and the simulation experiment is carried out in the Linux and Qt development environment. Based on the experimental results and embedded system hardware resources, discuss the advantages and disadvantages of each algorithm. On top of this, Kalman filter is introduced and evaluated objectively. In the end, this paper designs a moving target tracking algorithm which is suitable for embedded system based on CamShift and Kalman filter.In terms of embedded system function, first step is to design and make the S3C2440 A as the core of ARM development board, build the cross compiler environment, embedded Linux operating system and so on, in application layer, complete the migration of graphical interface library Qt and computer vision library OpenCV. Then the moving target detection and tracking applications are modular design, complete video capture, moving target detection, image pre-processing, moving target tracking, Qt interface design, achieve the moving target detection and tracking in the embedded platform. And through the Socket network communication, the moving target detection and tracking results can be transmitted to the host in real time, so that you can achieve remote monitoring and operation purposes.Experimental results prove that the improvement and optimization of moving target detection and tracking algorithm in the ARM board with S3C2440 A processor as the core is feasible and reliable, and has better robustness for the static background, and has wide application prospect.