The Design And The Achievement Of Target Tracking System Based On Embedded Technology

At present days,the intelligent supervisory system based on the embedded technology had been the substuition of the video-supervisory gradually with the development of the embedded technology and the hardware resources,the functions of the intelligent supervisory,which were different to the traditional one,are more than to extract the target information from the video sequence just like the motion detection,the position detection,the posture detection and the behivior detection so on,meanwhile,the property of immediacy and the robustness are also need to be advanced,under this circumstance,the performance of the embedded system must be guaranteened restrictly.This paper is aimed to construct the embedded system to the application of target tracking,we had applied the ARM-A9 to the hardware layer and to applied technology of the computer vision to the software layer,in the end,to applied the Linux os to the interlayer,which are vital to enhance the minlaturization,the immediacy and the robustness of the supervisory system,the jobs of this paper will be listed below this section.The major work:We had made some analysis of the current moving target detection algorithms and to figure out the principles of these algorithms according to these relative literature;To analyze the advantages and the drawbacks while they were applied to the embedded system,and then to analyze the basical theory of the compressed sensing and the application in the filed of moving target detection,To analyze the present problem of this method and to propose the optimization schemes;In the end,we had selected the develop board based on ARM-A9 and then transplant the Liunx os and OpenCV to the develop board.At the end of this paper,we had made some analysis for this system which were mainly about the speed and the accuracy,we can draw a conclusion from the consequnce of the testing that this system had tracking the target accurately,therefore,this system can also deduce the influence of the error,beside this,the performance of the real-time and the stability are advanced as well.