Moving Object Detection And Tracking Based On Embedded System

In recent years,many excellent algorithms and papers have came out in moving object detection and tracking areas.And these were used in many fields including monitoring,unmanned aerial vehicle,biomedical science,and even navigation and guidance.However,restricted by the complex algorithm,a powerful processor is needed to release this system,for example a personal computer is a minimum requirement.This paper studied in some common algorithm,and hoping to releasing a moving object detection and tracking system on the embedded platform.On object detection,this paper analyzed background subtraction method and frame difference bmethod,at last realized a method through improve three frame difference method.This method can find the object which moving slowly but hardly increase the complexity of algorithm.On object tracking,this paper has studied center of area method,mean shift method and camshaft method,and simulated 3 methods,finally,the kalman method is used to improve the camshift algorithm to solve the problems of similar color interference and occlusion.Througth above analysis and summary,we released a system on the embedded platform based on S3C6410,which achieved a good object tracking effect,and also ensure the real-time performance.