Research Of System Of Moving Object Detection And Tracking Based On DSP

Visual information,especially dynamic visual information, accounts for most of environmental information perceivable to human being. Therefore, perceiving dynamic visual information and processing these information is one of the most important research area in computer vision.The thesis focuses on the study of moving object detection and tracking on the platform of DSP. The human face is taken as the object in this study. The thesis includes the architecture design of DSP-based system, detection algorithm, the face tracking algorithm and the implementation using combination of pyramid search and sequential similitude method. Themajor works are shown as following:1. A number of moving object detection and tracking algorithms are introduced and analysed;2. Select face as moving object, use skin color model to extract face, and search the selectedface using detecting algorithm in YUV space, to obtain moving objects object;3.Designed Designed the pyramid and sequential similitude algorithm software, thisapproach is based on pyramid matching algorithm and sequential similiarity detectionalgorithm, we called this approach as coarse-to-fine matching algorithm, it used pyramidmatching algorithm to get coarse area, then get the fine result with sequential similitudealgorithm;4.The framework of the DSP-based moving target detection and tracking system includingsoftware diagram, Interface design, transplant of algorithm to the DSP platform, are presented. The algorthm is tested, some useful experiment results are achieved, MiniGUI is implemented with the Davinci DSP, laying foundation for future works.