Motion Detection Research Based On DM642 Network Communication

With the development of computer technology, digital image processing technologies and some other related theory, Computer vision technology has improved continuously, applications fields are rapidly widening, and is used in defense aerospace, industrial automation, traffic surveillance and so on. In the field of computer vision, motion detection and tracking are researched most and the application level is the highest. It is the important process of motion image analysis, intelligent surveillance and visualized human-machine interactive. One can gain motion information and extracting the motion object from the image through motion detection, and then locating and tracking, and motion detection is widely used in the security field of bank, transport and airport.The main content of this design Motion Detection Research Based on DM642 Networking Communication is based on the DSP technologies, and then a motion detecting system based on embedded networking communication is realized. The hardware and software design of this system is completed, and seeking a practical algorithm of motion detecting at the same time.The major work of this paper is as follows:1. Software and hardware design of embedded video system based on DSP networking transmission: a video surveillance system design and realization based on TI's TMS320DM642 is introduced, the hardware and software composition of the system is discussed, and analyzed the task scheduling of the operating system in TI's reference framework RF5 and also realization of TCP/IP protocol based on TI's NDK. Results show that the system can compute the real-time images captured by the camera, and the results can be transmitted via Ethernet to the client. This system can not only provides a hardware and software platform for various video processing algorithms, but also achieve remote monitoring, local display, local storage, and some other functions.2. Research of the motion detection algorithms: Background difference method for motion detection is used in this paper, a background model selection and adaptive updating method are also proposed, and applies particularly in the scenes of objects starting to move and stop. This method is used to dealing with a transport video in VC++ platform and the results show that the algorithm can avoid certain dynamic change and interference to some extent, and has a good stability and versatility.