The Design And Realization Of An Embed Color TV Tracker

In this theses, the history and actuality of TV tracking was introduced; the hardware and software foundations of realizing an Embed Color Video Tracker was analyzed; the working theory and structure of a TV tracking system was also expatiated; the key technology of realizing an Embed Color Video Tracker is presented. According to the demands of practical applications, an embed color video tracking platform was designed. An algorithm for separating the target, getting its characters and then identifying it using wavelet transform is presented. Using the embed color video tracking platform a lot of experiments have been done, and the satisfied results were gotten.This theses includes two parts: one is the hardware designe, the other is realization of the algorithms.In the first part the scheme of hardware realization is introduced and the methods to realize it is presented, the difficulties appeared is analyzed and how to resolve them is explained. An embed color video tracking platform that can satisfy the needs for more video application with a little modification was realized.The following are mainly used key techniques of the hardware realization:1) Using a color CCD camera and an enhanced video processor to sent the color information to the video tracking system, the dimension of the character space of a target is increased.2) Using two frame-memorizes to gather the data from the video signal and to get the data from the frame-memorizes for processing in the same time. It not only improved the Real-time of system, but also satisfied the time needs of a complex algorithm.3) Using the HPI port of the DSP as a general I/O port and simulating the data transfer function by software to realize an I2C-BUS without other circuits. It simplified the hardware circuit.4) Using the boot-load programs in the DSP processor to boot load the applied program, the reaction speed of the system was improved. Fixing the applied program in the external data space of the DSP and collocatingthe frame-memorize as the program space of the DSP, therefore theconflicts that there is not enough data space in the DSP for a lot of videodata was resolved. 5) The capability of integrating and renewing the system was improved byusing a CPLD.The algorithms of separating a target from its background and getting the characters of the target and locating the target in the wavelet transform space were presented in the second part. By using the embed color video tracking platform, the real-time tracking was realized. The key techniques used in our software design are as following:1) Using the capability of multi-resolution analysis in wavelet transform and the method of layered searching and matching to recognize and locate a target, the processing capability in real-time of the system was increased.2) Using the multi-character of the wavelet space as the characters of a target to increase the dimension of the character space of the target and to increase the reliability of identifying the target.3) Using the capability of TMS320C5409, all the algorithms of separating a target from its background, extracting the character and identifying the target are operated in "parallel". For example, during the separation of target from its background the characters of the target were picked up. This could quicken the operation obviously.The experiments using the Embed Color Video Tracking System and the methods designed shown that this tracking system is satisfied to the demands of real-time tracking .