Dynamic Target Detection And Tracking Algorithm Based On Video Image Of Hardware Implementation

The dynamic target detection and tracking technology based on video images is thebasis of research in the field of computer vision and intelligent control, which involvesimage processing, pattern recognition and artificial intelligence, and many other areas.The implementation process of moving target detection and tracking functions is asfollowed: Sending camera image sequences to PC firstly, and then realize it by thecorresponding algorithm. The amount of data processed through the entire process isvery large, there is a relatively large delay and poor real-time. FPGA hardwareprocessor has a fast processing speed, parallel processing characteristics, so it is suitableto video image processing area, it can be used for the video front-end intelligentanalytical processing. Therefore, the research based on the front-end hardwareprocessing of moving target detection and tracking system, improving the veracity andreal-time of intelligent monitoring is necessary. Therefore, a moving object detectionand tracking scheme with real-time has been introduceed in this thesis, which choseFPGA as a hardware development platform. The main work of this thesis is summarizedas follow:(1) Researching basic knowledge of the video signal. First the types andcharacteristics of the video input devices are introduced. Second several common videosignal formats and standards are analyzed. Then the scan mode of display is studied.Finally, the video signal processing color space theory is introduced.(2) Studying commonly used moving target detection and tracking algorithm. Firstthe basic principle and characteristics of three motion target detection algorithm areintroduced, that are the frame difference method, the background minus division andoptical flow method. Secondly the tracking principle and features of four trackingalgorithms are studied, that are based on target area, based on the characteristics, basedon the deformation of the templates, based on the model. Finally, according the analysed,the frame difference method and the background minus division are suitable to realizedon the FPGA device, and the tracking algorithm which based on the characteristics ischosed in this thesis. (3) Researching the architecture based on FPGA hardware development platformfor moving target detection and tracking algorithm. The background minus division andthe tracking algorithm which based on the characteristics are chosed in this thesis, andthe centroid of the moving target is chosed as the tracking characteristic. the principlesand the design realizations based on FPGA hardware have been described in this thesis.Then according to the system function module, using top-down design method, usingVerilog hardware description language for each module design and simulation. Finally,with comprehensive experiment method, the implementation achieved very good results,it can real-time monitoring and tracking a single objective in the monitor scene. It has agood guide for the implementation of front intelligent video analysis.