The Embedded Implementation Of Moving Object Detecting And Tracking System

Moving object detecting and tracking technology is relative to the field of signal processing,video processing,and computer vision.The current video detecting system only can simply track and record.The detection and analyse of video content is still mainly based on manual method.The development of video analyse technology is left behind,and is blocking the development and implemention of video monitor system.This thesis implements various kind of moving object detecting algorithm,binary image connected regions marking algorithm and moving object tracking algorithm.Via using C language,this thesis has implemented an embedded moving object detecting and tracking system on the ARM9 platform.The works that have been done in the thesis include:(1) The application background,research history of moving object detecting and tracking technology and embedded system are introduced.(2) The common pixel domain moving object detecting algorithms,binary image connected region marking algorithms,moving object tracking algorithms are described in detail.(3) Embedded system related knowledge is introduced,such as embedded system composing,embedded Linux operating system,ARM development tool and enviorment,etc.(4) A design scheme of object detecting and tracking system based on embedded system is proposed.The design of each module is described in detail,such as image fetching module, image decoding module,moving object detecting module,binary image connected region marking module,moving object tracking module,serial port transmit module and image display module for debugging.A determiniting criterion for YCrCb domain background subtraction is proposed.(5) The three main algorithm modules are implemented and the performance of the entire system is evaluated.The embedded moving object detecting and tracking system implemented in this thesis has quasi-realtime processing ability,and can be used in low-speed object detecting in simple scene.