The Research Of Algorithm Of Mining-Task-Machine Real-time Tracking System Base On Image

Moving object detection in complex environment is a basic problem in Computer Vision, Because of its good prospect and high application value. At the present time, the theory of moving object detection has been used widely in many field, such as military spy, television surveillance, traffic detection, medical research, experimentation measure and so on.This thesis is focused on research and development underwater Mining-Task-Machine moving object detection and track algorithms that position system's design, and it mainly includes 3 sections :1. Studied sub pixel location technique from the theory, include Gray weightedcentroid algorithm, Edge location algorithm, fitting method etc. we carried out these algorithms by real-time image processed tracking. Some improved algorithm, which can improve the deficiency of basic algorithms, is also researched. Studied Moving object detection algorithms that use extensively currently, and combine sub pixel location algorithms to put forward a kind of bright-point algorithms;.2. Studied to Mining-Task-Machine real-time tracking system's need analysis, include the hardware and environment request. Combine the experiment environment to put forward Real-time tracking system's design and work principle; We analyzed the factor that influence to measure accuracy;3. Studied camera Calibration of basic theory, in common use method, experiment result contents etc. Point out the existent problem and put forward improvement project. Sub pixel location technique and target tracking algorithms, develop system have good customer's interface, stability with multi-function of Mining-Task-Machine moving object detection and track measure system software.The prospect of moving object detection algorithm is analyzed simply.