The Research Of Moving Object Tracking System Based On Improved Hausdorff Distance Matching Algorithm

In recent years, with the requirement for military and civilian equipment in modern societygrowing and the demand gradually improving, the technology of moving object detection andtracking is also constantly developing rapidly, having important application and broad prospectsboth in military and civilian. How to establish a highly efficient and accurate real-time movingobject tracking system is a hot topic of research in the field of modern information processing.This paper aimed at some problems such as background disturbance, illumination changesand occlusion situation that may exists in the process of object detection and tracking，composited based on feature tracking methods and based on relevance tracking idea, explored aobject tracking method based on improved Hausdorff distance matching algorithm. The detailsteps is first using multi-fractal image edge extraction algorithm and based on edge backgroundsubtraction method to detect moving object and extract the object template, then using improvedHausdorff distance matching algorithm and multi-resolution search strategy which is improvedbased on hierarchical search algorithm to match images with template, updating adaptivelytemplate with the threshold selection compositing a weighted way to avoid mismatching orlosing tracking object, finally using comprehensive predictor to calculate the object trajectory. Itcan not only solve the problems of object blocking, but also can improve the real-timeperformance of the system.This paper implemented and optimized the matching algorithm based on improvedHausdorff distance and a variety of algorithms in the object detection and tracking system onARM embedded platform. After many tests, the performance is stable and the operation is good,but there are still some problems, remaining to be further research and exploration.