I. Introduction
Machine vision is a system that realizes observation and judgment by replacing human eyes with machines, and it has been applied in various fields of modern society. It is mainly manifested in: multi-sensor image fusion, vehicle pedestrian tracking, character and bar code positioning and recognition, and so on. At the same time, in special cases, only relying on human vision cannot meet the accuracy requirements, which can also be completed with the help of machine vision technology. At present, there are many researches on target location methods, which usually realize target location according to target prior knowledge and template image matching. Among them, target location based on image matching has a wider range of application and can achieve good application results [1]. In the case of no human intervention, the realization of video image intelligent processing with automatic extraction and positioning of moving targets in video image sequence as the core has important practical application value. Its application will change the traditional video monitoring, reduce the labor intensity of personnel on duty, and truly realize intelligent and automatic video monitoring [2].