视频监控系统中运动目标检测方法研究-信号前与信息处理专业毕业论文.docxVIP

西北大学硕士学位论文Abstract 西北大学硕士学位论文 Abstract Moving object detection in video surveillance system，which involves the advanced technologies in the fields of image processing，pattern recognition，artificial intelligence and SO on，has developed into the significant field of computer visual research．The thesis focuses on the technologies of moving object detection and extraction in a static situation． This thesis attempts to explore the internal consistency among various calculations through taxonomical research on the calculation of moving object detection in the hope of facilitating our comprehension and improving it eventually． The thesis gives an overall summary of the rationale of technology of video image segmentation and presents a theoretical model of moving object detection and segmentation and proposes a modified calculation based on the rationale，which combines both background subtraction method and frame subtraction with experimental results．By using frame subtraction，this method accumulates images，combines the testing effect of former frame，locates the moving objects in the front and back of the current frame respectively and updates background，and then tests the moving object by using background information． The experimental result proved that this method possesses not only background subtraction’S complete object test，but also frame subtraction’S robustness in the conditions of light varia．tion． Furthermore，after studying statistical characteristics of background image，it brings a self-adaptive method of background modeling estimation，which effectively adopts the fundamental thoughts of self-adaptive mixture gauss background modeling suggested by Stauffer,SO as to construct mixture gauss background modeling for each pixel and make with some improvements．It updates mean∥and variance 6 different updated rates 口aJld侈respectively．The experiment proved that this method improved testing sensitivity and reduced the degree of model’S adaptation． In the end，the th