智能视觉监控技术研究进展  

Intelligent Visual Surveillance Technology:A Survey

作　　者：王素玉[1] 沈兰荪[1] 

机构地区：[1]北京工业大学信号与信息处理研究室,北京100022

出　　处：《中国图象图形学报》2007年第9期1505-1514,共10页Journal of Image and Graphics

基　　金：国家自然科学基金项目(90304001;60472036);北京市自然科学基金项目(4052007);武器装备预研基金项目(51434050105QT0101);北京市教委基金项目(KM200410005022);北京市科技新星计划基金项目(2005B08)

摘　　要：新一代智能视觉监控技术的研究是一个极具挑战性的前沿课题,它旨在赋予监控系统观察分析场景内容的能力,实现监控的自动化和智能化,因而具有巨大的应用潜力。视觉监控系统的智能化分析过程由运动目标检测、分类、跟踪和视频内容分析等几个基本环节组成,其中视频内容分析又包括异常检测、人的身份识别以及视频内容理解描述等。本文在总结以上有关关键技术研究进展的基础上,进一步提出将超分辨率复原技术引入视觉监控领域,介绍了超分辨率复原的主要算法及其在智能视觉监控中的应用。The research on new generation of Intelligent Visual Surveillance is a new arising font field with many challenges. It aims to endow surveillance systems with the ability of analyzing scene contents, make surveillance tasks fulfilled automatically and intelligently, which has a large potential of application. There are four main issues in the process of intelligent analysis of typical visual surveillance system, detection, classification, tracking of objects and analysis of video contents. According to the real applications, video contents analysis can he abnormal detection, person recognition, understanding and description of video contents etc. In this paper, in addition to make a survey on the research progresses of the key technologies of visual surveillance, the concept of super resolution restoration is further introduced to this field to enhance the quality of the surveillance sequence. The main algorithms of super resolution restoration and its applications in visual surveillance are discussed in detail.

关 键 词：智能视觉监控 目标检测 目标跟踪 步态识别 行为理解和描述 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置] TP391[自动化与计算机技术—控制科学与工程]