不同应用场景下的监控视频前景目标提取方法研究  

Research on Foreground Target Extraction Method of Surveillance Video under Different Application Scenarios

作　　者：徐瑶瑶 冯笑雨 刘袁 XU Yaoyao;Feng Xiaoyu;LIU Yuan(Nanjing Center,China Geological Survey,Nanjing Jiangsu 210016;China United Network Communication Co.,Ltd.Beijing Branch,Beijing 100020)

机构地区：[1]中国地质调查局南京地质调查中心,江苏南京210016 [2]中国联合网络通信有限公司北京市分公司,北京100020

出　　处：《软件》2022年第3期165-169,共5页Software

摘　　要：目前,各类传感器广泛覆盖在人类生存的各个角落里,持续捕捉着地球表层的空间场景结构与变化过程,真实地记录着精细的场景信息和运动目标信息,为人们提供了丰富的视频信息。人们通常对视频监控中的前景目标(运动目标)较为感兴趣,而前景目标提取是运动目标跟踪和事件监测的前提,直接影响监测的精确性。本文针对监控视频的前景目标提取问题,提出了不同应用场景下的监控视频前景目标提取方法,并通过实验验证了方法的可行性。At present, various types of sensors are widely covered in all corners of human existence, continuously capturing the structure and change process of the spatial scene of the earth’s surface, truly recording the fine scene information and motion target information, providing people with rich video information. People are usually interested in foreground targets(motion targets) in video surveillance, and foreground target extraction is a prerequisite for motion target tracking and event monitoring, which directly affects the accuracy of monitoring. This paper addresses the foreground target extraction problem of surveillance video, proposes a foreground target extraction method for surveillance video under different application scenarios, and verifies the feasibility of the method through experiments.

关 键 词：前景目标提取 背景减除法 Vibe算法 光流法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]