YUV空间下基于码本模型的视频运动目标检测方法  

Moving target detection method based on codebook model under YUV space

作　　者：王雯[1] 陈丽[2] 李晨[1] 赵宇[1] 代荡荡 王先培[1] 

机构地区：[1]武汉大学系统集成与故障诊断实验室,湖北武汉430072 [2]华为技术有限公司,广东深圳518129

出　　处：《武汉大学学报（工学版）》2015年第3期412-416,423,共6页Engineering Journal of Wuhan University

基　　金：国家自然科学基金资助项目(编号:50677047);中国南方电网科技项目(编号:K-GX2011-019);湖北省科学条件专项基金(编号:2013BEC010)

摘　　要：针对传统运动目标视频检测算法中RGB空间背景建模计算、存储成本高的问题,提出采用YUV空间下基于码本模型的背景建模方法.码本模型对现场光线变化、随机噪声等具有较好的自适应性,采用对YUV模型中的亮度信号(Y分量)建立背景的码本模型,其优点在于亮度信号(Y)和色度信号(U,V)相互独立,避免了复杂的乘法与开平方运算,同时算法可单独编码,与MPEG标准兼容性更好,易于进行目标数据的压缩、处理和传输.视频仿真实验表明,该方法在视频监控系统中的使用效果较好,能获得比RGB空间码本模型更高的实时性,也更节约系统的计算内存.In light of the question that background modeling under RGB space in traditional video detection algorithms for moving targets costs a lot of calculation and storage, a background modeling method based on codebook model under YUV space is proposed. Codebook model has better self-adaptability to spot light changes and random noise, etc. In this detection method,luminance signal （Y component） in YUV model is adopted to establish the codebook model of background. The advantage is that luminance signal （Y） and chrominance signals （U, V） are mutual independent, which can be separately encoded. For that, complex multiplication and square root operations are avoided. This method has better compatibility with the MPEG standard, which can easily perform compression, processing and transmission of target data. Video simulation results show that this method behaves well in video surveillance system, which has better real-time performance and needs less internal memory compared with that using codebook model under RGB space.

关 键 词：运动目标检测 背景建模 码本模型 YUV空间 RGB空间 

分 类 号：TP29[自动化与计算机技术—检测技术与自动化装置]