基于帧间差分与码本模型的运动车辆检测算法  

Moving Vehicle Detection Algorithm Based on Frame Difference and Codebook Model

作　　者：杨燕妮[1] 吴向前[2] 刘鹏[3] 

机构地区：[1]新疆大学电气工程学院,新疆乌鲁木齐830049 [2]新疆大学信息科学与工程学院,新疆乌鲁木齐830049 [3]成都信息工程大学计算机学院,四川成都610225

出　　处：《新疆大学学报（自然科学版）》2016年第2期203-208,共6页Journal of Xinjiang University(Natural Science Edition)

基　　金：国家自然科学基金(61463047)

摘　　要：针对复杂背景下的视频车辆检测问题,采用一种在YUV空间下基于改进的帧间差分与简化码本模型相结合的运动车辆检测方法.首先,对亮度信号建立背景的码本模型,在模型训练中通过帧间差分提高训练速度.其次,在检测阶段引入分块思想,根据分块帧差提取粗略的前景像素,再用码本模型得到准确的前景车辆.视频测试实验结果表明,改进后的方法在树枝晃动、光照变化、前景融人背景的情况下能准确检测出运动车辆,具有较好的实时性和鲁俸性.In light of the complicated scenes in the video vehicle detection,the paper put forward a moving vehicle detection method based on the improved frame difference and simplified codebook model under YUV space.First of all,luminance signal is adopted to establish the codebook model of background,improving the training speed by frame difference in the process of model training.Secondly,the method introduces the thought of block when detection,extracts rough foreground pixels by consecutive blocks frame difference,then gets the exact vehicles by codebook model.Video test results show that this method can detect the moving vehicles effectively under the condition of branches shaking,brightness changing and foreground turning into background,and has the characteristics of fast operation and great robustness.

关 键 词：运动车辆检测 背景建模 码本模型 帧间差分 YUV空间 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]