CAJ | 학술논문

在视频监控的运动检测中,运动阴影常被误认为视频对象。为了解决这个问题,提出了一种结合RGB颜色特征和纹理特征的运动阴影检测方法。利用背景差和多帧差结合的方法,获得包含运动阴影的前景分割,在此基础上利用阴影区域中RGB特征的相似性(其RGB向量夹角较小)与局部二值纹理模式( Local Binary Pattern,LBP)相结合,进行阴影检测与去除；使用形态学的方法去除像素突变的边缘部分。实验表明,该方法在运动目标检测及跟踪方面取得了较好的效果,与其他消影算法比起来,速度较快。
재시빈감공적운동검측중,운동음영상피오인위시빈대상。위료해결저개문제,제출료일충결합RGB안색특정화문리특정적운동음영검측방법。이용배경차화다정차결합적방법,획득포함운동음영적전경분할,재차기출상이용음영구역중RGB특정적상사성(기RGB향량협각교소)여국부이치문리모식( Local Binary Pattern,LBP)상결합,진행음영검측여거제；사용형태학적방법거제상소돌변적변연부분。실험표명,해방법재운동목표검측급근종방면취득료교호적효과,여기타소영산법비기래,속도교쾌。
In video motion detection,moving shadow is often mistaken as video object. In order to solve this problem,the shadow detec-tion method of combining RGB color feature and texture feature is proposed. Video target segmentation is achieved by background differ-ence and multi-frame difference. The RGB similarity ( RGB vector angle becomes smaller) in shadow area and the Local Binary Pattern ( LBP) is employed to detect and eliminate the shadow. The morphological method is used to eliminate the edges of pixel mutation. The experimental results demonstrate the proposed method has good effect on target detection and tracking,is faster than others.