摘要
为去除阴影对运动目标检测的影响,在传统基于颜色空间特性的阴影检测法基础上,综合边缘特征进行快速双轮廓阴影去除;通过前景物体的轮廓线形状特征及阴影与光源之间投影位置关系快速找到阴影定位点,在定位点附近轮廓线上结合灰度及颜色特性进行有针对性的搜索,找阴影与物体的交界点位置,标记为阴影的外轮廓线;根据外轮廓线长度和位置,求出S空间与之相对应内轮廓线,此时的外轮廓线包围阴影区域;最后,对该封闭的不规则阴影轮廓线进行基于像素点的校正和标记,并对物体反光面的自阴影进行H空间去除,得到更精确的去阴影效果;实验结果表明,该方法快速且有针对性的锁定阴影区域进行阴影去除,避免物体区域误检测,有较好的实时性和准确性。
The double contours shadow elimination method is proposed based on edge and colour features to reduce the effort of shadow in the moving target detection. The method locate the position of shadow point using shape features of the foreground object contours and the position relationship between light source and shadow, scanning for interlock points between shadow and object on the outer contour by col-. our features. Find the corresponding points of inner contour in the S colour space, the shadow region is enclosed within the outer and inner contours. After that, refine and mark the irregular contour through the adjustment of pixels, remove self-shadow in the H colour space for a better shadow elimination effect. The final results show that this method is efficient in locating shadow region, avoid miss operation of fore ground object with similar chromaticity, and have good real-time performance and accuracy.
出处
《计算机测量与控制》
2017年第9期174-177,181,共5页
Computer Measurement &Control
基金
国家自然科学基金项目(U1504617)
关键词
运动目标检测
阴影去除
双轮廓线
边缘特征
moving object detection
shadow elimination
double contours
feature of Edge