一种基于区域的车辆阴影消除方法

Region-based Vehicle Cast Shadow Elimination Approach

该文给出了一种基于区域的车辆阴影检测方法。首先将运动前景像素检测出来,利用梯度特征分割运动前景的连通域获取投射阴影的方向,然后再结合区域内的角点及灰度相似性特征分割车辆和其阴影像素。实验结果表明该方法可以有效的检测阴影,并且能够满足实时性的要求。

A region-based vehicle cast shadow elimination approach was developed.It first identifies each moving object as a blob,then calculates gradient feature and therefore achieves the direction of the cast shadow,and finally uses corner detection and gray similarity to di-vide vehicle and its shadow.The experimental result indicates that the approach is effective and can process video sequence in real-time.