摘要： 提出了一种在图像序列中分割运动前景的算法,该算法首先对前后连续的两帧图像作差值,然后再对差值图像的直方图进行处理,并通过曲线拟合获取最佳分割门限,最后对分割结果进行形态滤波恢复出图像序列中的运动前景。实验结果表明:该算法简单易行,能够成功分割出图像序列中的运动物体,是一种较好的运动前景分割算法。

关键词: 运动物体, 位移帧差, 最佳门限, 图像分割

Abstract: This paper represents a moving foreground segmentation approach in image sequence. Firstly, it makes a difference between two consecutive frames. Secondly, histogram processing is made on the displaced frame. The optimum threshold is attained for segmentation by imitating the curve. At last, a morphological filter is used on the segmentation result to recover the moving foreground in image sequence. Experiment results show that the approach can successfully segment the moving objects from the image sequence, which is a better moving foreground segmentation approach.

中图分类号: 

• TN911.73