基于视差和帧差的图割优化运动目标分割算法

提出一种基于视差和帧差的图割优化运动目标分割算法,在运动目标分割过程中,利用能量函数优化方法得到较为准确的区域视差,同时在此基础上将视差和帧差特征采用图割算法的能量函数进行融合,以此提高前景目标分割的准确性。实验结果表明该方法当区域视差的优化受到灰度因素的影响时,利用图割结合视差和帧差特征能够有效地减少视差优化不够准确的区域被检测为前景目标的可能性,同时也能填补大多数帧差分割的空洞,增强了分割结果的稳定性。

Moving Object Segmentation Based on Graph Cut Optimization Integrating Disparity and Frame Difference

In the paper, an optimizing method of calculating the dense disparity based on energy function is used, but the algorithm can not get more accurate segmentation results for the foreground is similar to the background, and On this basis, this paper combine the features of frame difference and disparity by graph cut to improve the accuracy of the image segmentation. The experimental results show that the method reduces the possibility of mistaking some areas for the foreground because of not wholly accurate disparity and fill the cavities of segmentation results of frame difference by combining the features of frame difference and disparity in graph cut optimation algorithm. Finally, the stability of segmentation is improved.