基于多线索的运动手部分割方法

发布时间：2018-09-07 14:46

【摘要】：分割运动手部时,为了不依赖不合理的假设和解决手脸遮挡问题,该文提出一种基于肤色、灰度、深度和运动线索的分割方法。首先,利用灰度与深度光流的方差信息来自适应提取运动感兴趣区域(Motion Region of Interest,MRoI),以定位人体运动部位。然后,在MRoI中检测满足肤色与自适应运动约束的角点作为皮肤种子点。接着,根据肤色、深度与运动准则将皮肤种子点生长为候选手部区域。最后,通过边缘深度梯度、骨架提取和最优路径搜索从候选手部区域中分割出运动手部区域。实验结果表明,在不同情形下,特别是手脸遮挡时,该方法可以有效和准确地分割出运动手部区域。
[Abstract]:In order not to rely on unreasonable assumptions and solve the hand-face occlusion problem, this paper proposes a segmentation method based on skin color, gray level, depth and motion clues. Firstly, the variance information of gray level and depth optical flow is used to extract the region of interest (MRoI) adaptively to locate human motion. Then, the corners satisfying skin color and adaptive motion constraints are detected as skin seed points in MRI. Then, the skin seed points are grown into candidate hand regions according to skin color, depth and motion criteria. Finally, the motion hand is segmented from candidate hand regions by edge depth gradient, skeleton extraction and optimal path search. The experimental results show that the proposed method can effectively and accurately segment the moving hand region in different situations, especially in the case of hand-face occlusion.
【作者单位】：北京工业大学电子信息与控制工程学院;计算智能与智能系统北京市重点实验室;
【基金】：国家自然科学基金(61375086) 北京市教育委员会科技计划重点项目(KZ201610005010)~~
【分类号】：TP391.41

【相似文献】