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.%For moving hand segmentation, in order not to use unreasonable assumptions and to solve the hand-face occlusion, a segmentation method based on skin color, grayscale, depth and motion cues is proposed. Firstly, according to the variance information of grayscale and depth optical flow, Motion Region of Interest (MRoI) is adaptively extracted to locate the moving body part. Then, corners which satisfy skin color and adaptive motion constraints are detected as skin seed points in the MRoI. Next, skin seed points are grown to obtain candidate hand region utilizing skin color, depth and motion criterions. Finally, edge depth gradient, skeleton extraction and optimal path search are employed to segment moving hand region from candidate hand region. Experiment results show that the proposed method can effectively and accurately segment moving hand region under different circumstances, especially when the face is occluded by the hand.
展开▼