基于CAMSHIFT算法和形态学的人手动态分割

动态图像分割是计算机视觉分析中一个重要和困难的任务。现讨论了一种基于CAM-SHIFT算法和形态学的动态跟踪分割。首先把采集到的图像由RGB色彩空间转换到HSV色彩空间,因为肤色对HSV色彩空间的H色调具有恒常性。其后在HSV颜色空间建立肤色直方图模型,确定肤色概率度,运用CAMSHIFT算法对人手动态跟踪。然后运用形态学重构闭运算去除空洞和噪声。实验证明该方法能够有效的实现人手的动态分割。

Dynamic segmentation of hand based on the CAMSHIFT algorithm and morphological operators

Image segmentation is important step of low level vision analyzing. This paper discusses on a method on dynamic segmentation of hand using the CAMSHIFT algorithm and morphological operators. As a first step, image is converted into HSV space from RGB space, the reason of doing that is the hue of skin- color in HSV space having invariability. And then it sets up histogram model of skin - color and ensures the value of pmbilities of skin- color for carrying out hand dynamic tracing. Moreover the holes and noise of the images are removed by using morphological reconstruction and close operation, this make dynamic segmentation of hand do effectively.