Adaptive Motion Gesture Segmentation
Wen-kai Tsai, Chung-Chi Lin, Shyue-Wen Yang, Ming‐Hwa Sheu, Ching-Lung Su · 2008
In this paper, we propose a method to automatically obtain a user's gesture skin color space which is used to perform gesture segmentation. The proposed approach is based on gesture skin color space that is obtained by the least squares approximations solution and Gaussian distributed model. In addition, we can also improve gesture segmentation defacement, which is caused by a complex background, shadow effect and light-reflection effect. Compared with NCC and YCbCr, our proposed method can achieve more accurate motion gesture segmentation.