基于运动历史图像和椭圆拟合的手势分割

针对当前常用的手势分割方法较难适应复杂的光照环境，提出了一种新的方法。该方法先从视频序列获取运动历史图像（Motion History Image，MHI），对MHI进行运动区域分割，然后在这些运动区域筛选出手势区域。为了克服手势区域分割偏大的问题，提出了利用该区域内的当前运动轮廓做椭圆拟合，进而得到精度更高的手势分割结果。实验结果表明，提出的方法能够有效地分割出手势，并且和传统方法相比较更能适应不同的光照环境。

Hand gesture segmentation based on MHI and ellipse fitting

For general gesture segmentation method is more difficult to adapt to the complex light environment, this paper proposes a new method. Firstly, the method obtains Motion History Images（MHI）from a video sequence and segment motion regions from MHIs. Then it chooses the hand gesture region from these regions. In order to overcome the problem of excessive segmentation, it uses the outline of the current movement of the region to do ellipse fitting, so that it can obtain a more accurate hand gesture segmentation result. Experimental results show that the proposed method can segment hand gesture effectively and have a greater ability to adapt to different light environment compared with general method.