Hand detection based on self-organizing map and motion information

In this paper, a method combining SOM-color clustering and motion information is presented. This method can detect and segment dynamic hand gesture in complex background. Through HSV vector and the improved conventional SOM, the number of color clusters can be determined and color can cluster automatically. Experimental results show the proposed dynamic gesture detection and segmentation algorithm is adaptive, fast and robust. And this method can also be adopted in other research region.