Segmentation of hand gesture based on dark channel prior in projector-camera system

In order to achieve human-computer interaction based on hand gesture in a projector-camera system, an accurate and fast gesture segmentation is necessary. However, it is difficult for traditional method to obtain precise segmentation results because of the complex backgrounds, variable external illumination, and shadows of hand gesture. In this paper, based on the observation of the captured images' dark channel maps, we propose a dark channel prior in the projector-camera system, and thereby putting forward a saliency detection method in this scene. Then a graph-based segmentation method is utilized to refine the saliency map. After integrating the pixel saliency value with regional saliency value and making a segmentation, the hand gesture regions are finally obtained. Experimental results show that the proposed method has an excellent performance, and is reliable against complex illumination and backgrounds in the projector-camera system.