Segmentation of Hand Posture against Complex Backgrounds Based on Saliency and Skin Colour Detection

摘要

Hand posture segmentation is an important step in hand posture recognition process, which aims to eliminate the interference from other elements in an image, and get the hand posture region precisely. In this paper, we propose a hand segmentation method based on saliency and skin colour detection. This method mainly consists of a pixel-level hand detection method, a region-level hand detection method, and a multiple saliency maps fusion framework, thereby achieving the deep integration of the bottom-up saliency information and top-down skin colour information. Experimental results show that the proposed method has excellent performance, and is reliable against cluttered backgrounds. Moreover, evaluations based on comparisons with other approaches of the literature have demonstrated its effectiveness.