Hand segmentation with skin color fine-tuning using kinect depth sensor

Abstract

The hand segmentation is the critical pre-processing of the gesture recognition application. Nowadays, to achieve a robust hand segmentation under cluttered background is still challenging. Advanced research in model-driven approach based on the depth information has obtained impressive performance. However, it is unable to deal with the hand very close to the body part. Also, a large number of marked samples are needed in training process in order to obtain the compromise results. In this paper, we proposed a hand segmentation method which combining the depth and color information without off-line pre-training. It enables automatic adaptation using depth information and skin color model. An efficient active color zone selection with an elliptical model is proposed. The experiments show that our algorithm can handle the real-time hand segmentation with cluttered and similar depth background.