Hand gesture segmentation based on improved kalman filter and TSL skin color model

Abstract

The first step about the hand gesture recognition is the gesture segmentation. Frequently-used segmentation method is mainly about skin color segmentation based on color space, however skin color segmentation can't get rid of the similar skin color objects effectively. This paper proposes a new hand gesture segmentation method combined improved kalman filter and TSL skin color model, which efficiently solve the problem that the accuracy and efficiency could not be positively related to each other in traditional hand gesture segmentation. Firstly, use improved kalman filter to pre-judge the hand position for the sake of decreasing the interference of the similar skin color. Secondly, use TSL skin color model to realize hand gesture regions. Finally, holes and isolated small targets are removed by morphological approach. This novel approach could obtain a better segmentation.