Real-time hand detection and tracking with depth values

Abstract

Aim of this paper is to propose a methodology of real-time hand detection based on skin color model and background subtraction under any complex background with extracting the depth information. With the use of stereo camera calibration and disparity mapping, the depth information of the hand is extracted. Reasonable selection of threshold of skin color model and combining with background difference segmentation, then a bitwise-and operation results hand's binary outline feature clearly. Combining this result with the distance information experiments show accurate detection and tracking of hand with depth measuring from the camera.