Face stereo matching and disparity calculation in binocular vision system

In this paper, a novel framework for face stereo matching and disparity calculating based on binocular vision was proposed. We focus on generating facial disparity map without expensive devices and generic face model. First, we preprocess facial stereo images, including illumination compensation, face detection and feature point extraction. Then, we introduce camera calibration to obtain camera interior and exterior parameters. Stereo matching is extended and applied to color in addition to grayscale, which improves matching in a low texture image like parts of the skin of a human face. Prior knowledge of the face is used to determine erroneous disparity estimates. Some experimental results are presented to demonstrate the proposed technology. In our experiment results, the method proposed has a satisfied performance regarding calculating time and the matching accuracy for corresponding points.