Gaze-corrected View Generation Using Stereo Camera System for Immersive Videoconferencing

In this paper, we present an immersive videoconferencing system that enables gaze correction between users in the internet protocol TV (IPTV) environment. After we capture the object using stereo cameras, we perform preprocessing techniques, such as camera calibration, color correction, and image rectification. The preprocessed images are down-sampled and disparities are computed by using the down-sampled images. The disparity sequence is then filtered to improve temporal consistency. After central view synthesis, occlusion areas are decided and holes are filled. The entire system is implemented with parallel programming that is executed on the GPU for real-time processing. Finally, the user can observe the gaze-corrected image through display. From experimental results, we have verified that the proposed stereo camera system is sufficient to generate the natural gaze-corrected virtual image and realize immersive videoconferencing(1).