A new high resolution depth map estimation system using stereo vision and depth sensing device
Shuai Zhang, Chong Wang, S. C. Chan · 2013
Depth map estimation is a classical problem in computer vision. Conventional depth estimation relies on stereo/multi-view matching or depth sensing devices alone. In this paper, we propose a system which addresses high resolution and high quality depth estimation based on joint fusion of stereo and Kinect data. The problem is formulated as a maximum a posteriori probability (MAP) estimation problem and reliability of two devices are derived. The depth map estimated is further refined by color image guided depth matting and a 2D polynomial regression (LPR)-based filtering. Experimental results show that our system can provide high quality and resolution depth map, which complements the strengths of stereo vision and Kinect depth sensor.