Spatial-temporal depth de-noising for Kinect based on texture edge-assisted depth classification

The emergence of Kinect facilitates the real-time and low-cost depth capture. However, the quality of its depth map is still inadequate for further applications due to holes, noises and artifacts existing within its depth information. In this paper, a Kinect depth de-noising algorithm is proposed to enhance the stability and reliability of Kinect depth map by exploiting spatial-temporal depth classification beside edges. Depth edges are realigned by extracted texture edges. Spatial and temporal depth classification is retrieved and exploited adaptively to remove the blurs around the edges. Experimental results demonstrate that the proposed algorithm provides much sharper and clearer edges for the Kinect depth. Compared with the original depth and the depths refined by existing approaches, the spatial-temporal de-noised depth information provided by the proposed approach enhances the quality of some advanced processing e.g. 3D reconstruction prospectively.