The Depth Image Restoring Based on Harris Corner Detection

Obtaining the depth image is the foundation of the research on other direction of Kinect. When to obtain the depth image information in a Kinect, it can t obtain the depth image accurately if there is a shelter in front of the body, and then the recognition and the restoration of the skeleton will become very difficult. This paper uses the method of Harris corner detection to detect shelter corner which is used to detect the range of shelter and restore the depth image information. But different from the traditional way, this paper uses the depth data values in the corner detection instead of the gray date values. After detecting corner, a series of actions of depth value substitution and smoothing will be done to repair occluded depth image.