Self-taught Recovery Of Depth Data

Microsoft’s Kinect as a consumer-grade depth sensor, can be measured in real time to obtain the depth image of the scene as well as color images with synchronized. Kinect depth data obtained for the high-level computer vision problem of some areas, such as research object detection and recognition,3D reconstruction of providing cheap geometric information. In recent years, by virtue of its low price and real-time and other advantages, to be applied to more and more areas. However, because of its imaging works, which emits infrared light at the edge of the object and the material easy to absorb infrared light will change drastically, causing the reflected light can not get to this part of the object, and the limited scope of its work, leading to missing data affects subsequent depth research and applications.Depth data captured by Kinect provides inexpensive geometric information to higher level computer vision tasks such as object detection and recognition. However, there are missing values in the depth map at object boundaries and those beyond the working distance of Kinect due to the limitations of the hardware employed.In this paper, we proposed a self-taught regression method to recover the missing depth data. First a rough estimation of the scene depth was made based on the color image from Kinect. We then trained a random forest using the estimated depth and the intensity from the neighborhood of each pixel that the depth can be captured by Kinect. The random forest was used to predict missing depth data in a self-taught manner that the pixels with largest number of valid neighborhood were predicted first and then added to the training set for the next round prediction. This repeats until all missing data was recovered. The experiment results show that our method outperforms existing approaches to depth recovery.