Improved 3D Depth Image Estimation Algorithm for Visual Camera

A new 3D depth image estimation algorithm for visual camera is proposed in this paper. One visual camera and one Time-of-flight camera are used in our method. First, we propose a novel calibration method for visual and Time-of-Flight camera, in which both the intensity image and depth image of Time-of Flight camera are taken into consideration. Then the initial depth map is generated by project depth map of Time-of Flight camera to the view of visual camera. Using the color image of visual camera as reference, the initial depth map is iteratively refined according to a novel energy function which fuses two different kinds of data: the depth data and the color data. Experiment results show that the high resolution depth map of visual camera is successfully generated and of high quality.