In this study, a super-resolution (SR) reconstruction approach for Kinect 3D data is proposed. The proposed approach contains four steps: (1) extract the edge maps from the low-resolution (LR) depth map and the high-resolution (HR) color image using Canny edge detector, subsample the edge map of the HR color image, and segment the HR color image using mean shift segmentation, (2) detect and fill depth holes in the LR depth map, (3) upsample the LR depth map and reduce edge artifacts using local edge enhancement, and (4) perform HR depth map determination by energy cost minimization and refine the final HR depth map by joint bilateral filtering. Based on the experimental results obtained in this study, the performance of the proposed approach is better than those of four comparison approaches.