| 内容提要: |
Super-resolution reconstruction is to improve the resolution of a known low-resolution (LR) image and reconstruct its corresponding high-frequency information by some prediction methods. This process is called Super Resolution (SR).In this paper, we use color-guided multilevel fusion supervisory network to achieve SR of depth map, aiming to use the consistent object boundary between color map and depth map to assist the reconstruction of depth map, especially the edge part: |