Depth estimation based on adaptive support weight and SIFT for multi-lenslet cameras
Yuan Gao, Wenjin Liu, Ping Yang, Bing Xu · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2012
With a multi-lenslet camera, we can capture multiple low resolution subimages of the same scene and use them to reconstruct a high resolution image. The spatially variant shifts estimation between subimages is one of major problems. In this paper, a depth estimation algorithm has been proposed for multi-lenslet cameras. The stereo matching between the reference subimage and other subimages using segmentation-based Adaptive Support-Weight approach combined with Scale Invariant Feature Transform (SIFT) is introduced, which has an influence on the result of stereo matching. Then, disparity maps are converted to depth maps and these depth maps are merged into one map for quality improvement. At last, the average blending images at difference depth are calculated according to the depth map. The experimental results show that the proposed algorithm can extract accurate depth more concisely and efficiently.