Disparity Reconstruction Algorithm Based on YCbCr Light Field Data

In the light field computational imaging, scene depth reconstruction is transformed into a problem of disparity reconstruction. Efficient disparity reconstruction based on monomer light field data is realized by introducing YCbCr color space light field data. Region matching in Y channel can avoid the redundant calculation of RGB three-channel matching, and improve the matching efficiency. Monomer edge occlusion and internal disparity consistency constraint can be realized by Cb, Cr channel monomer to solve the problem of mismatching between occlusion region and smooth region. Cb, Cr chromaticity information provides effective clustering information for monomer. The accurate segmentation of monomer is realized by combining the region growth and dichotomy. At the edge of the monomer, the matching window shape and the visible viewpoint are selected according to the edge of the monomer to avoid the appearance of the edge occlusion mismatching. In the monomer, the disparity map is optimized based on disparity consistency prior. Experimental results of simulated data and real data show that the proposed method can achieve accurate reconstruction in texture, occlusion, and smooth regions, and has strong robustness for disparity reconstruction in occlusion regions.