ABSTRACT
Light field imaging has developed as an emerging technology in the field of digital imaging with the appealing demand of computational photography. Light field cameras like Lytro and Raytrix, is capable of capturing a three-dimensional scene acquiring spatial intensities and angular information which enables functionalities like refocusing, retrieving depth information and changing view perspectives. Advanced coding solutions are essential to remove redundancies making it suitable for effective storage and transmission, while keeping the number of bits low and perceived quality high over constrained bandwidth networks. In the proposed work, image reshaping regularizes the macropixels of a lenslet image to the coding units for a block-based coding architecture. The pre-processed light field image is further compressed with intra prediction process in High Efficiency Video Coding (HEVC)/H.265 based on inherent correlation among the macropixel structures. The least sum of absolute errors (SAE) is employed to predict the current block in HEVC.
