Light field imaging coding: Performance assessment methodology and standards benchmarking

It is well known that the conventional ways to capture the light around us are limited and thus provide a limited user experience, notably in terms of parallax capabilities. As this has been preventing 3D systems to explode in the market, significant advances are emerging in terms of light capturing technologies among which is relevant to highlight the socalled light field cameras which capture a richer representation of the visual scene by measuring the light intensity for each direction and for each pixel position. Considering the huge amount of light intensity data involved, efficient compression becomes a must. In this context, this paper addresses the light field coding challenge by using available image coding standard solutions. With this target in mind, this paper proposes first a compression performance assessment methodology and presents after the compression performance of the direct, fully compatible usage of the main image coding standards available, notably JPEG, JPEG 2000, H.264/AVC Intra and HEVC Intra for a representative set of light field images. While the expected conclusion is that HEVC Intra is the most efficient codec, other less expected conclusions emerge.