Patent attributes
Methods and systems for light field image encoding and decoding are disclosed. According to some embodiments, the method receives scene metadata and input light field images associated with a scene. The method further performs a first encoding operation on the scene metadata and the input light field images to generate reference views and reference disparity information. The method further performs a second encoding operation based on the reference views, the reference disparity information, and synthesized residuals to output encoded light field data, where the encoded light field data comprises encoded reference views, encoded reference disparity information, and encoded synthesized residuals. The method further randomly accesses and selects a group of reference views and corresponding disparity information from the encoded light field data based on one or more selected regions of interest. And the method transmits the selected group of reference views, the selected corresponding disparity information, and the encoded synthesized residuals.
