Video Coding Techniques for Multiview Video

摘要

Techniques for coding and decoding video captured as cube map images are disclosed. According to these techniques, padded reference images are generated for use while predicting input data. The reference image is stored in cube map format. From the reference image, a padded reference image is created in which image data of the first view included in the reference image is duplicated and disposed adjacent to the second view included in the cube map image. When coding a pixel block of the input image, a predictive search may be performed between the input pixel block and the content of the padded reference image. When the predictive search identifies a match, the pixel block may be coded with respect to the matching data from the padded reference image. The presence of replicated data in the padded reference image is expected to increase the likelihood that appropriate predictive matches are identified for the input pixel block data, which will increase the overall efficiency of video coding.