MULTIVIEW VIDEO ENCODING METHOD, MULTIVIEW VIDEO DECODING METHOD, MULTIVIEW VIDEO ENCODING APPARATUS, MULTIVIEW VIDEO DECODING APPARATUS, AND PROGRAM

摘要

Provided is a highly effective coding method for a multiview video having mismatches between cameras in terms of local brilliance and color. A viewpoint synthesized image corresponding to a coding target frame in a multiview video is synthesized from a previously coded reference viewpoint frame which has been photographed from a reference viewpoint which is different from a coding target viewpoint, and at the same time that the coding target frame from the coding target viewpoint was photographed. With respect to each processing unit area having a predetermined size, a reference area on the previously coded reference frame from the coding target viewpoint and corresponding to the viewpoint synthesized image is searched. A correction parameter for correcting the mismatches between cameras is estimated from the viewpoint synthesized image corresponding to the processing unit area, and the reference frame corresponding to the reference area. The estimated correction parameter is used to correct the viewpoint synthesized image corresponding to the processing unit area. The video from the coding target viewpoint is subjected to prediction coding using the corrected viewpoint synthesized image.