MOTION VECTOR ESTIMATION METHOD, MULTIVIEW IMAGE ENCODING METHOD, MULTIVIEW IMAGE DECODING METHOD, MOTION VECTOR ESTIMATION DEVICE, MULTIVIEW IMAGE ENCODING DEVICE, MULTIVIEW IMAGE DECODING DEVICE, MOTION VECTOR ESTIMATION PROGRAM, MULTIVIEW IMAGE EN

摘要

Proposed is a motion vector estimation method whereby motion vectors can be accurately estimated even without access to processed images, and multiview images can be efficiently encoded by simultaneous use of camera correlation and time correlation for image signal prediction. The composite-image view from the time the reference images were recorded are generated from reference camera footage taken on a camera other than the processing camera which recorded the processed images including the multiview images, by using the same settings as the reference camera. Motion vectors can be estimated without using processed images by using image signals from the composite-image view which correspond to the processing region of the processed images, and searching for the corresponding region of the reference images recorded by the processing camera.