Image predictive coding device, image predictive coding method, image predictive coding program, image predictive decoding device, image predictive decoding method, and image predictive decoding program

摘要

An image predictive encoding device of an embodiment obtains motion information, for acquiring from a previously-reconstructed reference picture, a signal highly correlated with a pixel signal of a target region serving as an encoding target. The motion information contains prediction modes of candidates for selection, the prediction modes being a uni-prediction using one reference picture and a bi-prediction using two reference pictures. When a prediction mode that is selected from the candidates of the two prediction modes is the uni-prediction, the motion information is obtained from one reference picture set in a single reference picture list used for uni-prediction in which frame numbers of a plurality of reference pictures are registered. At least the prediction mode is encoded as the motion information.

主权项

1. An image predictive encoding device comprising:
a processor; a region division unit executable by the processor to divide an input image into a plurality of regions; a motion information estimation unit executable by the processor to obtain motion information for acquiring, from a previously-reconstructed reference picture, a signal highly correlated with a pixel signal of a target region, the target region serving as an encoding target resulting from division of the input image; a predicted signal generation unit executable by the processor to generate a predicted signal of the target region based on the motion information; a residual signal generation unit executable by the processor to generate a residual signal based on the predicted signal of the target region and the pixel signal of the target region; a quantization unit executable by the processor to perform quantization of the residual signal to generate quantized coefficients; an encoding unit executable by the processor to encode the motion information and the quantized coefficients of the residual signal; an inverse quantization unit executable by the processor to perform inverse quantization of the quantized coefficients generated by the quantization means, to reconstruct a residual signal; and a recording unit executable by the processor to store, as a reference picture, a picture which includes a reconstructed pixel signal of the target region, the picture generated by addition of the predicted signal and the reconstructed residual signal, wherein the motion information comprises prediction modes of candidates for selection, the prediction modes being a uni-prediction using one reference picture from a first reference picture list, which is a single list used for uni-prediction, and a bi-prediction using two reference pictures from a second reference picture list and a third reference picture list, which are used for bi-prediction, the first, second, and third reference picture lists being different, a reference picture list management unit executable by the processor to determine the reference pictures to include in the first reference picture list based on the reference pictures included in the second and third reference picture lists, wherein the motion information estimation unit is executable by the processor to select one of the candidates from one of the two prediction modes and, in response to selection of the uni-prediction, the motion information estimation unit is further executable by the processor to obtain the motion information from one reference picture included in the first reference frame list in which frame numbers of a plurality of reference pictures are registered, and wherein the encoding unit executable by the processor to encode at least the prediction mode as the motion information.