Data compression for video utilizing non-translational motion information

摘要

A method of compressing a video signal. The method comprises: for each of a plurality of image portions, transforming from a spatial domain representation into a transform domain representation; and for each of a plurality of target image portions to be encoded, determining a respective reference portion, determining a prediction of the target image portion based on the reference portion, and encoding the target image portion based on the prediction. The prediction comprises predicting a non-lateral motion of the target image portion relative to the reference portion based on processing of their transform domain coefficients. The encoding comprises generating an encoded bitstream comprising information about the non-lateral motion for use by a decoder.

主权项

1. A method comprising:
transforming individual image portions of a plurality of image portions of frames of a video signal from a spatial domain representation into a transform domain representation so as to generate a respective set of transform domain coefficients for each of the individual image portions; determining for at least one of the plurality of image portions a respective reference portion represented by a respective set of the transform domain coefficients, the respective set of the transform domain coefficients being arranged in a first order; determining for the at least one of the plurality of image portions a prediction of a target image portion based on the reference portion by predicting a non-translational motion of the target image portion relative to the respective reference portion, said predicting being based on:
a reordering caused by the non-translation motion of transform domain coefficients from the first order of the respective reference portion to a second order expressed in similar transform domain coefficients for the target image portion; anda similarity between a pattern expressed in the first order of the transform domain coefficients of the respective reference portion, and a pattern expressed in the second order of the similar transform domain coefficients for the target image portion; and encoding the target image portion based on the prediction by generating an encoded bitstream that includes information about the non-translational motion.