Abstract: The process of compression of a block of a size L.times.H of a sequence of images comprises, in a repetitive manner, for said block: a search step, in one of the images of the sequence of images, for the zone L.times.H pixels the most similar to said block; a step of determining whether the resemblance between said zone and said block responds to predetermined criteria; if said resemblance responds to said criteria, a step of storing the motion vector which indicates the distance between the block and the most similar zone found; if said resemblance does not respond to said criteria, and if the block does not have a predetermined minimum size, a step of cutting out said block into sub-blocks and a supplemental repetition, for each of said sub-blocks, and if said resemblance does not respond to said criteria, and if the block has a predetermined minimum size, a step of storing the block in the course of which the block is compressed without reference to a reference image.
Abstract: The purpose of the invention is to enable, or not enable, motion compensation depending on whether it is profitable to do so, with profitability being expressed in terms of the costs in rate and in distortion. The method for decompressing an image portion comprises: a step (302) of determining at least one prediction of a said image portion, a step (307) of determining a distortion associated with each prediction, and a step (314) of deciding whether or not to encode the residual, based at least on the distortion value associated with each prediction.
Abstract: The process of compression of a block of a size L.times.H of a sequence of images comprises, in a repetitive manner, for said block: a search step, in one of the images of the sequence of images, for the zone L.times.H pixels the most similar to said block; a step of determining whether the resemblance between said zone and said block responds to predetermined criteria; if said resemblance responds to said criteria, a step of storing the motion vector which indicates the distance between the block and the most similar zone found; if said resemblance does not respond to said criteria, and if the block does not have a predetermined minimum size, a step of cutting out said block into sub-blocks and a supplemental repetition, for each of said sub-blocks, and if said resemblance does not respond to said criteria, and if the block has a predetermined minimum size, a step of storing the block in the course of which the block is compressed without reference to a reference image.
© 2026 Justia
