US 2014/0369616 A1 · IMAGE PREDICTIVE ENCODING AND DECODING SYSTEM

Abstract

In image predictive encoding according to an embodiment, one or more motion information sets are derived from plural pieces of motion information stored in a computer readable storage medium. Each of the one or more information sets includes two pieces of motion information between which any one of constituent elements differ in value. A predicted signal of a target region in a picture is generated by motion compensation, using two pieces of motion information in a motion information set selected from the one or more motion information sets. The two pieces of motion information in the selected motion information set are stored in the computer readable storage medium, and are used for generation of a predicted signal of another region.

Claims6

Dependent claims

An image predictive decoding method in which a plurality of blocks are merged into a single region that is assigned common motion prediction parameters, the method comprising:

decoding a bitsream from an encoder to extract a list index;
filling a pair of candidate lists with entries of predictive candidates collected for prediction of a target block, wherein at least some of the predictive candidates belong to the single region, and at least some of the entries comprising pairs of bi-predictive candidates, each pair being stored, respectively, in the pair of lists and commonly indexed in the lists;
identifying, in the pair of candidate lists, a pair of bi-predictive candidates indexed in agreement with the extracted list index and retrieving their associated motion prediction parameters in pair;
and
averaging the pair of retrieved motion prediction parameters and generating a predictive block of the target.

2 more elements

Dependencies are inferred from the claim text.

Image predictive encoding and decoding system

Life of this patent

Image predictive encoding and decoding system

Life of this patent

Abstract

Claims6