Patent №
US 6,567,555
Granted
2003-05-20
Filed 2001
Owner
—
Lab
—
AI components
2
nlp · vision
Assignment
None on record
Dataset
AIPD
2023_r1 edition
Application
10000131
An image decoding method is provided for decoding each block of a digital image divided into a plurality of blocks, from a coded image signal including codewords of the each block and codewords of coding mode information of the each block, the method comprising separating the codewords of a target block and the codeword of coding mode information of the target block from the coded image signal, decoding the coding mode information of a target block from the separated codeword of the coding mode information using a codeword table. The codeword table is selected from a plurality of codeword tables by referring to a combination of temporary mode information of the blocks at the periphery of the target block when the blocks at the periphery are out of an image, and the codeword tables each have codewords, each codeword being determined based on a combination of coding mode information of previously decoded blocks at the periphery of the target block, or on a combination of temporary mode information that replaces the coding mode information of the blocks at the periphery of the target block when the blocks at the periphery are out of an image. The method further comprises decoding pixel data of the target block from the separated codewords of the target block, based on the decoded coding mode information of the target block.