Intra Prediction Encoding Sub-Block Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image compression methods, such as MPEG-1, MPEG-2, and H.264/AVC, repeat information about intra prediction modes in bit streams due to encoding based on standard 8×8 blocks, despite adjacent unit blocks having the same prediction direction, leading to inefficient compression rates.
Innovation Solution
Divide current blocks into sub blocks based on intra prediction directions, allowing only one intra prediction direction to be encoded per sub block, which can be of varying sizes or rectangular shapes, thereby reducing redundant information in the bit stream.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If intra prediction encoding is performed based on standard 8×8 blocks, then the encoding process is simple and standardized, but information about intra prediction modes is repeated in the bit stream even when adjacent unit blocks have the same prediction direction, leading to inefficient compression rates
Solution Approach 1:
The current block is divided into sub blocks in a manner different from the standard 8×8 block segmentation. By segmenting the block into sub blocks that align with regions having the same intra prediction direction, the patent avoids repeating prediction mode information for adjacent unit blocks with identical directions, thereby improving compression efficiency while maintaining encoding simplicity
Solution Approach 2:
Adjacent unit blocks that share the same intra prediction direction are merged into a single sub block for encoding purposes. This merging eliminates the redundancy of encoding the same prediction mode information multiple times across adjacent blocks, directly addressing the compression efficiency problem while preserving the standardized encoding approach
2Measurement precision
If intra prediction direction information is encoded for each unit block, then the encoding is thorough and accurate, but redundant information is inserted into the bit stream when adjacent blocks have equal prediction directions
Solution Approach 1:
The patent extracts and eliminates the redundant intra prediction direction information from the bit stream by identifying adjacent unit blocks with identical prediction directions and encoding them as a single sub block. This extraction approach maintains the accuracy of prediction direction encoding while removing unnecessary duplicate data, directly reducing bit stream volume
Solution Approach 2:
The patent changes the granularity parameter of intra prediction encoding from the standard 8×8 unit block level to a sub block level that adapts to the actual distribution of prediction directions. By dynamically adjusting the encoding unit size based on prediction direction homogeneity, the patent reduces the quantity of encoded information while preserving measurement precision
Data Source
AI summary
Methods and apparatuses for encoding and decoding based on intra prediction. The method of encoding an image based on intra prediction comprises dividing a current block into a plurality of sub blocks so that a plurality of unit blocks adjacent to each other are included in one of the sub blocks based on an intra prediction direction of each of the unit blocks included in the current block, and performing intra prediction encoding on each of the sub blocks so that only information about one intra prediction direction is encoded in each of the sub blocks.