Intra-Prediction Weighting for Video Compression
Solution Overview
Problem
Current video compression techniques face challenges in achieving improved compression ratios with minimal sacrifice in picture quality, especially in scenarios with limited network resources and increasing demands for higher video quality.
Innovation Solution
The method involves intra-predicting a block of a picture by determining predicted sample values using DC, planar, or angular intra-prediction modes, calculating weights based on block dimensions, and applying these weights to reference samples to generate a weighted predicted sample value, which is then normalized through arithmetic right shift.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If video compression is applied to reduce data size for transmission and storage, then compression ratio is improved, but picture quality deteriorates
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting prediction weights based on block position, dimensions, and intra-prediction mode. Different weight values are assigned to top and left reference samples depending on the specific block characteristics and prediction mode used, optimizing the balance between compression and quality
Solution Approach 2:
The patent implements local quality by applying position-dependent prediction combination where different regions of the block receive different prediction treatments. The weight values for top and left reference samples are determined based on local block properties such as position, width, height, and prediction mode, allowing optimized prediction for each local region
2Productivity
If compression techniques are enhanced to improve compression ratio, then data size is reduced, but processing complexity increases
Solution Approach 1:
The patent uses parameter changes by deriving weight values from block dimensions and prediction mode through mathematical operations (logarithms, divisions, multiplications). These parameter-based weight calculations enable adaptive prediction that improves compression efficiency while maintaining manageable processing complexity through deterministic formulas
Data Source
AI summary
A method of coding comprises determining a predicted sample value P(x, y) based on one or more reference sample values for a sample from a plurality of samples of a block using a DC intra-prediction mode; determining a vertical weighting factor based on one of a height of the block, a width of the block, or both; determining a horizontal weighting factor based on one of the height, the width, or both; obtaining a modified predicted sample value {circumflex over (P)}(x, y) without a weighting factor of the top-left reference sample R−1,−1 based on: (wL×R-1,y+WT×Rx,−1+(64−wL−wT)×P(x,y)+32))>>6; and reconstructing the block based on the modified predicted sample value {circumflex over (P)}(x, y) and a residual block; wherein wT is the vertical weighting factor, wL is the horizontal weighting factor, Rx,−1 represents a value of nearest reference samples located above the sample, and R−1,y represents a value of nearest reference samples located to the left of the sample.