Permuting Residual Data for Transform Skipping in Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Transform skipping in video coding results in statistically different data distributions within blocks, invalidating assumptions underlying conventional entropy coding, which negatively impacts coding efficiency.
Innovation Solution
Permuting residual data before quantization and entropy coding to align data distribution with the assumptions of the entropy coding process, using techniques like 180-degree rotation to rearrange data, and applying inverse permutation during decoding to reconstruct the original block of residuals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If transform skipping is applied to intra-coded blocks, then rate-distortion performance is improved for mixed content video, but the data distribution becomes statistically different from transform coefficients, invalidating entropy coding assumptions and reducing coding efficiency
Solution Approach 1:
The patent applies a permutation transformation to the residual data before quantization and entropy coding. This changes the statistical parameters of the data distribution to match the assumptions of the entropy coding process, specifically arranging data so that significant values appear earlier in the scan order, thereby restoring the validity of entropy coding assumptions while maintaining the benefits of transform skipping
Solution Approach 2:
The patent performs a preliminary permutation of the residual data block before the quantization and entropy coding steps. This preliminary reorganization of data ensures that when entropy coding is applied, the data distribution conforms to expected statistical patterns, preventing the invalidation of coding assumptions that would otherwise occur with transform skipping
Data Source
AI summary
Methods and devices for modified coding of blocks of residuals in the case of transform skipping. To better align the data with assumptions upon which the entropy coding scheme is based, the block of residual data is permuted at the encoder prior to entropy coding. The block of reconstructed data is then inverse permuted at the decoder to recover the reconstructed block of residuals. The permutation may include 180 rotation of the block, which may be of particular benefit in the case of intra-coded residuals.