Residual Block Rotation for Transform Skip Entropy Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques face inefficiencies in entropy coding due to the positioning of residual data with higher energy values, as they are not optimally aligned at the top left corner of the residual block when a transform is skipped, affecting compression efficiency.
Innovation Solution
The proposed solution involves a rotation unit in video encoders and decoders that determines whether to rotate residual blocks based on the type of boundaries at their edges, positioning higher energy residual data at the top left corner for improved entropy coding efficiency, either by explicitly signaling a rotation value or independently determining the rotation based on edge types.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If transform skipping is applied to residual blocks, then computational complexity is reduced, but entropy coding efficiency deteriorates due to mispositioning of high energy residual data
Solution Approach 1:
The patent applies preliminary rotation to residual blocks before entropy coding when transform skipping is enabled. This preliminary action repositions high energy residual data to the top left corner in advance, ensuring optimal entropy coding efficiency without requiring complex transform operations, thus resolving the contradiction between reduced computational complexity and maintained coding efficiency
Solution Approach 2:
The patent changes the spatial arrangement parameter of residual data by rotating the residual block. This parameter change (rotation angle) transforms the positioning of residual data elements, concentrating high energy values at the top left corner to match entropy coding expectations, thereby maintaining coding efficiency while keeping the transform skip mode computationally simple
2Loss of information
If residual blocks are rotated to align high energy data, then entropy coding efficiency is improved, but processing complexity increases
Solution Approach 1:
The patent applies local quality by selectively rotating only those residual blocks where transform skipping is enabled and high energy data misalignment is detected. Not all residual blocks undergo rotation, only the specific local regions (residual blocks) that benefit from it, thus improving entropy coding efficiency without unnecessarily increasing processing complexity for all blocks
Solution Approach 2:
The rotation operation is performed as a preliminary step before entropy coding, with simple detection criteria based on boundary types. This preliminary action with minimal processing requirements achieves the alignment needed for efficient entropy coding while keeping the added complexity low
3Manufacturing precision
If rotation is applied to residual blocks, then residual data positioning is optimized for entropy coding, but decoding complexity increases
Solution Approach 1:
The patent incorporates feedback by signaling rotation information in the bitstream from encoder to decoder. The decoder receives this feedback and applies the exact same rotation operations that the encoder performed, ensuring synchronized processing. This feedback mechanism optimizes residual data positioning at the encoder while keeping decoder complexity manageable through explicit instruction rather than complex analysis
Solution Approach 2:
Both encoder and decoder perform rotation as a preliminary action before entropy coding/decoding respectively. The rotation is applied early in the process based on simple boundary type detection, optimizing data positioning without requiring complex operations during the main coding/decoding phases
4Productivity
If transform is skipped for residual blocks, then coding speed is improved, but compression performance deteriorates due to non-optimal residual data distribution
Solution Approach 1:
The patent applies preliminary rotation to residual blocks before entropy coding when transform skipping is enabled. This preliminary action repositions high energy residual data to the top left corner in advance, ensuring optimal entropy coding efficiency without requiring complex transform operations, thus resolving the contradiction between reduced computational complexity and maintained coding efficiency
Solution Approach 2:
The patent changes the spatial arrangement parameter of residual data by rotating the residual block. This parameter change (rotation angle) transforms the positioning of residual data elements, concentrating high energy values at the top left corner to match entropy coding expectations, thereby maintaining coding efficiency while keeping the transform skip mode computationally simple
Data Source
AI summary
Techniques are described for coding residual data of a prediction residual block with transform skipping. A transform may be skipped for a residual block when the residual block is coded using either a lossless coding mode or a lossy coding mode in a transform skip mode. According to the techniques, based on a transform being skipped for a residual block, a rotation unit included in a video encoder or a video decoder determines whether to rotate the residual block prior to coding residual data of the residual block. In some examples, a rotation value may be explicitly signaled between the video encoder and the video decoder. In other examples, the video encoder and the video decoder may each independently determine whether to rotate the residual block based on a type of boundary at two or more edges of the residual block.