Low-Frequency Non-Separable Transforms for Intra/Inter Video Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding systems face inefficiencies in compressing digital video signals, particularly in handling block sizes and slice characteristics that require non-separable inverse transforms, leading to suboptimal performance in delay and complexity management.
Innovation Solution
A device determines the size of a video block and its slice characteristics to selectively apply separable or non-separable inverse transforms based on predetermined values, using high-level syntax elements to indicate the appropriate transform type for inter-predicted or intra-predicted blocks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If non-separable inverse transform is applied to video blocks in P-slices or B-slices, then video coding efficiency is improved, but computational complexity increases
Solution Approach 1:
The patent applies non-separable inverse transform selectively to specific video blocks based on local characteristics (block size, slice type, temporal depth) rather than uniformly to all blocks. This localized application improves coding efficiency for blocks that benefit from the transform while avoiding unnecessary complexity for blocks where traditional transforms suffice.
Solution Approach 2:
The patent changes the transform type parameter dynamically based on multiple conditions including slice type (P-slice, B-slice), block size, and temporal depth. By adjusting this parameter according to local video characteristics, the system optimizes the balance between coding efficiency and computational complexity.
2Manufacturing precision
If non-separable inverse transform is applied to all video blocks, then decoding quality is improved, but processing delay increases
Solution Approach 1:
The patent applies non-separable inverse transform only to specific video blocks where it provides significant quality improvement, rather than to all blocks. This selective application reduces the total processing time while maintaining high decoding quality for the blocks that require enhanced transformation.
Solution Approach 2:
The patent applies the computationally intensive non-separable inverse transform partially - only to blocks where the condition (slice type = P-slice or B-slice, block size < threshold, temporal depth < threshold) is met. This partial application achieves sufficient decoding quality without the excessive processing delay that would result from applying the transform to all blocks.
3Productivity
If non-separable inverse transform is used for small video blocks, then transform efficiency is improved, but device complexity increases
Solution Approach 1:
The patent applies non-separable inverse transform specifically to small video blocks (block size < predetermined threshold) where the transform provides superior efficiency. This localized application targets the blocks that benefit most from the advanced transform while avoiding the complexity overhead for larger blocks where traditional separable transforms remain effective.
Solution Approach 2:
The patent dynamically changes the transform type parameter based on block size, switching to non-separable inverse transform for small blocks where it provides efficiency gains, and maintaining traditional transforms for larger blocks. This parameter adaptation optimizes transform efficiency while managing device complexity.
Data Source
Figure 1A
Figure 1B
Figure 1C
AI summary
A device (e.g., a decoder for video decoding) may obtain a slice comprising a video block. The device may determine the size of the video block. The device may determine, based on a slice characteristic of the slice and the size of the video block, an inverse transform technique associated with the video block. The device may decode the slice based on the inverse transform technique associated with the video block. For example, the slice characteristic of the slice may include a slice type. The device, based on a determination that the slice type of the slice is a predicted slice (P-slice) or a bidirectional slice (B-slice) and a determination that the size of the video block is less than a predetermined value, may determine that a non-separable inverse transform associated with the video block is to be performed.