Intra-Prediction Transform Mapping to Reduce Video Signaling Bits
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Next-generation video contents with high spatial resolution, high frame rate, and high dimensionality require more efficient transform methods to reduce memory storage, memory access rate, and processing power demands.
Innovation Solution
An adaptive transform mapping method dependent on intra prediction mode is proposed, which includes checking for a transform skip and determining transform types based on intra prediction mode, with optional grouping and binarization of transform indices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If transform is applied to video blocks, then encoding efficiency is improved, but processing complexity and computational load increase
Solution Approach 1:
The patent changes the parameters of transform selection by adapting the transform type to the intra prediction mode. Different transform types are selected based on the prediction mode (e.g., planar, angular, etc.), which optimizes the transform for each specific coding situation and improves encoding efficiency while managing computational complexity through targeted transformations rather than applying all possible transforms universally.
Solution Approach 2:
The transform selection is made dynamic by linking it to the intra prediction mode. The system dynamically determines which transform type to apply based on the prediction mode of the current block, allowing the transform process to adapt to varying video content characteristics and coding conditions, thereby optimizing the balance between encoding efficiency and processing complexity.
2Productivity
If multiple transform types are used to adapt to different prediction modes, then encoding performance is improved, but signaling overhead increases
Solution Approach 1:
The intra prediction mode serves as an intermediary that indirectly determines the transform type. Instead of directly signaling multiple transform types, the system uses the prediction mode (which is already being signaled) as a mediator to select the appropriate transform, thereby avoiding additional signaling overhead while still achieving adaptive encoding performance.
Solution Approach 2:
The intra prediction mode signaling serves a dual function: it signals the prediction type for decoding purposes and simultaneously determines the transform type for encoding/decoding efficiency. This multi-functionality eliminates the need for separate transform type signaling, reducing signaling overhead while maintaining encoding performance across different prediction modes.
3Productivity
If transform index is binarized using truncated unary method, then coding efficiency is improved, but implementation complexity increases
Solution Approach 1:
The patent replaces complex transform selection mechanisms with a simplified binarized transform index approach. Instead of directly encoding multiple transform types, the system uses a binarized index that can be efficiently coded using truncated unary method, substituting a more complex selection process with a simpler, more efficient coding mechanism that improves coding efficiency while managing implementation complexity.
Data Source
AI summary
The present disclosure, in a method of decoding a video signal, provides a method including checking whether a transform skip is applied to a current block; acquiring a transform index indicating transform types applied to a horizontal direction and a vertical direction of the current block from among a plurality of predefined transform types from the video signal when the transform skip is not applied to the current block; and performing an inverse primary transform on the current block using the transform types indicated by the transform index, wherein the transform types applied to the horizontal direction and the vertical direction of the current block are determined from among transform types defined according to an intra prediction mode of the current block based on the transform index.