Intra Prediction with Reordered Templates and Linear Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in efficiently encoding and decoding video sequences, particularly for screen content, due to the complexity and redundancy in predicting intra prediction modes, leading to suboptimal compression and quality trade-offs.
Innovation Solution
The implementation of reconstructed-reordered template matching prediction (RRTMP) with a linear filter model (LFM) for intra prediction, which enhances the prediction accuracy by reordering and filtering reference blocks to improve template matching, thereby optimizing the encoding and decoding processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional intra prediction methods are used, then encoding complexity is reduced, but prediction accuracy and compression performance deteriorate
Solution Approach 1:
The patent applies preliminary action by reordering reference blocks before template matching is performed. The reference blocks are sorted based on their positional relationship with the current block, with blocks closer to the current block placed earlier in the order. This pre-processing step improves prediction accuracy by ensuring that more relevant reference blocks are considered first, without significantly increasing encoding complexity since the reordering follows a simple positional criterion.
Solution Approach 2:
The patent changes the parameter of reference block ordering from conventional sequential or spatial ordering to a reordered sequence based on positional distance metrics. By modifying the ordering parameter of reference blocks, the template matching process can more effectively identify suitable reference blocks, thereby improving prediction accuracy while maintaining manageable encoding complexity through efficient sorting algorithms.
2Productivity
If template matching prediction is applied, then compression performance improves, but encoding and decoding time increases
Solution Approach 1:
The patent applies preliminary action by pre-ordering reference blocks before the template matching process. This reordering is performed once per prediction unit and establishes an optimal search sequence in advance, allowing the template matching to proceed more efficiently by examining reference blocks in order of their likelihood to provide good matches, thereby reducing the effective search time while maintaining compression performance.
Solution Approach 2:
The patent modifies the temporal parameter by changing the reference block sequence from conventional ordering to position-based reordering. This parameter change optimizes the template matching process by ensuring that the most relevant reference blocks are evaluated first, improving compression performance while limiting the time increase through efficient ordering algorithms that avoid exhaustive searches.
3Manufacturing precision
If reference blocks are not reordered, then processing speed is maintained, but prediction quality for screen content deteriorates
Solution Approach 1:
The patent applies preliminary action by reordering reference blocks based on their positional distance to the current block before template matching. For screen content, this reordering ensures that reference blocks with similar spatial characteristics are examined first, significantly improving prediction quality. The reordering uses simple positional calculations that maintain processing speed by avoiding complex computations.
Solution Approach 2:
The patent changes the ordering parameter of reference blocks from conventional sequences to position-based sequences calculated using distance metrics. This parameter change improves prediction quality for screen content by aligning the reference block evaluation order with the spatial characteristics of screen content, while the efficiency of the positional calculation maintains acceptable processing speed.
Data Source
AI summary
A coder (encoder or decoder) determines, for a current block (CB), a reference block (RB) based on a template matching (TM) cost associated with a first reference template of the RB and a first current template of the CB. The coder further determines, based on the first reference template matching the first current template being flipped in a direction, coefficients of a linear spatial filter based on filtered values, resulting from the linear spatial filter applied to first samples of a second reference template of the RB, being set to corresponding values of second samples of a second current template flipped in the direction. An adjusted RB is generated by applying the linear spatial filter with the determined coefficients to the RB. The coder codes the CB based on a residual of the CB determined based on the adjusted RB.