Video Compression Prediction Models Using Adaptive Template Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in efficiently utilizing spatial and temporal redundancies for improved compression efficiency, particularly in deriving accurate prediction models for video encoding and decoding.
Innovation Solution
The method involves determining a similarity value between candidate templates and the current block to derive a prediction model, using intra template matching prediction (IntraTMP) to select the most suitable template for encoding and decoding, and applying linear or non-linear functions to refine the prediction process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple candidate templates are evaluated to derive accurate prediction models, then prediction accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent segments the template matching process by dividing candidate templates into different types (e.g., left template, top template, L-shaped template) and evaluating them separately. This segmentation allows the system to manage computational complexity by processing templates in organized groups rather than as a monolithic operation, while still maintaining comprehensive evaluation for accuracy.
Solution Approach 2:
The patent employs parameter changes by using different similarity metrics and evaluation criteria for different template types. The system dynamically adjusts which templates to evaluate and how to evaluate them based on block characteristics, thereby optimizing the balance between computational complexity and prediction accuracy through adaptive parameter selection.
2Productivity
If template matching is performed across multiple regions and candidate templates, then compression efficiency is improved, but processing time increases
Solution Approach 1:
The patent applies preliminary action by pre-identifying and categorizing candidate templates into specific types (left, top, L-shaped) before the actual matching process. This preliminary organization allows the decoder to quickly determine which templates to evaluate based on block position and characteristics, reducing processing time while maintaining compression efficiency through targeted template selection.
Solution Approach 2:
The system implements partial action by selectively evaluating only the most relevant candidate templates based on block characteristics and position, rather than exhaustively processing all possible templates. This selective approach achieves sufficient compression efficiency by focusing computational resources on the most promising candidates while reducing overall processing time.
3Measurement precision
If similarity values are calculated for all candidate templates, then template selection accuracy is improved, but computational load increases
Solution Approach 1:
The patent applies local quality by calculating similarity values with different levels of detail for different template types and block regions. The system concentrates computational resources on evaluating templates in regions and positions where high accuracy is most beneficial, while using simplified evaluation for less critical areas, thereby optimizing the balance between selection accuracy and computational load.
Solution Approach 2:
The system implements self-service through adaptive template selection where the decoding process automatically determines which candidate templates require full similarity evaluation based on block characteristics. The decoder serves itself by intelligently identifying which templates warrant detailed analysis versus those that can be evaluated more simply, reducing overall computational load while maintaining accuracy where needed.
Data Source
AI summary
In a method, a similarity value is calculated based on samples in at least one of a plurality of candidate templates of a current block and samples of the current block corresponding to the samples in the at least one of the plurality of candidate templates. The plurality of candidate templates includes at least one of a left template, a top template, and an L-shaped template that is adjacent to the current block. A template of the current block is determined as one of the plurality of candidate templates based on whether the similarity value is larger than a threshold value or corresponds to the one of the plurality of candidate templates. A prediction model is determined based on the template of the current block and a template of a reference block of the current block. The current block is encoded based on the prediction model and the reference block.