Video Decoding Block Prediction Using Filtered Reference Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Intra template matching prediction in video coding faces challenges in accurately predicting block units due to difficulties in deriving precise prediction models, leading to decreased coding efficiency.
Innovation Solution
A method and device for predicting a block unit using multiple reference matching blocks and filter models, where filtered reference blocks are generated based on derived filter models and combined to reconstruct the block unit, with weighting parameters determined by comparing template costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a prediction mode using the prediction model is used to predict most blocks in the current frame, then the coding efficiency is decreased, but using multiple reference matching blocks and filter models increases the device complexity
Solution Approach 1:
The patent segments the prediction process into multiple stages: first selecting multiple reference matching blocks from the current frame, then deriving multiple reference filter models from template samples, and finally combining filtered reference blocks to generate the predicted block. This segmentation allows the system to handle complex prediction tasks through manageable steps, improving coding efficiency while maintaining controlled complexity through structured processing.
Solution Approach 2:
The patent introduces multiple dimensions to the prediction process by using multiple reference matching blocks (spatial dimension) and multiple reference filter models (parameter dimension). Instead of relying on a single reference block or filter model, the system operates in a multi-dimensional search space, enabling more accurate predictions and improving coding efficiency through enhanced prediction accuracy.
2Measurement precision
If multiple reference matching blocks and filter models are used to predict block units, then the prediction accuracy is improved, but the processing time increases
Solution Approach 1:
The patent performs preliminary actions by pre-deriving multiple reference filter models from template samples before the actual prediction process. The reference matching blocks are also selected and prepared in advance. This preliminary preparation reduces the computational burden during the prediction step, allowing the system to achieve high prediction accuracy through multiple reference blocks while minimizing processing time by avoiding redundant calculations.
Solution Approach 2:
The patent applies partial action by selecting a subset of the most relevant reference matching blocks and filter models rather than using all possible references. The system determines the optimal number and selection of reference blocks based on the specific block being predicted, applying only the necessary computational resources to achieve accurate predictions without excessive processing time consumption.
Data Source
AI summary
A method of decoding video data performed by an electronic device is provided. The method receives the video data and determines a block unit from a current frame included in the video data. The method further selects, for the block unit, multiple reference matching blocks from the current frame, determines, from the current frame, a current matching template, neighboring the block unit, and multiple reference matching templates, and determines multiple reference filter models. Each of the reference matching templates neighbors a corresponding one of the reference matching blocks. Each of the reference filter models is derived based on a corresponding one of the reference matching templates and associated with a corresponding one of the reference matching blocks. The method then filters the reference matching blocks based on the reference filter models to generate multiple filtered reference blocks, and reconstructs the block unit based on the filtered reference blocks.