Template Matching Intra-Prediction Coding for Video Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing intra-frame predictive coding methods based on L-shaped-template matching struggle with accuracy when compressing images with complex and variable textures, as the correlation between the image texture in the prediction unit and the L-shaped template is low, leading to reduced prediction accuracy.
Innovation Solution
The method employs multiple template shapes (L-shaped, straight-line, and 1-shaped) to calculate predicted pixel values, determining a scanning sequence based on energy distribution to increase matching accuracy and select the optimal predicted pixel value, which is then used for coding, and sends template information to the decoder for decoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If L-shaped template matching is used for intra-frame prediction, then prediction quality is improved for simple textures, but prediction accuracy deteriorates when image texture is complex and variable
Solution Approach 1:
The patent dynamically selects template shapes (L-shaped, straight-line, or 1-shaped) based on the actual texture characteristics of the image region being predicted. This dynamic adaptation allows the system to switch between different template configurations to match varying texture patterns, thereby maintaining high prediction accuracy across diverse image content
Solution Approach 2:
The patent changes the shape parameter of the template based on texture complexity analysis. By adjusting the template shape parameter to match the underlying texture structure (e.g., using L-shaped for corner patterns, straight-line for linear patterns), the system optimizes the correlation between template and prediction unit, resolving the contradiction between simple and complex texture performance
2Measurement precision
If multiple template shapes are used to improve texture matching probability, then prediction accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent segments the template matching process into distinct shape categories (L-shaped, straight-line, 1-shaped templates). By dividing the prediction process into these segments and selecting only the most appropriate shape for each region based on texture analysis, the system avoids the computational burden of evaluating all possible template configurations, thus balancing accuracy improvement with complexity management
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
According to intra-frame predictive coding and decoding methods based on template matching, an array scanning method and apparatus, and an apparatus that are disclosed in embodiments of the present invention, templates of at least two shapes are used to calculate predicted pixel values of a to-be-predicted unit. Intra-frame prediction based on template matching is performed based on a correlation between a texture of a template region and a texture of a to-be-predicted unit. Compared with a template of a single shape, templates of different shapes can represent different textures; when a correlation between a texture of a template of a shape and a texture of a to-be-predicted unit is low, a template of another shape may be selected and used, and a matching degree between a texture of a template and a texture of a to-be-predicted unit can be increased. In this way, an optimal predicted pixel value is determined from at least two predicted pixel values, and accuracy of the optimal predicted pixel value is higher than accuracy of a predicted pixel value determined by using a template of a single shape.