Texture Synthesis Clustering for Video Coding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding methods, such as HEVC, face inefficiencies in encoding complex textures due to their reliance on predictive coding, which struggles with high-complexity and irregular textures, leading to poor bit rate efficiency and subjective quality.
Innovation Solution
The proposed solution involves refining texture synthesis by clustering video images using polynomial fitting to identify synthesizable regions, encoding these regions separately, and using motion compensation and luminance adjustment to improve encoding efficiency and subjective quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional predictive coding (intra prediction or motion compensation) is used for high-complexity and irregular textures, then the encoding process is simple, but the coding efficiency is poor and bit rate consumption is high
Solution Approach 1:
The image is segmented into texture regions and non-texture regions using clustering algorithms. Texture regions are identified by analyzing pixel similarity and spatial continuity, then processed separately using texture synthesis rather than conventional predictive coding. This segmentation allows each region to be encoded with the most appropriate method, improving overall coding efficiency for complex textures.
Solution Approach 2:
The invention changes the encoding parameters by introducing texture synthesis parameters (such as patch size, patch position, and synthesis algorithm parameters) instead of using traditional prediction parameters. This parameter change enables efficient representation of high-complexity textures by capturing their statistical properties rather than attempting pixel-wise prediction.
2Manufacturing precision
If texture synthesis is applied to high-complexity textures, then subjective quality improves, but device complexity increases due to additional processing steps
Solution Approach 1:
The clustering and texture region identification is performed as a preliminary step before encoding. By pre-identifying texture regions and preparing synthesis parameters in advance, the actual encoding process becomes more efficient. The preliminary segmentation and parameter extraction reduce the computational burden during the main encoding phase, balancing quality improvement with processing complexity.
Solution Approach 2:
Texture synthesis works by copying and replicating small patches from reference regions to synthesize larger texture areas. This copying mechanism, combined with polynomial fitting for smooth transitions, achieves high subjective quality for complex textures while maintaining computational efficiency through parameter-based representation rather than full pixel processing.
3Measurement precision
If polynomial fitting is used for cluster refinement, then accuracy of texture region identification improves, but computational cost increases
Solution Approach 1:
Polynomial fitting is applied selectively only to texture regions that are identified by initial clustering, not to the entire image. This partial application of the computationally intensive operation reduces overall computational cost while maintaining high accuracy for the critical texture regions where it matters most for encoding quality.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present disclosure relates to encoding a decoding video employing texture coding. In particular, a texture region is identified within a video picture and a texture patch is determined for said region. Clustering is performed to identify a texture region within the video image. The clustering is further refined. In particular, one or more brightness parameters of a polynomial is determined by fitting the polynomial to the identified texture region. In the identified texture region, samples are detected with a distance to the fitted polynomial exceeding a first threshold and identify a refined texture region as the texture region excluding one or more of the detected samples. Finally, the refined texture region is encoded separately from portions of the video image not belonging to the refined texture region.