Image Encoding Motion Prediction Model Selection for Boundary Distortion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The HEVC standard's deblocking processing method fails to effectively reduce image distortion at block boundaries due to its reliance on translational motion models, neglecting the use of different motion prediction models in adjacent blocks.
Innovation Solution
An image encoding and decoding method that selects between translational and non-translational motion models for each block, performing filtering only when a non-translational model is used for adjacent blocks to reduce boundary distortion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If deblocking processing is performed on all block boundaries, then image distortion at boundaries is reduced, but processing complexity increases and computational resources are wasted on blocks that do not require filtering
Solution Approach 1:
The patent applies different processing treatments to different block boundaries based on the motion prediction models used in adjacent blocks. Filtering is selectively applied only to boundaries where a non-translational motion model is used in at least one adjacent block, while uniform translational motion blocks receive no filtering. This local differentiation resolves the contradiction by concentrating processing resources where they are actually needed to reduce distortion, rather than uniformly processing all boundaries.
2Manufacturing precision
If filtering is applied to reduce boundary distortion, then image quality improves, but processing time and computational load increase
Solution Approach 1:
The patent changes the parameter used to determine filtering application from a universal criterion to one based on motion prediction model types. By using the motion model parameter (translational vs. non-translational) as the decision criterion, the system efficiently identifies which blocks require filtering without extensive analysis, thus reducing processing time while maintaining image quality improvements where needed.
3Manufacturing precision
If filtering is performed on all blocks, then boundary distortion is reduced uniformly, but computational resources are wasted on blocks with translational motion models that do not require filtering
Solution Approach 1:
The patent implements local quality by making filtering application dependent on the specific characteristics of adjacent blocks' motion prediction models. Blocks with non-translational motion models (which produce more distortion) receive filtering, while blocks with uniform translational motion (which produce less distortion) skip filtering. This selective approach reduces computational resource waste while maintaining effective distortion reduction where needed.
Data Source
AI summary
An image encoding method includes: selecting one motion prediction model from a plurality of motion prediction models including a translational motion model and a non-translational motion model for each of blocks in the image; performing motion prediction using the selected motion prediction model to generate a prediction image; generating a reconstructed image using the prediction image; determining that filtering is to be performed if the non-translational motion model is selected for at least one of a first block and a second block that is adjacent to the first block; and performing the filtering on a boundary between the reconstructed image for the first block and the reconstructed image for the second block if it is determined that the filtering is to be performed.