Segmentation-Based Deblocking Filtering for Object Boundary Preservation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression techniques, such as VVC, require improved deblocking filtering to handle increasing image sizes and resolutions, necessitating a new approach to enhance coding efficiency and image quality without filtering at object boundaries.
Innovation Solution
A video coding method and apparatus that utilizes segmentation information to partition restored areas into objects, adjust filter lengths, and determine the type of deblocking filter based on spatial changes in sub-areas to avoid filtering at object boundaries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If deblocking filtering is applied to all boundaries in restored areas, then blocking artifacts are reduced, but image quality deteriorates due to unnecessary filtering at object boundaries
Solution Approach 1:
The patent segments the restored area into multiple coding units and further divides boundaries into filtering boundaries and non-filtering boundaries based on segmentation information. This allows selective application of deblocking filtering only where needed, avoiding unnecessary filtering at object boundaries while maintaining artifact reduction in appropriate regions.
Solution Approach 2:
The patent applies different filtering strategies to different regions: strong filtering is applied to certain boundaries while weak or no filtering is applied to others based on segmentation information. This local differentiation optimizes image quality by preserving object boundaries while reducing blocking artifacts in appropriate regions.
2Productivity
If filtering strength is increased to reduce blocking artifacts, then coding efficiency improves, but image quality deteriorates due to over-filtering
Solution Approach 1:
The patent dynamically adjusts filter strength based on boundary characteristics and segmentation information. The filtering strength is not fixed but adapts to local requirements, applying strong filtering where blocking artifacts are present and weak or no filtering where object boundaries exist, thus optimizing both coding efficiency and image quality.
Solution Approach 2:
The patent changes the filtering parameter (filter strength) based on the type of boundary identified through segmentation. Different filter strengths are selected for different boundary types, allowing the system to optimize coding efficiency when needed while preserving image quality by reducing filtering strength at object boundaries.
3Ease of operation
If deblocking filtering is applied uniformly across all boundaries, then processing simplicity is maintained, but coding efficiency decreases due to unnecessary filtering operations
Solution Approach 1:
The patent uses segmentation information to classify boundaries into filtering and non-filtering categories, enabling selective processing. This maintains relative simplicity through automated classification while significantly improving coding efficiency by avoiding unnecessary filtering operations at object boundaries.
Solution Approach 2:
The patent extracts and identifies boundaries that should not be filtered based on segmentation information. By taking out object boundaries from the set of boundaries requiring filtering, the system reduces unnecessary filtering operations and improves coding efficiency while maintaining simple processing logic.
Data Source
AI summary
A video coding method and an apparatus using a deblocking filtering based on segmentation information are disclosed. The video coding method and the apparatus perform deblocking filtering using segmentation information of an image to avoid performing filtering at the boundary of an object within the image.