Intra Prediction Boundary Filtering Control for Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multi-view video coding techniques suffer from inefficiencies due to the application of boundary filtering and neighboring sample filtering, which can degrade image quality, especially in depth video and screen content materials with sharp transitions, and lack control mechanisms for disabling these filters in lossy coding modes.
Innovation Solution
A method and apparatus for adaptive control of boundary filtering and neighboring sample filtering in intra prediction coding, where filtering-disable-flags are used to selectively disable filters for specific intra prediction modes, particularly in depth and screen content data, based on block sizes and coding levels, allowing for flexible filtering control at various video data levels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If boundary filtering and neighboring sample filtering are applied to all intra prediction modes, then blocking artifacts are reduced, but image quality degrades in depth video and screen content materials with sharp transitions
Solution Approach 1:
The patent applies different filtering strategies to different intra prediction modes and block types. Boundary filtering is selectively applied only to DC prediction modes and certain angular modes, while being disabled for vertical and horizontal modes in depth video and screen content. This local differentiation allows the system to maintain artifact reduction where needed while preserving sharp transitions in depth maps and screen content.
Solution Approach 2:
The patent introduces dynamic control mechanisms through filtering-disable-flags that can be adjusted at sequence, picture, slice, or block level. The system can adaptively enable or disable boundary filtering based on the specific content type (depth video, screen content, or conventional video) and prediction mode, allowing flexible optimization of image quality versus artifact reduction trade-offs.
2Reliability
If boundary filtering is always enabled for DC mode, then blocking artifacts are reduced, but coding efficiency decreases due to unnecessary filtering in lossy coding modes
Solution Approach 1:
The patent changes the filtering parameter state based on coding mode and block characteristics. Through the filtering-disable-flags, the system can transition between filtered and unfiltered states depending on whether the current block is lossy-coded, the prediction mode type, and the block size. This parameter adjustment enables the system to maintain coding efficiency by avoiding unnecessary filtering operations while preserving artifact reduction where beneficial.
3Manufacturing precision
If filtering control mechanisms are added to disable boundary filtering for specific modes, then image quality improves for depth video and screen content, but device complexity increases
Solution Approach 1:
The filtering-disable-flags serve multiple functions: they control boundary filtering for DC modes, manage neighboring sample filtering for angular modes, and adapt to different content types (depth video, screen content, conventional video). By using a single flag mechanism with multiple conditions, the patent achieves comprehensive filtering control without proportionally increasing system complexity.
Solution Approach 2:
The patent segments the filtering control into multiple independent decision points at different levels (sequence level, picture level, slice level, block level). Each level can independently configure filtering behavior for specific prediction modes, allowing fine-grained control while distributing the complexity across manageable hierarchical levels rather than requiring a single complex control system.
Data Source
AI summary
A method and apparatus for Intra prediction coding in multi-view video coding, three-dimensional video coding, or screen content video coding are disclosed. A first filtering-disable-flag associated with a high-level video data is determined to indicate whether to disable at least one filter from a filter group. If the first filtering-disable-flag is asserted, one or more selected Intra prediction modes from an Intra prediction mode group is determined, and at least one filter from the filter group for the current block is skipped if the current Intra prediction mode of the current block belongs to the selected Intra prediction modes. The system may further determine a second filtering-disable-flag associated with a low-level video data corresponding to a current block level or a higher level than the current block level to disable said at least one filter from a filter group for the low-level video data.