Constrained Intra Prediction Reference Sample Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding technologies, such as HEVC, face challenges in efficiently encoding video data due to spatial reconstruction errors caused by encoder-decoder mismatches, especially in unreliable networks and lossy compression scenarios, where constrained intra prediction methods are limited in flexibility and effectiveness.
Innovation Solution
The implementation of a method that enables constrained intra prediction by selecting a reference sample for encoding a current block based on the availability of neighboring samples from intra block copy (IBC) coded blocks, using a Quad Tree Binary Tree (QTBT) structure and cross-component linear model prediction, to enhance encoding efficiency and robustness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional intra prediction is used, then encoding flexibility is maintained, but spatial reconstruction errors occur due to encoder-decoder mismatches in lossy compression scenarios
Solution Approach 1:
The patent changes the parameter of reference sample selection by introducing a flag-based mechanism that switches between conventional intra prediction and constrained intra prediction modes. When the constrained intra prediction flag is set, the encoder uses only intra-coded neighboring blocks as reference samples, excluding inter-coded blocks, thereby preventing spatial reconstruction errors while maintaining encoding adaptability through parameter control.
Solution Approach 2:
The patent implements dynamic reference sample selection by conditionally enabling constrained intra prediction based on the constrained intra prediction flag and the coding mode of neighboring blocks. This dynamic approach allows the system to adaptively choose between conventional and constrained prediction methods, resolving the contradiction between reliability and flexibility by making the reference sample selection process responsive to current encoding conditions.
2Reliability
If constrained intra prediction is enabled with strict reference sample selection, then spatial noise propagation is reduced, but encoding effectiveness decreases due to limited reference sample availability
Solution Approach 1:
The patent enhances the universality of the constrained intra prediction mechanism by allowing it to work effectively with both intra-coded and inter-coded neighboring blocks through the flag-based control. The system universally applies the constrained prediction rule (using only intra-coded references when flag is set) across different coding scenarios, ensuring spatial noise resistance while maintaining encoding effectiveness through versatile applicability.
Solution Approach 2:
The patent uses parameter changes in the form of the constrained intra prediction flag to dynamically adjust reference sample selection criteria. When the flag is enabled, the system changes the selection parameter to exclude inter-coded blocks, reducing spatial noise propagation. When disabled, conventional selection is used to maintain encoding effectiveness, thus resolving the contradiction through parameter-based adaptability.
3Reliability
If neighboring samples from inter-coded blocks are excluded from reference samples, then encoder-decoder mismatch is prevented, but the number of available reference samples decreases
Solution Approach 1:
The patent implements dynamic reference sample selection based on the coding mode of neighboring blocks and the constrained intra prediction flag. When the flag is set and a neighboring block is inter-coded, the system dynamically excludes it from reference samples to prevent encoder-decoder mismatch. When the flag is not set or the neighboring block is intra-coded, the sample is included, thus maintaining encoder-decoder consistency while maximizing the number of available reference samples through conditional dynamics.
4Adaptability or versatility
If flexible block partitioning is implemented using QTBT structure, then encoding adaptability is improved, but processing complexity increases
Solution Approach 1:
The patent applies segmentation through the Quad Tree Binary Tree (QTBT) structure, which divides the current block into smaller sub-blocks that can be independently coded with different prediction modes. This segmentation provides encoding adaptability by allowing flexible block partitioning according to content characteristics, while the systematic tree-based approach keeps processing complexity manageable through structured decomposition.
Data Source
AI summary
A method of selecting a reference sample used to encode a current block in an encoded video bitstream using at least one processor includes determining that a constrained intra prediction (CIP) mode is enabled; determining that a neighboring sample of the current block belongs to a coding block that is coded using intra block copy (IBC) coding; based on the determining that the neighboring sample belongs to the coding block that is coded using the IBC coding, selecting the neighboring sample as the reference sample; and encoding the current block using intra prediction coding based on the reference sample.