Video Coding Intra Prediction Using Multiple Reference Lines
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video coding methods rely solely on the closest reconstructed pixel line for intra prediction, which may include different textures or signal noise, leading to suboptimal prediction accuracy due to the lack of utilization of multiple reference lines.
Innovation Solution
A method and device for video coding that employs multiple reference lines for intra prediction, allowing the encoder to select and signal multiple reference lines, enabling the decoder to determine the optimal reference lines based on directional modes, thereby improving prediction accuracy without the need for explicit flag signaling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If only the closest reconstructed pixel line is used for intra prediction, then the coding complexity is low and the decoding process is simple, but the prediction accuracy deteriorates due to texture differences and signal noise
Solution Approach 1:
The patent divides the reference pixel lines into multiple segments (closest pixel line and other adjacent pixel lines) and selectively uses different segments for prediction based on the directional mode. This segmentation allows the system to overcome the limitations of using only a single reference line while maintaining manageable complexity through structured processing.
Solution Approach 2:
The patent extends the reference from a single-dimensional closest pixel line to multiple adjacent pixel lines, adding a dimensional aspect to the reference selection. By considering multiple reference lines in addition to the closest one, the system achieves better prediction accuracy without proportionally increasing complexity.
2Measurement precision
If multiple reference lines are exploited to generate reference pixels and predictors, then the prediction accuracy is improved, but the bitstream complexity increases due to flag signaling requirements
Solution Approach 1:
The patent enables the decoder to autonomously determine which reference lines to use based on the directional mode information already present in the bitstream. The decoder derives the reference line selection internally without requiring explicit signaling flags, making the system self-sufficient and avoiding additional bitstream overhead.
Solution Approach 2:
The patent changes the parameter of reference line selection from being explicitly signaled to being derived from existing directional mode parameters. By repurposing existing parameters (directional modes) to control reference line selection, the system avoids increasing bitstream complexity while still enabling multiple reference line usage.
3Reliability
If multiple adjacent pixel lines are used for prediction, then the impact of noise and texture differences is reduced, but the processing complexity increases
Solution Approach 1:
The patent implements dynamic reference line selection where the number and type of reference lines used varies based on the directional mode. Different directional modes utilize different numbers of reference lines (e.g., planar mode uses multiple lines while directional modes may use fewer), allowing the system to adapt processing complexity to the specific prediction scenario and maintain reliability without uniform high complexity.
Data Source
AI summary
A method of decoding a bitstream by an electronic device is provided. A directional mode of a block unit is determined from the bitstream. A plurality of reference lines neighboring to the block unit are determined based on the block unit. Each of the reference lines includes a plurality of neighboring samples. A plurality of reference samples are determined along an orientation of the directional mode based on the neighboring samples in at least one of the reference lines. A smoothed sample is generated for reconstructing an image frame including the block unit based on the reference samples.