Intra Prediction Mode Selection for Image Encoding Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image compression methods, particularly in planar modes of intra-prediction, face challenges with low prediction accuracy when top or left reference samples have similar values or significant differences, leading to inefficient encoding and decoding processes.
Innovation Solution
A new planar prediction method is proposed that selects reference samples based on the planar mode type, using a combination of neighboring blocks to derive the most probable mode candidates and generate prediction blocks, thereby improving prediction performance and encoding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a traditional planar mode of intra-prediction is applied, then the encoding process is simple, but prediction accuracy is low when top or left reference samples have similar values or significant differences
Solution Approach 1:
The patent applies local quality by selecting different reference samples based on the specific planar mode type (horizontal, vertical, or diagonal). Instead of using a uniform reference sample selection method, the patent adapts the reference sample choice to the local characteristics of each prediction mode, thereby improving prediction accuracy while maintaining manageable encoding complexity through context-based adaptation.
Solution Approach 2:
The patent implements dynamics by making the reference sample selection adaptive rather than static. The encoder dynamically determines which reference samples to use based on the planar mode type and neighboring block characteristics. This dynamic adaptation allows the system to optimize prediction accuracy for each specific case while the decoder can efficiently reconstruct using the same adaptive logic.
2Measurement precision
If more reference samples are used to improve prediction accuracy, then prediction performance improves, but the bit requirement for encoding increases
Solution Approach 1:
The patent segments the reference sample selection process by dividing it into distinct cases based on planar mode type (horizontal, vertical, diagonal). Each segment uses a specific reference sample selection strategy optimized for that mode. This segmentation allows the system to use multiple reference samples when needed without requiring the encoder to signal all possibilities, as the decoder can infer the appropriate selection based on the mode type.
Solution Approach 2:
The patent changes parameters by adapting the reference sample selection based on planar mode characteristics rather than using a fixed approach. The system modifies which reference samples are used according to the specific prediction mode and neighboring block patterns, achieving higher accuracy without increasing bit requirement because the parameter changes are driven by mode classification rather than explicit signaling.
3Productivity
If existing intra-prediction methods are used, then encoding efficiency is maintained, but residual signals remain large due to low prediction accuracy
Solution Approach 1:
The patent implements feedback by using neighboring block intra-prediction modes to inform the current block's reference sample selection. The encoder analyzes the prediction modes of adjacent blocks and uses this information to adaptively select reference samples for the current block. This feedback mechanism improves prediction accuracy by leveraging spatial correlations, thereby reducing residual signals while maintaining encoding efficiency through context-based adaptation rather than exhaustive search.
Data Source
AI summary
Disclosed are an intra prediction mode based image processing method, and an apparatus therefor. In particular, a method for processing an image on the basis of an intra prediction mode comprises the steps of: inputting, in most probable mode (MPM) candidates, intra prediction modes of a plurality of neighboring blocks which neighbor a current block, and then deriving the MPM candidates by inputting, in the MPM candidates, intra prediction modes within substitute modes consisting of a plurality of intra prediction modes, to which priorities are given, according to the priorities until the total number of MPM candidates is satisfied; deriving, from the MPM candidates, an intra prediction mode applied to the current block using an MPM index indicating a specific intra prediction mode; and generating a prediction block of the current block by applying the derived intra prediction mode.