Wiener Filtered Prediction Blocks for Image Decoding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-resolution and high-quality images require efficient compression techniques to reduce transmission and storage costs, as conventional methods result in increased data amounts due to high information bits.
Innovation Solution
The method involves generating and filtering prediction blocks using Wiener filters based on bi-predicted motion information from reference picture lists L0 and L1, minimizing the difference between prediction blocks to reduce residual signal data and derive filtering information from neighboring blocks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional image compression techniques are used for high-resolution images, then image quality is maintained, but transmission cost and storage cost increase due to increased data amount
Solution Approach 1:
The patent applies preliminary filtering actions to prediction blocks before they are used in the decoding process. By pre-filtering the prediction blocks using Wiener filters or other filtering methods, the system reduces the residual signal energy that needs to be encoded and transmitted, thereby reducing the overall data amount while maintaining image quality
Solution Approach 2:
The patent changes the parameters of the prediction blocks by applying various filtering operations (Wiener filtering, bilateral filtering, guided filtering, etc.) that modify the pixel values and statistical properties of the prediction blocks. This parameter transformation reduces the difference between prediction and actual blocks, reducing the residual data that needs to be transmitted
2Measurement precision
If multiple prediction blocks are generated from reference picture lists L0 and L1, then prediction accuracy improves, but filtering complexity and computational load increase
Solution Approach 1:
The patent applies partial filtering actions by selectively filtering only certain prediction blocks based on specific conditions (e.g., prediction mode, block size, neighboring block properties). Instead of filtering all prediction blocks uniformly, the system applies filtering only where necessary, reducing overall computational complexity while maintaining prediction accuracy where it matters most
Solution Approach 2:
The patent applies different filtering methods and parameters to different regions or types of prediction blocks. For example, different filtering strategies are applied based on whether the block is in a motion boundary region or a smooth region, or based on the specific prediction mode used. This localized approach optimizes the balance between prediction accuracy and computational complexity
3Measurement precision
If Wiener filter coefficients are derived for each prediction block, then filtering precision improves, but computational time and processing complexity increase
Solution Approach 1:
The patent performs preliminary derivation of Wiener filter coefficients based on available information (such as neighboring block statistics or simplified models) before the actual decoding process. By pre-computing or pre-estimating the filter coefficients, the system reduces the computational burden during real-time decoding while maintaining filtering precision
Solution Approach 2:
The patent uses simplified or approximate Wiener filter coefficient derivation methods that are computationally cheaper than exact methods. Instead of deriving precise coefficients for every block using full statistical analysis, the system uses approximate methods that provide sufficient filtering precision at lower computational cost, accepting a trade-off between exact precision and processing speed
Data Source
AI summary
According to the present invention, an image decoding method performed by a decoding device comprises the steps of: generating a first prediction block and a second prediction block of a current block; selecting a prediction block, to which a Wiener filter is to be applied, among the first prediction block and the second prediction block; deriving Wiener filter coefficients of the selected prediction block based on the first prediction block and the second prediction block; filtering the selected prediction block based on the derived Wiener filter coefficients; and generating a reconstructed block of the current block based on the filtered prediction block. According to the present invention, the overall coding efficiency can be improved by minimizing the difference between prediction blocks.