Video Coding Device Prediction Mode Selection Energy Calculation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In VVC, when the width or height of a TU exceeds 32, the high frequency transform coefficients are eliminated, leading to unnecessary processing and increased load when selecting the optimal prediction mode, even if the image quality is not degraded.
Innovation Solution
A video coding device that includes transform means, quantization means, arithmetic coding means, local decoding means, prediction mode selection means, original energy calculation means, and modified energy calculation means. The device eliminates specified transform coefficients when the TU size exceeds 32 and calculates a second energy value. It decides not to consider a prediction mode as optimal without performing quantization, arithmetic coding, and local decoding if the second energy value is significantly smaller than the first energy value.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If transform coefficients in high frequency area are eliminated when TU size exceeds 32, then image quality is maintained without degradation, but processing load increases due to unnecessary processing in prediction mode selection
Solution Approach 1:
The patent calculates the energy of transform coefficients before performing quantization, arithmetic coding, and local decoding during prediction mode selection. When the TU size exceeds 32, it eliminates high frequency transform coefficients in advance and uses only low frequency coefficients for energy calculation. This preliminary elimination avoids unnecessary processing while maintaining image quality, as the eliminated high frequency coefficients would not significantly contribute to the final image quality anyway.
2Reliability
If all prediction modes are evaluated with complete processing (quantization, arithmetic coding, local decoding), then optimal prediction mode is accurately selected, but processing time increases
Solution Approach 1:
The patent extracts and eliminates the high frequency transform coefficients when TU size exceeds 32, keeping only the low frequency coefficients for subsequent processing. This extraction reduces the amount of data that needs to be processed during prediction mode evaluation, thereby reducing processing time while maintaining reliability because the low frequency coefficients contain the majority of the image information.
Solution Approach 2:
Instead of performing complete processing (quantization, arithmetic coding, local decoding) for all prediction modes, the patent applies partial action by eliminating high frequency coefficients and using only low frequency coefficients for energy calculation. This partial processing is sufficient to determine the optimal prediction mode while significantly reducing processing time, as the eliminated high frequency components would not significantly affect the selection accuracy.
Data Source
AI summary
The video coding device calculates a first value representing energy of the prediction error signals of the block to be processed, eliminates specified transform coefficients when at least one of a width and a height of the block to be processed exceeds a predetermined value and calculates a second value representing energy of the transform coefficients other than the eliminated transform coefficients, and decides not to make the candidate of prediction mode the optimal prediction mode, without executing quantizing, arithmetic-coding and local decoding, in a case where the second value is smaller than a predetermined degree as compared to the first value.