Quantization Coefficient Correction for HEVC Hardware Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In High Efficiency Video Coding (HEVC), rate-distortion optimized quantization (RDOQ) processes require iterative operations to determine optimal quantization coefficients, making parallel processing difficult and hardware implementability low due to high calculation costs.
Innovation Solution
An information processing device with a quantization processing unit that divides images into blocks and a coefficient change unit that alters coefficient values in specific frequency ranges, allowing for efficient determination of quantization coefficients with reduced calculation costs by counting non-zero coefficient appearances and applying threshold-based changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If rate-distortion optimized quantization is performed using iterative operations to determine optimal quantization coefficients, then coding efficiency is improved, but calculation cost increases and hardware implementability decreases
Solution Approach 1:
The patent segments the quantization coefficient determination process into two distinct parts: (1) initial quantization coefficient calculation using standard methods, and (2) subsequent correction by setting specific frequency range coefficients to zero based on a threshold. This segmentation eliminates the need for iterative operations while maintaining coding efficiency, as the correction step can be implemented through simple conditional logic suitable for hardware.
Solution Approach 2:
The patent performs preliminary quantization to obtain initial quantization coefficients before applying the correction step. By pre-calculating the coefficients and then applying a simple threshold-based correction rule (setting high-frequency coefficients to zero when exceeding a threshold), the system avoids iterative processing during the actual optimization phase, enabling efficient hardware implementation.
2Manufacturing precision
If iterative operations are used to determine optimal quantization coefficients, then coding precision is improved, but processing time increases
Solution Approach 1:
The patent extracts only the essential correction step from the full iterative optimization process. Instead of performing complete iterative rate-distortion optimization, the method extracts the key insight that high-frequency coefficients exceeding a threshold can be set to zero, applying this extracted rule directly to achieve precision improvement without the time-consuming iterative loop.
Solution Approach 2:
The patent applies partial action by implementing only the necessary correction step (setting specific coefficients to zero) rather than performing complete iterative optimization. This partial approach achieves sufficient quantization precision for practical purposes while dramatically reducing processing time compared to full iterative methods.
Data Source
AI summary
A scaling unit (30) (quantization processing unit) quantizes a coding target image divided into blocks. A coefficient determination unit (47) (coefficient change unit) changes a coefficient value in a specific frequency range among quantization coefficients corresponding to respective sub-blocks calculated by the scaling unit (30) by quantization. Thus, the image coding device (10) determines a final position of a coefficient just by counting the number of appearances of non-zero coefficients, and thus corrects the quantization coefficients without correcting coding distortion by iterative operation.