Neural Visual Coding With Cross-Component Sample Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Neural network-based image and video coding technologies face challenges in achieving improved coding quality due to the inherent difficulty of processing cross-component information independently, leading to suboptimal reconstruction quality and efficiency in compression algorithms.
Innovation Solution
A method is proposed that adjusts samples of a first component of visual data using offsets and utilizes cross-component information by comparing adjusted first samples with a threshold to enhance the quality of reconstructed visual data, employing a neural network-based model for conversion and bitstream generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If cross-component information is processed independently in neural network-based coding, then device complexity is reduced, but coding quality deteriorates
Solution Approach 1:
The patent merges the processing of first component samples and second component samples by using adjusted samples from the first component to adjust samples of the second component. This combining approach allows cross-component information to be utilized, improving reconstruction quality without requiring completely separate processing paths, thus resolving the contradiction between processing complexity and coding quality.
2Manufacturing precision
If cross-component information is utilized to enhance reconstruction quality, then coding quality improves, but processing complexity increases
Solution Approach 1:
The patent applies preliminary action by first adjusting the first component samples with offsets before using them to adjust the second component samples. This pre-processing step organizes the cross-component information in advance, making the subsequent processing more efficient and reducing the overall processing complexity while still achieving improved reconstruction quality.
Solution Approach 2:
The patent applies local quality by selectively adjusting second component samples based on specific conditions (comparing adjusted first samples with thresholds). This targeted approach focuses computational resources only where needed, improving reconstruction quality in critical areas while avoiding unnecessary processing elsewhere, thus managing processing complexity effectively.
Data Source
AI summary
Embodiments of the present disclosure provide a solution for visual data processing. A method for visual data processing is proposed. The method comprises: obtaining, for a conversion between visual data and one or more bitstreams of the visual data with a neural network (NN)-based model, a set of adjusted first samples by adjusting a first sample of a first component of the visual data with a set of offsets, each of the set of adjusted first samples corresponding to one of the set of offsets; adjusting a second sample of a second component of the visual data based on at least one adjusted first sample, wherein the at least one adjusted first sample is determined from the set of adjusted first samples by comparing each of the set of adjusted first samples with a threshold; and performing the conversion based on the adjusted second sample.