Adaptive Offset Filtering for Reconstructed Image Quality
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Next-generation video contents with high spatial resolution, high frame rate, and high dimensionality require efficient processing, but existing compression techniques lead to loss of information due to lossy compression, resulting in quality degradation and increased memory and processing demands.
Innovation Solution
A method is introduced that filters peaked samples in reconstructed images by determining a category based on neighboring pixel values and applying an offset to compensate for information loss, using a device that extracts encoded image data, decodes it, determines the category, calculates an offset, and applies it to the pixel value, thereby enhancing image quality and compression efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If lossy compression is applied to reduce data size, then compression efficiency is improved, but information loss occurs resulting in quality degradation
Solution Approach 1:
The patent converts the harmful effect of quantization error (information loss) into a beneficial filtering process. By identifying peaked samples caused by quantization and applying adaptive offset compensation, the method transforms the distortion artifact into an opportunity for targeted correction, thereby improving reconstructed image quality while maintaining compression efficiency
Solution Approach 2:
The patent changes the parameter of pixel values by adding adaptive offsets to peaked samples. The offset values are derived based on the magnitude of quantization error and the category of neighboring pixels, dynamically adjusting the compensation amount to reconstruct more accurate original pixel values and reduce information loss
2Manufacturing precision
If high spatial resolution and high frame rate are implemented, then video quality is improved, but memory storage and processing power requirements increase drastically
Solution Approach 1:
The patent segments the image processing into distinct stages: encoding with quantization, decoding with reconstruction, and post-processing with adaptive filtering. By separating the compensation function into a dedicated filtering stage using decoded pixel values and neighboring samples, the method reduces the processing burden on the main decoding pipeline and enables efficient handling of high resolution and frame rate content
Solution Approach 2:
The patent uses neighboring pixel values as copies to estimate and compensate for quantization error at the current pixel location. By referencing already-decoded neighboring samples, the method creates a low-cost approximation of the original signal without requiring additional high-precision storage or complex processing
Data Source
AI summary
Disclosed in the present invention are an image processing method and an apparatus therefor. Specifically, the image processing method can comprise the steps of: extracting encoded image data from a bitstream received from an encoder; decoding the encoded image data so as to generate a reconstructed image; determining a category of a current pixel by comparing a pixel value of the current pixel in the reconstructed image with pixel values of neighboring pixels of the current pixel; generating an offset applied to the current pixel on the basis of a difference value between pixel values of pixels, determined according to the category, among the neighboring pixels and the pixel value of the current pixel; and compensating for the current pixel by adding the offset to the pixel value of the current pixel.