Chroma Subsampling Using Luminance Contrast Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Chroma subsampling in high dynamic range (HDR) images and videos leads to artifacts like color bleeding due to mixing of chroma values from dark and bright regions, especially at high contrast edges, which existing methods fail to adequately address without causing de-saturation issues.
Innovation Solution
An apparatus and method that detect high contrast areas in the luminance channel to partition the image into regions, using a first chroma subsampler with higher edge-preserving properties in high contrast areas and a second subsampler in low contrast areas, thereby avoiding color bleeding artifacts while preserving image saturation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If chroma subsampling is applied to HDR images, then compression efficiency is improved, but color bleeding artifacts occur at high contrast edges
Solution Approach 1:
The image is divided into multiple regions based on luminance contrast characteristics. High contrast regions (with luminance ratios exceeding a threshold) are separately identified and processed using different chroma subsampling strategies compared to low contrast regions, preventing color bleeding at edges while maintaining compression efficiency in appropriate areas
Solution Approach 2:
Different chroma subsampling methods are applied to different regions of the image based on local luminance characteristics. In high contrast regions, chroma subsampling is restricted or avoided to prevent color bleeding, while in low contrast regions, full chroma subsampling is applied to maximize compression efficiency
2Productivity
If chroma values from dark and bright regions are mixed during subsampling, then compression is improved, but image saturation is degraded
Solution Approach 1:
The image is segmented into high contrast and low contrast regions based on luminance ratios. By preventing chroma mixing across region boundaries in high contrast areas, the original chroma information and saturation are preserved where it matters most, while still achieving compression through subsampling in appropriate regions
Solution Approach 2:
The chroma subsampling operation is adapted locally to preserve saturation in high contrast regions while enabling compression in low contrast regions. This selective approach ensures that chroma values are not mixed across luminance boundaries where such mixing would cause saturation loss
3Device complexity
If a single chroma subsampling method is applied globally, then processing complexity is reduced, but edge preservation is insufficient
Solution Approach 1:
The image is automatically segmented into high contrast and low contrast regions by detecting luminance ratios in local areas. This segmentation enables the application of region-appropriate chroma subsampling methods, significantly improving edge preservation at high contrast boundaries while maintaining manageable processing complexity through automated region classification
Solution Approach 2:
The chroma subsampling approach transitions from a static global method to a dynamic region-adaptive method. The processing complexity is justified by the significant improvement in edge preservation, as the system automatically adjusts subsampling behavior based on local image characteristics without requiring manual intervention
Data Source
AI summary
A chroma subsampling having reduced artifacts is achieved by detecting high contrast areas in a luminance channel of an image to be chroma subsampled so as to partition the image into a first region composed of the high contrast areas and a second region distinct from the first region, with chroma subsampling the image in the first region using a first chroma subsampler and chroma subsampling the image in the second region using a second chroma subsampler, with the first chroma subsampler having a higher edge preserving property than compared to the second chroma subsampler. Thereby, bleeding artifacts may be avoided at least partially, while the saturation of the image may be substantially preserved.