Saturation Processing for HDR to SDR Dynamic Range Conversion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies lack a simple and effective saturation specification strategy for handling significant dynamic range conversions in high dynamic range (HDR) image processing, particularly when converting between HDR and standard dynamic range (SDR) images, leading to issues like banding and clipping artifacts.
Innovation Solution
A saturation specification method using a saturation modification function defined by a set of gains for color difference values, applied through a color-dependent transformation that adjusts saturation based on the V-Y index, allowing for flexible and efficient conversion between different dynamic ranges.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional saturation processing is used for HDR to SDR conversion, then color transformation can be performed, but banding and clipping artifacts occur
Solution Approach 1:
The patent applies parameter changes by modifying the saturation gain as a function of luminance value. Instead of using a fixed saturation gain, the patent dynamically adjusts the gain based on the luminance of each pixel, using different gain values for different luminance ranges. This prevents over-saturation in bright regions that causes clipping artifacts while maintaining color accuracy in mid-tone and shadow regions.
Solution Approach 2:
The patent implements dynamics by making the saturation processing adaptive rather than static. The saturation gain varies dynamically according to the luminance characteristics of the image content, allowing the processing to respond to different regions of the image. This dynamic adjustment prevents uniform over-saturation that leads to banding artifacts while preserving color fidelity where needed.
2Manufacturing precision
If complex saturation processing is applied to maintain color accuracy, then color fidelity improves, but computational complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the luminance range into multiple segments or bins, where each segment is assigned a specific saturation gain value. This segmentation approach simplifies the computational complexity by avoiding continuous gain calculation while still achieving adaptive saturation control. The segmented approach maintains color fidelity through region-specific processing without requiring complex real-time calculations.
Solution Approach 2:
The patent changes the saturation gain parameter based on luminance segmentation, using discrete gain values for different luminance ranges. This parameter change strategy simplifies computation by replacing complex continuous functions with discrete lookup tables or simple conditional logic, reducing computational complexity while maintaining color fidelity through luminance-aware saturation control.
3Productivity
If fixed saturation gain is used for conversion, then processing speed is high, but color accuracy deteriorates in different luminance regions
Solution Approach 1:
The patent changes the saturation gain parameter from a fixed value to a luminance-dependent value. By adjusting the gain based on the luminance of each pixel or region, the patent achieves accurate color transformation across different luminance ranges. This parameter change maintains processing efficiency by using simple luminance-based conditional logic or lookup tables rather than complex calculations.
Solution Approach 2:
The patent applies local quality by using different saturation gain values for different luminance regions of the image. Instead of applying a uniform gain across the entire image, the patent tailors the saturation processing to local luminance characteristics, ensuring color accuracy in each region while maintaining overall processing efficiency through the systematic approach.
4Device complexity
If simple saturation processing is applied, then computational complexity is low, but artifact generation increases
Solution Approach 1:
The patent changes the saturation gain from a constant to a luminance-varying parameter, which prevents artifacts by adapting the processing to local image characteristics. This simple parameter change based on luminance values effectively prevents over-saturation artifacts in bright regions and under-saturation in shadow regions without requiring complex processing algorithms.
Solution Approach 2:
The patent implements dynamics by making the saturation processing adaptive to luminance conditions. This dynamic adjustment prevents artifacts by responding to the actual image content rather than applying static processing. The dynamic nature of the solution keeps computational complexity low by using simple luminance-based control logic rather than complex artifact detection and correction algorithms.
Data Source
AI summary
For serving the new different needs of high dynamic range image handling and conversion of images to a different dynamic range in decoding strategies the inventor invented image encoder (201) for encoding at least one high dynamic range image (HDR_ORIG) for at least one time instant, and at least one image (Im_LDR_out) of a second dynamic range which is different from the one of the high dynamic range image by at least a factor two for the same time instant, which is arranged to encode a matrix of pixel colors for at least one of those two images, comprising a specification unit (212) which comprises a color saturation modification apparatus (202) as claimed in claim 1, and a saturation specification unit (204) arranged to specify a saturation modification specification function specifying a color-dependent saturation transformation, as a set of gains (g) for each possible value of a maximal one of at least two color difference values (R′-Y′, B′-Y′) of a pixel color (Y′,Cb,Cr), which color difference values are defined as the value of a non-linear RGB additive color component of the pixel color minus the luma of the pixel color, and the image encoder further comprising a formatter (210) arranged to output in an image signal (S_im) an encoding of an image (Im_in_DCT) comprising the matrix of pixel colors, and as metadata the set of gains (g).


