Image Processing Device Selective Color Appearance Model Application
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current color appearance models, such as CIECAM02, require significant computational resources due to complex transformations and numerous parameters, especially when applied to entire images, leading to processing burdens and inefficiencies, particularly in determining background luminance for each pixel.
Innovation Solution
The proposed solution involves using a combination of the CIECAM02 model for pixels within a predetermined color gamut and the von Kries adaptation transformation formula for pixels outside this gamut, reducing processing load by applying detailed and simplified color appearance models selectively based on luminance and hue conditions, and dividing images into regions for uniform transformation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the CIECAM02 color appearance model is applied to the whole image, then accurate color appearance prediction is achieved, but the processing load becomes very heavy
Solution Approach 1:
The patent divides the image into multiple regions and applies different color appearance models to different regions. Specifically, it segments pixels into those within a predetermined color gamut (requiring CIECAM02 for accurate color appearance) and those outside (using simpler models), thereby reducing overall processing load while maintaining accuracy where needed
Solution Approach 2:
The patent applies different levels of processing quality to different parts of the image based on local characteristics. Pixels within the predetermined color gamut receive the full CIECAM02 treatment for high accuracy, while pixels outside use simplified models, creating local quality variation that optimizes the balance between accuracy and processing speed
2Measurement precision
If the CIECAM02 model is used for each pixel, then accurate color transformation is achieved, but the number of calculation steps increases significantly
Solution Approach 1:
The patent applies the complex CIECAM02 calculation only partially - specifically to pixels within the predetermined color gamut - rather than to all pixels. This partial action approach maintains color transformation accuracy for critical regions while avoiding unnecessary complex calculations in regions where simpler models suffice
3Measurement precision
If background luminance is calculated for each pixel, then accurate color appearance is achieved, but the processing load increases heavily
Solution Approach 1:
The patent performs preliminary calculation of background luminance only for pixels within the predetermined color gamut that require accurate color appearance modeling. By pre-identifying which pixels need this treatment and calculating their background luminance only when necessary, it avoids the heavy processing load of calculating background luminance for every pixel in the image
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An image processing device that transforms first image data of an image captured under an image-capturing condition into second image data that reproduce a color appearance under a viewing condition, includes: an image data acquiring unit that acquires the first image data of the image captured under the image-capturing condition; a first transformation unit that transforms the acquired first image data into the second image data pixel by pixel based on the image-capturing condition and the viewing condition by a first method that takes into consideration chromatic adaptation; a second transformation unit that transforms the acquired first image data into the second image data pixel by pixel based on the image-capturing condition and the viewing condition by a second method that is simpler than the first method and takes into consideration chromatic adaptation; and a controlling unit that controls such that either the first transformation unit or the second transformation unit is selected depending on a state of a transformation target pixel or a neighborhood pixel neighboring the transformation target pixel in the first image data.