Inverse Tone Mapping via Sublayer Segmentation and Category-Specific Transformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for converting low-contrast-ratio images to high-contrast-ratio images fail to effectively restore detailed information, as they apply uniform expansion techniques that do not account for varying image characteristics, resulting in loss of contrast components.
Innovation Solution
The method involves separating low-contrast-ratio images into sublayer images, determining image categories, and learning transformation matrices to apply targeted conversions, allowing for the generation of high-contrast-ratio images by categorizing and processing each patch separately using learned matrices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If uniform dynamic range expansion is applied to the entire image, then the processing simplicity is maintained, but the detailed information restoration capability deteriorates
Solution Approach 1:
The image is divided into multiple local regions or patches, and each region is processed independently using appropriate transformation parameters. This segmentation allows the system to apply different processing strategies to different image areas, thereby restoring detailed information effectively without requiring complex global processing.
Solution Approach 2:
Different transformation parameters are applied to different local regions of the image based on their specific characteristics. This local quality approach ensures that each region receives the most suitable processing, improving detailed information restoration while maintaining reasonable processing complexity through parameter adaptation rather than structural complexity.
2Loss of information
If the image is divided into sublayer images with different transformation parameters, then the detailed information restoration capability is improved, but the device complexity increases
Solution Approach 1:
The image is segmented into manageable patches or regions that can be processed independently. This segmentation reduces the overall complexity by breaking down the complex task of entire-image processing into simpler, repeated operations on smaller units, while still achieving improved detailed information restoration through localized parameter adaptation.
Solution Approach 2:
The system changes transformation parameters (such as scaling factors, offset values, or filter coefficients) based on local image characteristics rather than using fixed parameters throughout. This parameter adaptation improves restoration capability while maintaining relatively simple processing structures, as the complexity is managed through parameter variation rather than structural complexity.
3Loss of time
If conventional dynamic range expansion is used, then the processing time is short, but the image quality improvement is insufficient
Solution Approach 1:
The image processing is segmented into independent regional operations that can be executed in parallel. This segmentation enables the system to apply more sophisticated processing algorithms to each region without significantly increasing total processing time, as the parallel execution of simpler regional tasks compensates for the increased per-region complexity, achieving both speed and quality improvement.
Solution Approach 2:
The system applies transformation parameters that may exceed conventional processing intensity locally, particularly in regions where detailed information restoration is most needed. This partial or excessive action approach improves image quality in critical areas while maintaining acceptable processing times by concentrating enhanced processing only where necessary rather than uniformly across the entire image.
Data Source
AI summary
The present invention provides a technology that separates a low-contrast-ratio image into sublayer images, classifies each sublayer image into several categories in accordance with the characteristics of each sublayer image, and learns a transformation matrix representing a relationship between the low-contrast-ratio image and a high-contrast-ratio image for each category. In addition, the present invention provides a technology that separates an input low-contrast-ratio image into sublayer images, selects a category corresponding to each sublayer image, and applies a learned transformation matrix to generate a high.