Image Highlight Detection and Rendering for Display Adaptation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image display technologies struggle to render images with high dynamic range accurately across different displays, leading to inconsistent visual experiences due to limitations in dynamic range and color gamut, causing highlights to appear differently on various devices.
Innovation Solution
The method involves detecting highlights in images using algorithms like histogram-based and difference-of-Gaussians techniques, classifying them, and applying gain factors based on viewer preferences to scale luminance amplitudes, allowing for consistent rendering across different dynamic ranges and display devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If highlights are mapped to different maximum brightness levels for different displays, then the image can be adapted to each display's dynamic range, but the same image will look disparately different across displays
Solution Approach 1:
The patent applies preliminary action by pre-classifying highlights into different groups (e.g., specular highlights, emissive highlights, reflective highlights) with distinct characteristics before rendering. This classification is performed in advance so that when the image is displayed on different devices, the appropriate highlight group can be selected and rendered with consistent visual intent, rather than adapting highlights post-processing which would lead to disparate appearances.
2Device complexity
If standard dynamic range displays are used, then device complexity is reduced, but the ability to render images with large lightness variations is severely limited
Solution Approach 1:
The patent applies segmentation by dividing the image processing into distinct segments: first processing non-highlight regions with standard dynamic range techniques, then separately processing highlight regions with specialized algorithms. This allows SDR displays to render images with large lightness variations by treating highlights as a separate component that can be independently adjusted and blended, effectively extending the display's capability without requiring HDR hardware.
3Measurement precision
If highlight detection algorithms are applied, then highlight rendering accuracy is improved, but processing time and computational complexity increase
Solution Approach 1:
The patent applies partial action by implementing a two-stage highlight detection approach: first using a fast, simple threshold-based method to identify potential highlight regions, then applying more complex detection algorithms only to those identified regions. This selective application of processing intensity maintains high detection accuracy while significantly reducing overall computational time compared to applying full-strength algorithms to the entire image.
Data Source
AI summary
A highlight mask is generated for an image to identify one or more highlights in the image. One or more highlight classifiers are determined for the one or more highlights in the image. One or more highlight gains are applied with the highlight mask to luminance amplitudes of pixels in the one or more highlights in the image to generate a scaled image. The one or more high-light gains for the one or more highlights are determined based at least in part on the one or more highlight classifiers determined for the one or more highlights.