Hybrid Tone Mapping for High-Resolution Images Without Halos
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing tone mapping algorithms for high resolution images face challenges in processing time and memory constraints, leading to artifacts such as stains in smooth regions, halos around edges, and detail loss when up-sampled and filtered gain maps or profile gain table maps are used.
Innovation Solution
A hybrid approach combining a global lookup table with a local gain map is applied, where the global LUT is derived through histogram matching and regularization, and the local gain map is up-sampled and denoised to minimize artifacts, allowing for enhanced detail and contrast without halos or stains.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If tone mapping is applied to low resolution images to meet processing time and memory constraints, then processing speed and memory usage are improved, but image quality deteriorates with artifacts such as stains, halos, and detail loss
Solution Approach 1:
The tone mapping operation is segmented into two distinct components: a global lookup table (LUT) that captures overall tone mapping characteristics, and a local gain map that captures spatially varying adjustments. This segmentation allows the computationally intensive global LUT to be pre-computed at low resolution while the local gain map refines the results at high resolution, resolving the contradiction between processing speed and image quality.
Solution Approach 2:
The global lookup table is pre-computed by performing tone mapping on low resolution images before the final high resolution processing. This preliminary action establishes the foundational tone mapping characteristics efficiently, and the local gain map then refines these pre-computed results, achieving both speed and quality.
2Productivity
If up-sampled and filtered gain maps are used for tone mapping, then processing efficiency is improved, but artifacts such as stains in smooth regions and halos around edges are introduced
Solution Approach 1:
The local gain map component applies different adjustment characteristics to different regions of the image based on their content. By computing and applying local adjustments rather than using a uniform up-sampled gain map, the method preserves smooth regions without stains and maintains edge sharpness without halos, while still benefiting from the efficiency of the global LUT approach.
Solution Approach 2:
Instead of directly copying the noisy low resolution gain map to high resolution (which would introduce artifacts), the method creates a cleaned-up local gain map by comparing the low resolution tone mapped image with the high resolution source image. This copying approach with correction eliminates artifacts while maintaining processing efficiency.
3Manufacturing precision
If contrast enhancement is applied to improve image quality, then detail and contrast are enhanced, but artifacts such as halos and stains are excessively enhanced
Solution Approach 1:
The method extracts the tone mapping operation from the local detail processing by separating it into a global LUT component. This allows contrast enhancement to be applied through the global LUT without the harmful side effect of enhancing local artifacts, as the local gain map only handles residual adjustments after the global tone mapping is applied.
Data Source
AI summary
To employ low resolution, noisy tone mapping operations for high resolution images, at least one raw image frame is converted to a first image at a higher resolution and a second image at a lower resolution. Tone mapping is applied to the second image to derive a third image at the lower resolution. Histogram matching and regularization are performed to determine a lookup table approximating histogram matching of the second image to the third image. A global gain map is derived based on luma for the first image and the lookup table. A local gain map is derived by up-sampling and denoising residual differences between the third image and the lookup table applied to the second image. Based on the global gain map and the local gain map, a total gain map is determined for tone mapping the first image to produce a fourth image at the first resolution.