Neural Local Tone Mapping for Luminance Compensation Gain Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Digital images captured under varying lighting conditions often exhibit disparities in luminance and contrast, leading to underdefined, blurry, or washed-out portions that negatively impact visual fidelity and perception.
Innovation Solution
A neural-network-generated luminance compensation gain map is used to enhance images by applying a trained neural network that encodes and decodes lighting characteristics, reducing luminance differences through a residual block and postprocessing techniques to improve visual fidelity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional image processing methods are used to correct luminance disparities, then computational resources and processing time increase significantly, but image quality improvement is limited
Solution Approach 1:
The neural network is trained in advance on a large dataset of images with varying lighting conditions to learn optimal luminance compensation patterns. During actual image processing, the pre-trained network directly applies learned transformations without requiring iterative calculations, thereby achieving high precision luminance compensation while maintaining fast processing speed.
2Manufacturing precision
If conventional tone mapping algorithms are applied to reduce luminance differences, then processing complexity increases, but visual fidelity improvement is insufficient
Solution Approach 1:
The patent replaces traditional mechanical/mathematical tone mapping algorithms with a data-driven neural network model. The neural network learns complex luminance compensation patterns from training data and applies them through learned transformations, achieving superior visual fidelity while simplifying the processing pipeline compared to conventional iterative algorithms.
3Manufacturing precision
If region-specific processing is applied to correct local luminance variations, then processing time increases, but image quality improvement is achieved
Solution Approach 1:
The neural network processes the image by dividing it into regions or patches, applying localized transformations to each region based on its specific luminance characteristics. This segmentation approach enables region-specific luminance compensation to be performed efficiently in parallel, achieving local contrast enhancement without proportionally increasing overall processing time.
Data Source
AI summary
To generate a compensation map used to enhance a captured image, an accelerator unit (AU) is configured to implement a trained neural network. This trained neural network is configured to encode one or more lighting characteristics from the capture image and decode these lighting characteristics into a compensation map. The AU uses the compensation map generated by the trained neural network to modify at least a portion of the captured image to produce an enhanced image. Then, the AU applies one or more additional postprocessing techniques to further improve quality of the enhanced image prior to rendering the enhanced image on a display.