Image Enhancement via Illumination Map Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image enhancement methods using network models trained with pairs of original and annotated images are inefficient and have low robustness, particularly in handling underexposure and backlight issues.
Innovation Solution
The method involves obtaining an original image, performing synthesis processing to create a low-resolution illumination map, establishing a mapping relationship to achieve an original-resolution illumination map, and using this map for image enhancement, thereby improving efficiency and robustness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If network models are trained using pairs of original images and annotated images, then image enhancement capability is achieved, but training efficiency is low
Solution Approach 1:
The patent extracts the illumination map from the image enhancement process, separating it as an independent intermediate representation. Instead of directly mapping original images to enhanced images, the method extracts illumination information first, then uses it to guide the enhancement process. This extraction principle resolves the contradiction by creating a more efficient training pathway that maintains enhancement capability while reducing training complexity and time.
2Measurement precision
If high-resolution illumination maps are generated directly from original images, then image enhancement accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent segments the illumination map generation process into two distinct stages: first generating a low-resolution illumination map, then upscaling it to high resolution. This segmentation principle resolves the contradiction by dividing the computationally intensive task into manageable parts, maintaining accuracy through the two-stage approach while significantly reducing overall computational complexity compared to direct high-resolution generation.
Solution Approach 2:
The patent performs preliminary action by generating the low-resolution illumination map first before upsampling to high resolution. This preliminary generation of illumination information at lower resolution reduces the immediate computational burden, while the subsequent upscaling step ensures high-resolution accuracy is achieved without the full computational cost of direct high-resolution processing.
3Manufacturing precision
If image enhancement is performed on underexposed images, then image quality is improved, but risk of over-exposure increases
Solution Approach 1:
The patent employs feedback mechanisms through the illumination map that provides guidance on where enhancement is needed and to what extent. The illumination map acts as a feedback signal that modulates the enhancement process, allowing the system to improve image quality in underexposed regions while preventing over-exposure by using the illumination information to control the enhancement intensity spatially.
Data Source
AI summary
Embodiments of this disclosure include an image enhancement method and apparatus. The image enhancement may include obtaining an original image and performing synthesis processing on features of the original image to obtain a first illumination map corresponding to the original image. A resolution of the first illumination map may be lower than a resolution of the original image. The image enhancement may further include obtaining, based on the first illumination map, a mapping relationship between an image to an illumination map and performing mapping processing on the original image based on the mapping relationship to obtain a second illumination map. A resolution of the second illumination map may be equal to the resolution of the original image. The image enhancement may further include performing image enhancement processing on the original image according to the second illumination map to obtain a target image.