Composite Image Fusion Using Selective NIR Detail Thresholding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for enhancing low light images using near-infrared (NIR) images often result in unnatural artifacts or lack significant contrast, as they blindly fuse visible and NIR images without differentiating noise, leading to suboptimal image quality under low light conditions.
Innovation Solution
A method and electronic device that receive visible and NIR image data from a composite sensor, identify portions of the NIR image with a level of detail greater than or equal to a threshold, and generate a composite image by selectively fusing the visible and NIR image data, thereby enhancing image quality and contrast while avoiding noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If visible and NIR images are completely fused to enhance low light image quality, then image detail and contrast are improved, but noise and unnatural artifacts increase
Solution Approach 1:
The patent applies local quality by selectively fusing NIR image details only in specific regions where the NIR image quality exceeds the visible image quality. The processing unit compares image qualities in different regions and applies fusion selectively, ensuring that noise is not introduced in regions where NIR quality is insufficient while maximizing detail enhancement in regions where NIR quality is superior.
2Illumination intensity
If NIR image details are fused with visible image to improve contrast, then visibility is enhanced, but unnatural artifacts such as red eye and harsh shadows appear
Solution Approach 1:
The patent prevents unnatural artifacts by applying fusion selectively in regions where it benefits image quality. The processing unit determines which regions should receive NIR detail fusion based on quality comparison, avoiding regions where fusion would create artifacts such as red eye or harsh shadows, thus maintaining natural appearance in those areas.
Solution Approach 2:
The patent applies partial action by using only the necessary portion of NIR image details rather than fusing the entire NIR image. The fusion is applied partially in specific regions where quality improvement is needed, avoiding excessive fusion that would introduce artifacts in other regions.
3Illumination intensity
If flash is used to capture low light images, then image brightness is improved, but color temperature mismatch and harsh shadows occur
Solution Approach 1:
The patent uses NIR image details as an intermediary to enhance visible image quality without using flash. Instead of directly illuminating with flash light that causes color cast, the system transfers detail information from NIR images captured under ambient light, serving as an intermediary mechanism to improve brightness and detail while maintaining natural color temperature.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
A method and an electronic device for producing a composite image are provided. The method includes receiving visible image data and near infrared (NIR) image data from a composite sensor, determining whether at least one portion of the NIR image data having a level of detail greater than or equal to a threshold, and generating a composite image by fusing the visible image data with the at least one portion of the NIR image data based on the determination and storing the composite image in a memory.