Contrast-Based Image Fusion for Low Light IR and Visible Blending
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing night vision systems that fuse infrared and visible light images struggle to maximize scene detail, especially in low light conditions, bright lights, and fog or smoke environments, as they lack effective methods to emphasize structural content and correct for noise and artifacts.
Innovation Solution
A system and method that detects which image type has more structural information and increases the weight of pixels in that type, using a fusion module to blend co-registered low light level visible images with thermal infrared images, employing contrast detection and weighting techniques to enhance image detail and situational awareness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If traditional image fusion algorithms are used to blend infrared and visible light images, then the fusion process can be completed, but the structural content information is not emphasized sufficiently and image detail is not maximized
Solution Approach 1:
The patent changes the weighting parameter dynamically based on structural content detection. By calculating structural content metrics for both infrared and visible light images and adjusting the fusion weight accordingly, the system emphasizes regions with higher structural information while maintaining computational efficiency through parameter-based control rather than complex iterative optimization.
Solution Approach 2:
The patent replaces complex mechanical image processing operations with a contrast-based weighting mechanism. Instead of using multiple iterative fusion algorithms or complex constraint optimization, the system uses a simplified contrast detection and weight adjustment approach to achieve effective structural emphasis, substituting computational complexity with a more elegant parameter-driven solution.
2Measurement precision
If contrast enhancement algorithms are applied to improve image detail, then structural information is emphasized, but noise and artifacts may be amplified
Solution Approach 1:
The patent applies local quality by detecting structural content at different spatial locations and applying different weighting factors accordingly. Regions with high structural content receive higher weights for emphasis, while regions with low structural content (likely to contain noise) receive lower weights, thereby selectively enhancing detail without amplifying noise throughout the entire image.
Solution Approach 2:
The system uses feedback from structural content detection to adjust the fusion weighting. By continuously monitoring the structural information content and adapting the weight parameters based on this feedback, the system can enhance image detail in regions where it is most needed while suppressing enhancement in regions where it would amplify noise, creating a self-regulating process.
Data Source
AI summary
A system for two color image fusion blending co-registered low light level images in the visible region of the electromagnetic spectrum with thermal infrared images maximizes the information content of the scene by detecting in which of the two image types, IR and visible, there is more structural information and increasing the weight of the pixels in the image type having the most structural information. Additionally, situational awareness is increased by categorizing image information as “scene” or “target” and colorizing the target images to highlight target features when raw IR values are above a predetermined threshold. The system utilizes Red, Green and Blue (RGB) planes to convey different information such that for targets the Red plane is used to colorize regions when raw IR exceeds the predetermined threshold. For scene images, the Green plane provides improved situational awareness due to the above weighted blend of the two image types.