Infrared Image Registration Using Camera Response Normalization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image registration systems face challenges in accurately aligning and normalizing infrared images captured during unmanned aerial system (UAS) flights due to varying environmental conditions and image capture parameters, leading to imperfect normalization and stitching artifacts.
Innovation Solution
A camera response function normalization process is implemented, which includes an algorithm that applies offset and gain corrections to captured images, using meta data for precise registration within 2% of the field of view, and employs pairwise and global optimization techniques to ensure consistent radiometric values across images, minimizing deformation and stitching errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If multiple infrared images are captured during UAS flight under varying environmental conditions, then the quantity of image data increases, but the normalization accuracy deteriorates due to temperature changes and parameter variations
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting offset and gain parameters for each image based on environmental conditions captured during flight. The system modifies camera response function parameters (offset, gain) to compensate for temperature variations and other environmental factors, enabling accurate normalization across multiple images taken under different conditions.
2Productivity
If standard image registration techniques are used without response normalization, then the processing speed is maintained, but stitching artifacts increase due to imperfect normalization
Solution Approach 1:
The patent implements preliminary action by performing camera response function normalization (applying offset and gain corrections) before the image stitching process. This preliminary normalization step ensures that images are properly aligned and radiometrically consistent before being combined, preventing stitching artifacts from occurring in the final composite image.
3Manufacturing precision
If offset and gain corrections are applied to normalize camera response, then the radiometric consistency improves, but the computational complexity increases
Solution Approach 1:
The patent manages computational complexity through parameter changes by using a streamlined offset and gain correction approach. Instead of complex multi-parameter optimization, the system applies simplified linear transformations (offset and gain) to the camera response function, achieving radiometric consistency with computationally efficient operations that can be processed in real-time or near real-time.
Data Source
AI summary
Systems and methods for registration of image pairs, including camera response function normalization are disclosed and include selecting a pair of images from a set of images, each of the pair of images having associated metadata, determining a camera response function for each of the images in the pair of images using the associated metadata, normalizing each camera response function for across the set of images, and applying the normalized camera response function to the pair of images. A deformation map is generated in a multi-scale process using registration parameters, to deform one of the pair of images to align with another of the pair of images. The image pairs may be selected by identifying image pairs having an overlap that exceeds an overlap threshold, having a sequential proximity in an image series satisfying a proximity threshold, and/or having estimated image capture attitudes within an attitude threshold.