Infrared Image Stabilization via Edge Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image stabilization methods for infrared (IR) images are computationally expensive and ineffective due to the high dynamic range and noise levels of IR data, which differ significantly from visual light images, leading to suboptimal stabilization results.
Innovation Solution
The method involves generating edge information representations of IR images, reducing them to lower information content, determining displacements between frames based on these representations, and using a feedback control loop to optimize threshold values for improved stabilization, thereby achieving image stabilization at a low computational cost.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If visual light image stabilization methods are applied to IR images, then image stabilization can be achieved, but the computational cost becomes excessively high due to the 16-bit dynamic range of IR data
Solution Approach 1:
The patent segments the IR image data processing by separating edge detection from full image processing. Only edge information is extracted and processed for stabilization calculations, while the remaining image data is processed separately. This segmentation reduces the amount of data requiring computationally intensive operations from the entire 16-bit image to only the edge portions.
Solution Approach 2:
The patent extracts only the essential edge information from the full IR image data, discarding redundant information. By taking out only the edge components that are necessary for determining displacement and stabilization, the method avoids processing the complete 16-bit dynamic range data, thereby reducing computational complexity while maintaining stabilization effectiveness.
2Measurement precision
If complex preprocessing methods are used for IR image stabilization, then stabilization accuracy improves, but computational burden increases significantly
Solution Approach 1:
The patent extracts only the necessary edge information from IR images for stabilization processing, eliminating the need for complex preprocessing of the entire image. This extraction approach maintains stabilization accuracy by focusing on the most relevant features (edges) while avoiding unnecessary computational operations on the full 16-bit image data.
Solution Approach 2:
The patent applies different processing quality levels to different parts of the image: high-quality edge detection and processing is applied only to edge regions where it is most needed for accurate displacement measurement, while the rest of the image receives minimal processing. This local quality approach maintains overall stabilization accuracy without requiring complex preprocessing of the entire image.
3Loss of information
If full 16-bit IR image data is processed for stabilization, then comprehensive information is utilized, but processing time and computational resources increase
Solution Approach 1:
The patent extracts only the essential edge information from the full 16-bit IR image data, utilizing the most critical portions of the information while discarding redundant data. This extraction enables processing to focus on the most informative elements (edges that define object boundaries and movements) without the time penalty of processing the complete image dataset.
Solution Approach 2:
The patent segments the image processing task to handle only edge portions separately from the rest of the image data. This segmentation allows the system to process edge information quickly for stabilization calculations while the remaining image data can be processed separately or with less urgency, thereby reducing overall processing time without losing critical stabilization information.
Data Source
AI summary
A method and systems of stabilizing a sequence of infrared (IR) images captured using an infrared (IR) imaging system, includes: generating edge information representations of selected captured IR images in said sequence; for each generated edge information representation, generating a second representation having a reduced amount of information compared to the edge information representation; determining displacements between captured IR images in relation to previous IR images in said sequence based on a comparison between said second representations; generating a stabilized sequence of IR images by shifting said captured IR images based on said determined displacements.