Super-Resolution Thermal Imaging via Motion-Based Image Registration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing cameras for non-visible spectral ranges, such as thermal imaging cameras, face challenges in improving resolution without increasing production costs, as increasing pixel numbers is complex and costly.
Innovation Solution
A method that records and combines non-visible image data streams during or after random camera movements to generate super-resolution images with higher output resolution than the recorded images, using energy functions and point spread functions to optimize image processing and registration, allowing for improved resolution without the need for additional hardware.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the number of pixels in the detector apparatus is increased to improve resolution, then the obtainable resolution is improved, but the device complexity and production costs increase
Solution Approach 1:
The patent uses multiple low-resolution images (copies of the same scene) captured during random movements to reconstruct a high-resolution image. Instead of using a single high-resolution detector, the system creates multiple copies at lower resolution and combines them through image processing, thereby achieving high resolution without increasing detector complexity
Solution Approach 2:
The patent transitions from a single static image to a sequence of images captured during camera movement. By adding the temporal dimension and utilizing random movements in space, the system reconstructs resolution information that would otherwise require a much more complex detector apparatus
2Measurement precision
If the number of pixels in the detector apparatus is increased to improve resolution, then the obtainable resolution is improved, but the production costs increase
Solution Approach 1:
The system creates multiple copies of the scene at lower resolution through repeated imaging during random movements, then combines these copies to achieve high resolution. This approach is more cost-effective than manufacturing a single high-resolution detector, as it uses standard components and computational resources instead of expensive high pixel-count hardware
Solution Approach 2:
By utilizing temporal sequences and spatial movements rather than increasing spatial resolution directly, the patent avoids the high production costs associated with high pixel-count detectors. The solution leverages time and motion dimensions to achieve resolution improvement at lower hardware cost
3Measurement precision
If multiple images are combined to generate super-resolution images, then the output image resolution is improved, but the computational processing time increases
Solution Approach 1:
The system captures multiple images during random camera movements, creating a periodic sequence of low-resolution images. By processing these images in a structured sequence rather than attempting to process a single high-resolution image, the computational burden is distributed across multiple simpler processing steps, reducing overall time loss
Data Source
AI summary
In a thermal imaging camera (1), an infrared image data stream (5) of infrared images (4) is captured during a random movement of the thermal imaging camera (1), and the infrared images (4) are combined into a higher-resolution infrared image (9).