Infrared Image Noise Reduction via Column-Row Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Infrared imaging systems face challenges in reducing spatial, temporal, and 1/f type noise without relying on a shutter or uniform scene, which can lead to image blurring and performance degradation.
Innovation Solution
The system processes infrared image data by separating it into column and row noise filter portions, calculating noise offset terms, and applying corrections to reduce noise while minimizing blur, using high pass filtering and histogram analysis to determine median differences for noise correction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If low pass filters are used to reduce noise, then noise is reduced, but the image is blurred and system performance is lowered
Solution Approach 1:
The patent segments the noise reduction process into separate column and row filtering operations. By dividing the image processing into independent column-wise and row-wise operations, the system can reduce noise while preserving spatial relationships and edges, avoiding the blurring effect of conventional low pass filters.
Solution Approach 2:
The patent applies different processing characteristics to different regions of the image by calculating column and row noise offset terms independently. This allows localized noise correction that adapts to the specific noise characteristics in each column and row, reducing noise while maintaining local image quality and detail.
2Measurement precision
If a shutter is used for calibration, then calibration can be performed, but manufacturing costs increase and the system cannot capture images during calibration
Solution Approach 1:
The patent extracts the calibration function from the mechanical shutter component. By removing the shutter and performing calibration through software-based column and row noise offset term calculation using non-uniform scene images, the system eliminates the need for mechanical moving parts while maintaining calibration capability.
Solution Approach 2:
The patent replaces the mechanical shutter system with a computational approach. Instead of using physical shutter mechanisms for calibration, the system uses image processing algorithms that calculate noise offset terms from images captured during normal operation, substituting mechanical calibration with computational calibration.
3Measurement precision
If calibration is performed using a uniform scene, then noise can be corrected, but the system requires a uniform target which may not be available in all scenarios
Solution Approach 1:
The patent makes the calibration process dynamic by calculating column and row noise offset terms from actual scene images captured during normal operation. Instead of requiring a static uniform scene, the system adapts to any scene type by dynamically computing noise characteristics from the captured image data, making the calibration versatile for different应用场景.
Solution Approach 2:
The system performs self-calibration by using its own captured images to calculate noise offset terms. The infrared imaging system processes its own output images to determine column and row noise characteristics, eliminating the need for external uniform calibration targets and enabling autonomous calibration in any scene condition.
Data Source
AI summary
Various techniques are provided to process infrared images. In one implementation, a method of processing infrared image data includes receiving infrared image data associated with a scene. The infrared image data comprises a plurality of pixels arranged in a plurality of rows and columns. The method also includes selecting one of the columns. The method also includes, for each pixel of the selected column, comparing the pixel to a corresponding plurality of neighborhood pixels. The method also includes, for each comparison, adjusting a first counter if the pixel of the selected column has a value greater than the compared neighborhood pixel. The method also includes, for each comparison, adjusting a second counter if the pixel of the selected column has a value less than the compared neighborhood pixel. The method also includes selectively updating a column correction term associated with the selected column based on the first and second counters.