Cross-Sensor Iris Matching via Frequency Domain Normalization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing ocular recognition systems fail to accurately match iris images captured using different sensors, such as close range and far range sensors, leading to ineffective identification of subjects.
Innovation Solution
The system captures and compares iris images from different sensors by normalizing and matching them based on sensor characteristics and image quality, using modulation transfer functions and a mixture of local experts method to account for differences in resolution, blur, and illumination, and fuses scores across frequency bands for accurate matching.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional iris matching processes are used, then matching accuracy is maintained when using the same sensor, but matching effectiveness deteriorates when using different sensors
Solution Approach 1:
The patent transforms iris images from different sensors into a common frequency domain representation, changing the parameter space from spatial domain to frequency domain. This allows images captured by different sensors (with different resolutions, blur levels, and illumination characteristics) to be compared on an equal footing, resolving the contradiction between maintaining matching accuracy and achieving cross-sensor adaptability
Solution Approach 2:
The patent introduces frequency band decomposition as an intermediary step between image capture and matching. By decomposing images into multiple frequency bands and processing each band separately, the system mediates the differences between sensors, allowing accurate matching across diverse sensor types while maintaining the core matching functionality
2Productivity
If images from different sensors are matched directly, then processing speed is maintained, but matching accuracy deteriorates due to sensor differences
Solution Approach 1:
The patent segments the iris image into multiple frequency bands, processing each band separately through the matching process. This segmentation allows the system to handle sensor differences in a structured manner, maintaining precision by treating each frequency component appropriately while preserving overall processing efficiency through parallelizable operations
3Measurement precision
If sensor-specific matching processes are used, then matching accuracy is maintained for that sensor type, but system versatility deteriorates
Solution Approach 1:
The patent creates a universal matching process that works across different sensor types by transforming all images into a common frequency domain representation. This universal approach maintains the precision needed for accurate matching while achieving multi-sensor versatility, as the frequency domain transformation naturally accommodates different sensor characteristics without requiring sensor-specific processing pipelines
Data Source
AI summary
Methods, devices, and systems for cross-sensor iris matching are described herein. One method includes capturing a first image of an iris using a first sensor, capturing a second image of an iris using a second sensor, and determining whether the iris in the first image matches the iris in the second image based on characteristics of the first sensor and the second sensor and image quality of the first image and the second image.