Bilateral Iris Imaging Head Tilt Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing iris recognition systems face inefficiencies due to head tilt and iris rotation issues, requiring extensive processing to account for unknown rotations, especially in large databases, and lack effective methods to remove head tilt and utilize interpupillary distance (IPD) information for improved identification and spoof detection.
Innovation Solution
An apparatus and method that captures images of both eyes to determine head tilt and IPD, using a virtual line and processor to rotate iris images to zero tilt, and utilizes IPD-to-iris ratio for enhanced identification and spoof detection, reducing comparative search times and improving processing efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single eye image is used for iris recognition, then the device complexity is reduced, but head tilt causes unknown rotation angles that require extensive processing to account for
Solution Approach 1:
The patent combines images from both eyes into a single unified image processing stream. By merging the left and right eye images and processing them together, the system can determine head tilt angle from the relationship between the two eyes and then apply correction to both eyes simultaneously, reducing overall processing time compared to handling each eye separately with multiple rotation attempts
2Reliability
If multiple rotations are performed to account for head tilt, then identification accuracy is improved, but processing time increases significantly
Solution Approach 1:
The patent performs preliminary action by determining the head tilt angle from the relationship between left and right eye images before the actual iris recognition process. This preliminary determination of the rotation angle allows the system to correct for head tilt in advance, eliminating the need for subsequent multiple rotation attempts during identification and significantly reducing processing time while maintaining accuracy
3Productivity
If head tilt is not removed, then processing steps are minimized, but spoof detection accuracy and eyelid segmentation speed are reduced
Solution Approach 1:
The patent applies preliminary action by removing head tilt from the eye images before proceeding to spoof detection and eyelid segmentation. By first determining the head tilt angle from the bilateral eye relationship and correcting the images accordingly, the system prepares the images in advance for subsequent processing steps, thereby improving both the speed and accuracy of spoof detection and eyelid segmentation without requiring additional separate processing steps
Data Source
AI summary
The apparatus represents a device having one or two sensors for capturing a single image or two images having the subject's eyes, and processor(s) in a housing with the one or two sensors and/or in a computer system which receives the single image or two images. Such processor(s) determine a head tilt angle between a virtual line extending between the two eyes of the subject in accordance with a predefined features associated with the eyes and a dimension characterizing zero head tilt in the single image or two images, segment left and right iris images, and rotate the segmented left and right iris images in accordance with the angle to substantially remove head tilt when present. The apparatus may also determine head tilt using predefined features associated with a single eye in the image. The resulting iris image(s) are utilized for enrollment or identification.