Biometric Template Matching with Dynamic Image Ranking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing fingerprint matching processes that capture multiple images during a single sensor activation are computationally extensive, requiring more time and power, and often result in high latency and error rates due to the need for sequential matching of each image against pre-stored templates.
Innovation Solution
A biometric object matching method that dynamically ranks and processes multiple images captured during a single sensor activation, using a combination of quality metrics to select the highest ranked image for initial matching, and subsequent non-highest ranked images for further matching attempts, with adaptive threshold adjustments to optimize computational efficiency and reduce error rates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple images are captured during a single sensor activation for fingerprint matching, then the reliability of biometric recognition is improved, but the computational overhead and processing time increase significantly
Solution Approach 1:
The patent segments the template matching process into multiple stages: first matching the highest-ranked image against all templates, then conditionally matching additional images only against templates that showed partial matches in previous stages. This segmentation reduces the total number of comparisons needed while maintaining recognition reliability through progressive filtering.
Solution Approach 2:
The system performs preliminary ranking of captured images based on quality metrics before matching. By pre-ranking images and identifying the highest-quality candidate first, the system can prioritize processing resources on the most promising match, reducing overall computational overhead while maintaining high recognition accuracy.
2Reliability
If sequential matching is performed for each captured image against all templates, then comprehensive biometric verification is achieved, but the latency and power consumption increase
Solution Approach 1:
The patent implements partial matching by comparing images only against a subset of templates rather than all templates in every case. After the first image is matched against all templates, subsequent images are compared only against templates that showed partial matches previously, performing fewer than the maximum possible comparisons while maintaining verification reliability.
Solution Approach 2:
The matching process is made dynamic by adaptively adjusting the number of templates to compare based on results from previous matching attempts. The system dynamically determines whether to continue matching additional images against remaining templates based on whether a sufficient match has already been found, optimizing processing time for each specific case.
3Measurement precision
If all captured images are compared against all pre-stored templates, then the matching accuracy is maximized, but the power consumption and processing time become excessive
Solution Approach 1:
The patent extracts and utilizes matching scores from previous comparison attempts to inform subsequent matching decisions. By taking out and reusing this intermediate information, the system avoids redundant full comparisons, reducing power consumption while maintaining matching accuracy through informed selective processing.
Solution Approach 2:
The system changes the parameter of template comparison scope based on matching results. Initially all templates are compared, but as matching progresses and partial matches are identified, the parameter changes to limit subsequent comparisons to only those relevant templates, reducing overall computational energy requirements while preserving accuracy for critical comparisons.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A biometric object matching method is disclosed. The method comprises capturing (210) a plurality of images of a single biometric object during a single activation of a biometric sensor and ranking (220) the images based on a ranking metric value of each of the images to determine a highest ranked image and at least one non-highest ranked image. The method also comprises comparing (230), in a first matching attempt for the biometric object, the highest ranked image to templates of a plurality of pre-stored biometric object templates, wherein a matching metric is determined for each compared template, and generating (270) a match indicating signal when it is determined (240) that at least one of the matching metrics exceeds amatching threshold value. Whenit is determined that none of the matching metrics exceed the matching threshold value,the method comprisescomparing (260), in a further matching attempt, one of the non- highest ranked images to templates of the plurality of templates based on the matching metrics. Corresponding arrangement, server node, systemand computer program product are also disclosed.