Biometric Matching via Distance Transform Attribute Filtering

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Biometric identification and verification systems face high latency due to the need for numerous comparisons between input images and enrolled templates, leading to user frustration, especially in fingerprint matching where multiple views of partial biometric objects require extensive processing.

Innovation Solution

A biometric input system that generates a feature map from the input image, performs a distance transform, and determines an attribute, allowing for the elimination of enrolled images from the matching process based on this attribute, reducing the number of comparisons needed by focusing on other attributes for the remaining images.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If multiple enrolled templates are compared to each input biometric image, then matching accuracy is improved, but processing time increases

Engineering Contradiction:

Improvematching accuracyVSAvoidprocessing time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system performs preliminary actions by pre-computing and storing distance transform attributes for all enrolled biometric templates during enrollment. When verification or identification is needed, these pre-computed attributes are immediately available for quick comparison with input images, eliminating the need for time-consuming full image comparisons for all templates.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent segments the biometric matching process into two distinct stages: (1) a fast rejection stage using distance transform attribute comparison that quickly eliminates non-matching templates, and (2) a detailed matching stage that performs comprehensive comparison only on the small subset of templates that survived the rejection stage. This segmentation dramatically reduces overall processing time while maintaining accuracy.

Inventive Principle:

Principle #1Segmentation

2Adaptability or versatility

If a large number of enrolled templates are stored in the database, then identification capability is improved, but the number of comparisons required increases

Engineering Contradiction:

Improveidentification capabilityVSAvoidcomparison speed

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent introduces distance transform attributes as an intermediary representation that mediates between the raw biometric images and the matching process. These attributes serve as a compact, computationally efficient proxy that captures essential geometric characteristics, enabling fast preliminary filtering without requiring comparison of the full high-dimensional image data.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Area of stationary object

If small sensors taking partial views are used, then device size is reduced, but the number of views per template increases

Engineering Contradiction:

Improvesensor sizeVSAvoidnumber of views per template

Core Design Contradiction:

Area of stationary objectVSDevice complexity

Solution Approach 1:

The patent transforms the biometric images into a different parameter space using distance transform, which converts geometric information into distance maps. This parameter transformation allows the system to effectively combine and compare partial views by operating on the transformed distance attributes rather than the original image data, reducing the complexity of handling multiple views.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS9946917B2Efficient determination of biometric attribute for fast rejection of enrolled templates and other applications

Publication Date: 2018.04.17 SYNAPTICS INC

AI summary

A biometric input system includes: a biometric sensor, configured to generate a biometric image comprising features of an input biometric object; and a processing system, configured to receive the biometric image, generate a feature map from the biometric image, perform a distance transform on the feature map, and determine an attribute of the biometric image from the distance transform. The processing system is further configured to compare the determined attribute of the biometric image with corresponding attributes of enrolled biometric images, and, based on the comparison, to eliminate enrolled biometric images from a matching process for the biometric image, wherein the matching process is performed after the elimination and comprises comparison of the biometric image to remaining enrolled biometric images based on one or more attributes of the biometric image other than the determined attribute.