Fingerprint Recognition Using Local Texture Descriptors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current fingerprint recognition methods face computational inefficiencies due to the limited number of minutiae, leading to high complexity in matching algorithms and memory requirements, especially when dealing with small active areas or insufficient feature overlap in fingerprint images.
Innovation Solution
The method involves extracting fingerprint descriptors from a larger area surrounding detected features, using nonlinear transformations to capture sufficient information, allowing for the discarding of duplicated descriptors and reducing complexity by matching a single descriptor from a larger area to previously enrolled templates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the number of features is increased beyond minutiae to improve matching reliability, then the biometric score reliability is improved, but the optimization algorithm becomes computationally expensive
Solution Approach 1:
The fingerprint image is divided into multiple local regions, each centered around a detected feature point. This segmentation allows the system to process and describe local fingerprint patterns independently, reducing the overall computational complexity while maintaining reliability through multiple localized descriptors
Solution Approach 2:
The patent changes the parameter of feature description from traditional minutiae-based global descriptors to local descriptors based on fingerprint texture patterns in surrounding areas. This parameter change enables more reliable matching through texture information while avoiding computationally expensive optimization algorithms by using direct local pattern analysis
2Device complexity
If traditional minutiae-based methods are used, then the algorithm complexity is low, but the number of available features is limited leading to insufficient matching information
Solution Approach 1:
The patent extracts additional information from the fingerprint image by analyzing local texture patterns surrounding each feature point. This extraction of local texture information provides significantly more features and matching information compared to traditional minutiae-only approaches, while maintaining relatively low algorithmic complexity through direct pattern matching
Data Source
AI summary
The present disclosure relates to methods and devices for fingerprint recognition. In an aspect, a method of a fingerprint sensing system of extracting at least one fingerprint descriptor from an image captured by a fingerprint sensor for enrolment in the fingerprint sensing system is provided. The method comprises capturing images of a finger contacting the fingerprint sensor, detecting at least one fingerprint feature in each captured image, extracting fingerprint data from an area surrounding a location of the detected feature in each captured image, and extracting a fingerprint descriptor by performing a transform of the fingerprint data extracted from the area surrounding the location of the detected feature in each captured image, wherein a size of the area is selected such that the area comprises sufficient information to allow a duplicated descriptor to be discarded for the captured images.