Lightweight CNN Fingerprint Binarization for Resource-Constrained Devices

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional biometric fingerprint recognition methods require high computing power and memory resources, making them unsuitable for consumer devices like smartphones, where performance is restrictive.

Innovation Solution

A method using a lightweight Convolutional Neural Network (CNN) for binarizing fingerprint images, which reduces resource requirements and facilitates efficient feature extraction by generating a binary image, enabling effective classification on consumer devices.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional biometric approaches use traditional feature extraction methods, then classification accuracy is improved, but computing power and memory resources are excessively consumed

Engineering Contradiction:
Improveclassification accuracyVSAvoidcomputing power and memory resources
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent segments the fingerprint image processing into distinct levels: Level 1 extracts general pattern characteristics (ridge flow, orientation), Level 2 extracts minutiae points, and Level 3 extracts detailed features. This segmentation allows the system to process only the necessary level of detail for each authentication scenario, reducing overall computational burden while maintaining accuracy where needed.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent extracts only the essential characteristic information from fingerprint images rather than processing the entire image data. By focusing on specific features (ridge flow matrices, minutiae points, orientation maps) and discarding redundant information, the system achieves accurate classification with significantly reduced computational resources.

Inventive Principle:
Principle #2Taking out (Extraction)

2Ease of operation

If fingerprint recognition is implemented on consumer equipment like smartphones, then accessibility and ease of operation are improved, but device complexity and resource constraints increase

Engineering Contradiction:
Improveaccessibility on consumer devicesVSAvoidresource constraints on smartphones
Core Design Contradiction:
Ease of operationVSDevice complexity

Solution Approach 1:

The patent implements a dynamic feature extraction approach where the system can adapt the level of processing based on available resources and authentication requirements. The multi-level framework allows consumer devices to adjust computational depth dynamically, making fingerprint recognition accessible on resource-constrained smartphones while maintaining flexibility for different use cases.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system performs self-optimization by automatically selecting appropriate processing levels and feature extraction depths based on the specific authentication context and device capabilities. This eliminates the need for manual configuration and allows consumer devices to efficiently handle fingerprint recognition without requiring specialized hardware adjustments.

Inventive Principle:
Principle #25Self-service

3Measurement precision

If detailed feature extraction is performed to improve classification precision, then measurement precision is improved, but processing time increases

Engineering Contradiction:
Improvefeature extraction accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary feature extraction at Level 1 (general patterns, ridge flow) before proceeding to more detailed analysis. This preliminary processing quickly filters and organizes the fingerprint data, so that when Level 2 or Level 3 extraction is needed, the system already has pre-processed information ready, significantly reducing the time required for detailed feature extraction while maintaining accuracy.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3620970B1Method for extracting characteristics of a fingerprint represented by an input image
Publication Date: 2021.10.27 IDEMIA IDENTITY & SECURITY FRANCE SAS
  • EP3620970B1 patent drawingFigure 1
  • EP3620970B1 patent drawingFigure 2
  • EP3620970B1 patent drawingFigure 3

AI summary

The present invention relates to a method for extracting desired features from a fingerprint represented by an input image, the method being characterized in that it comprises the implementation, by data processing means (21) of a client (2), of steps of: (a) Binarization of said input image by means of a convolutional neural network, CNN, so as to generate a so-called binary image; (b) Processing of said binary image so as to extract said desired features from the fingerprint represented by said input image.