Fingerprint Verification Blending Low Quality Images

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing fingerprint verification technologies face challenges in maintaining high accuracy when dealing with low-quality fingerprint images, which can lead to false acceptances or rejections, reducing the recognition rate and overall performance.

Innovation Solution

A processor-implemented method that involves obtaining a query fingerprint image, performing image processing to generate a processed fingerprint image, blending the query and processed images based on determined blending ratios, and performing fingerprint verification using the blended images and registered fingerprint images, with filtering techniques adapted based on the frequency characteristics of both images.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If image processing filtering is applied to low-quality fingerprint images, then noise is reduced and image quality is enhanced, but processing time and computational complexity increase

Engineering Contradiction:
Improvefingerprint verification accuracyVSAvoidimage processing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system performs image quality assessment before deciding whether to apply filtering processing. By evaluating the fingerprint image quality in advance, the system only applies computationally intensive filtering when necessary, avoiding unnecessary processing time while ensuring accuracy when needed.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system applies filtering selectively based on image quality assessment rather than universally. For low-quality images, filtering is applied to reduce noise and enhance features. For high-quality images, filtering is skipped to save processing time. This partial application of processing optimizes the balance between accuracy and speed.

Inventive Principle:
Principle #16Partial or excessive action

2Measurement precision

If blending ratio is dynamically adjusted based on image quality, then verification accuracy is optimized, but computational complexity increases

Engineering Contradiction:
Improvefingerprint matching precisionVSAvoidprocessing algorithm complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The blending ratio is dynamically adjusted based on the assessed image quality rather than using a fixed ratio. For low-quality images, the blending ratio emphasizes the processed (filtered) version to enhance features. For high-quality images, the original image is weighted more heavily to preserve natural characteristics. This dynamic adaptation optimizes matching precision across different image conditions.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system changes the blending parameter (ratio between original and processed images) based on image quality assessment. By adjusting this parameter dynamically, the system optimizes the trade-off between noise reduction and feature preservation, thereby improving verification accuracy without requiring completely different processing algorithms.

Inventive Principle:
Principle #35Parameter changes

3Reliability

If frequency characteristic filtering is applied, then noise reduction is achieved, but loss of fingerprint pattern details may occur

Engineering Contradiction:
Improvenoise reduction effectivenessVSAvoidfingerprint pattern detail
Core Design Contradiction:
ReliabilityVSLoss of information

Solution Approach 1:

The filtering process uses frequency characteristic analysis to selectively enhance or suppress specific frequency components. By adjusting filtering parameters based on the assessed image quality and detected fingerprint patterns, the system reduces noise while preserving essential ridge and valley details necessary for accurate verification.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The system uses image quality assessment and pattern detection results as feedback to adjust the filtering strength and parameters. The assessment of image quality and fingerprint pattern characteristics informs the filtering process, allowing the system to adaptively balance noise reduction with detail preservation based on the specific input image conditions.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP3937075B1Method and apparatus with fingerprint verification
Publication Date: 2024.03.20 SAMSUNG ELECTRONICS CO LTD
  • EP3937075B1 patent drawingFigure 1
  • EP3937075B1 patent drawingFigure 2
  • EP3937075B1 patent drawingFigure 3

AI summary

A method and apparatus with fingerprint verification are disclosed. A processor implemented method includes obtaining a query fingerprint image through a fingerprint sensor, obtaining a processed fingerprint image by performing image processing on the obtained query fingerprint image, obtaining a blended fingerprint image by blending the query fingerprint image and the processed fingerprint image, and performing fingerprint verification based on the blended fingerprint image and a registered fingerprint image.