FJFX Minutiae Quality

The FJFX Minutiae Quality module evaluates the quality of detected minutiae points using the FingerJetFX minutiae extraction algorithm. It computes quality metrics based on local image characteristics at minutiae locations, including contrast (Mu) and orientation certainty (OCL).

Overview

Minutiae quality assessment is critical for fingerprint matching accuracy. This module analyzes the local image quality surrounding each detected minutia point to determine its reliability. High-quality minutiae are more likely to be consistently detected and matched correctly.

The module integrates with the FingerJetFX feature extraction library and computes quality measures based on both contrast characteristics and orientation certainty at minutiae positions.

Class Definition

Header: quality_modules/FJFXMinutiaeQuality.h

class FJFXMinutiaeQuality : public Algorithm {
public:
    struct MinutiaData {
        int x;          ///< x-coordinate from top-left corner
        int y;          ///< y-coordinate from top-left corner
        double quality; ///< computed minutiae quality value
    };
    
    FJFXMinutiaeQuality(
        const NFIQ2::FingerprintImageData &fingerprintImage,
        const std::vector<FingerJetFX::Minutia> &minutiaData);
    
    virtual ~FJFXMinutiaeQuality();
    
    std::string getName() const override;
    static std::vector<std::string> getNativeQualityMeasureIDs();
    std::vector<FingerJetFX::Minutia> getMinutiaData() const;

private:
    std::unordered_map<std::string, double> computeFeatureData(
        const NFIQ2::FingerprintImageData &fingerprintImage);
    
    std::vector<FingerJetFX::Minutia> minutiaData_ {};
    
    std::vector<MinutiaData> computeMuMinQuality(
        int bs, const NFIQ2::FingerprintImageData &fingerprintImage);
    
    std::vector<MinutiaData> computeOCLMinQuality(
        int bs, const NFIQ2::FingerprintImageData &fingerprintImage);
    
    double computeMMBBasedOnCOM(
        int bs,
        const NFIQ2::FingerprintImageData &fingerprintImage,
        unsigned int regionSize);
};

Ideal Quality Constants:

#define IDEALSTDEV 64   // Ideal standard deviation for local regions
#define IDEALMEAN 127   // Ideal mean intensity for local regions

Constructor

FJFXMinutiaeQuality()

FJFXMinutiaeQuality(
    const NFIQ2::FingerprintImageData &fingerprintImage,
    const std::vector<FingerJetFX::Minutia> &minutiaData)

Creates a FJFX Minutiae Quality module instance with pre-detected minutiae. Parameters:

fingerprintImage - Input fingerprint image data
minutiaData - Vector of detected minutiae from FingerJetFX

Methods

getName()

std::string getName() const override

Returns the module identifier: "MinutiaeQuality" Returns: Module name as string

getNativeQualityMeasureIDs()

static std::vector<std::string> getNativeQualityMeasureIDs()

Returns the list of quality measure identifiers produced by this module. Returns: Vector containing:

FJFXPos_Mu_MinutiaeQuality_2 - Percent of minutiae with moderate contrast quality
FJFXPos_OCL_MinutiaeQuality_80 - Percent of minutiae with high orientation certainty

getMinutiaData()

std::vector<FingerJetFX::Minutia> getMinutiaData() const

Returns the minutiae data used for quality computation. Returns: Vector of FingerJetFX minutiae Throws:

NFIQ2::Exception - If template could not be extracted

Quality Features

The FJFX Minutiae Quality module generates 2 quality measures:

FJFXPos_Mu_MinutiaeQuality_2

Description: Percentage of minutiae with contrast quality in the moderate-high range. Algorithm:

For each minutia, extract the local region around its position
Compute contrast-based quality metric (difference from ideal mean/stddev)
Classify quality into 4 ranges:
- Range 0: quality ≤ -0.5 (very poor)
- Range 1: -0.5 < quality ≤ 0.0 (poor)
- Range 2: 0.0 < quality ≤ 0.5 (moderate-good)
- Range 3: quality > 0.5 (excellent)
Return the percentage of minutiae in range 2

Interpretation:

High values (> 0.5): Most minutiae are in well-contrasted regions
Low values (< 0.2): Many minutiae are in poor contrast areas

FJFXPos_OCL_MinutiaeQuality_80

Description: Percentage of minutiae with orientation certainty level above 80%. Algorithm:

For each minutia, extract the local block at its position
Compute OCL (Orientation Certainty Level) for that block
Convert OCL to percentage scale (0-100)
Count minutiae with OCL > 80
Return as percentage of total minutiae

Interpretation:

High values (> 0.6): Most minutiae are in regions with clear ridge orientation
Low values (< 0.3): Many minutiae are in uncertain orientation areas

Both quality measures range from 0.0 to 1.0 (representing 0% to 100%). Higher values indicate better overall minutiae quality.

Supporting Data Structures

MinutiaData

struct MinutiaData {
    int x;          ///< x-coordinate from top-left corner
    int y;          ///< y-coordinate from top-left corner  
    double quality; ///< computed minutiae quality value
};

Internal structure for storing minutia position and its computed quality score.

Private Methods

computeMuMinQuality()

std::vector<MinutiaData> computeMuMinQuality(
    int bs,
    const NFIQ2::FingerprintImageData &fingerprintImage)

Computes contrast-based quality for each minutia. Parameters:

bs - Block size for local region analysis
fingerprintImage - Input fingerprint image

Returns: Vector of MinutiaData with quality scores based on contrast

computeOCLMinQuality()

std::vector<MinutiaData> computeOCLMinQuality(
    int bs,
    const NFIQ2::FingerprintImageData &fingerprintImage)

Computes orientation certainty-based quality for each minutia. Parameters:

bs - Block size for local region analysis
fingerprintImage - Input fingerprint image

Returns: Vector of MinutiaData with quality scores based on OCL

computeMMBBasedOnCOM()

double computeMMBBasedOnCOM(
    int bs,
    const NFIQ2::FingerprintImageData &fingerprintImage,
    unsigned int regionSize)

Computes Mean of Block Means based on the center of mass (COM) of minutiae distribution. Parameters:

bs - Block size
fingerprintImage - Input fingerprint image
regionSize - Size of region around COM

Returns: MMB value for the central minutiae region

Usage Example

#include <quality_modules/FJFXMinutiaeQuality.h>
#include <quality_modules/FingerJetFX.h>

NFIQ2::FingerprintImageData fpImage = /* load image */;

try {
    // First, extract minutiae using FingerJetFX
    std::vector<NFIQ2::QualityMeasures::FingerJetFX::Minutia> minutiae;
    // ... extract minutiae ...
    
    // Compute minutiae quality
    NFIQ2::QualityMeasures::FJFXMinutiaeQuality fjfxQuality(fpImage, minutiae);
    
    // Get quality features
    auto features = fjfxQuality.getFeatures();
    double muQuality = features["FJFXPos_Mu_MinutiaeQuality_2"];
    double oclQuality = features["FJFXPos_OCL_MinutiaeQuality_80"];
    
    // Assess overall minutiae quality
    if (muQuality < 0.3 || oclQuality < 0.4) {
        // Poor minutiae quality - many minutiae in low-quality regions
    }
    
    // Get individual minutiae data
    auto minutiaeData = fjfxQuality.getMinutiaData();
    std::cout << "Total minutiae detected: " << minutiaeData.size() << std::endl;
    
} catch (const NFIQ2::Exception &e) {
    // Handle error
}

Integration Example

// Complete workflow with minutiae extraction and quality assessment

#include <quality_modules/FingerJetFX.h>
#include <quality_modules/FJFXMinutiaeQuality.h>

NFIQ2::FingerprintImageData fpImage = /* load image */;

try {
    // Step 1: Extract minutiae
    NFIQ2::QualityMeasures::FingerJetFX fjfx(fpImage);
    std::vector<NFIQ2::QualityMeasures::FingerJetFX::Minutia> minutiae = 
        fjfx.getMinutiae();
    
    if (minutiae.empty()) {
        throw NFIQ2::Exception(
            NFIQ2::ErrorCode::NoMinutiaeDetected,
            "No minutiae detected in fingerprint");
    }
    
    // Step 2: Assess minutiae quality
    NFIQ2::QualityMeasures::FJFXMinutiaeQuality quality(fpImage, minutiae);
    auto features = quality.getFeatures();
    
    // Step 3: Filter minutiae based on quality thresholds
    double muQualityThreshold = 0.3;
    double oclQualityThreshold = 0.5;
    
    if (features["FJFXPos_Mu_MinutiaeQuality_2"] >= muQualityThreshold &&
        features["FJFXPos_OCL_MinutiaeQuality_80"] >= oclQualityThreshold) {
        // Sufficient high-quality minutiae for reliable matching
    } else {
        // Quality issues detected
    }
    
} catch (const NFIQ2::Exception &e) {
    std::cerr << "Error: " << e.what() << std::endl;
}

Algorithm Details

Contrast Quality (Mu) Assessment

Local region extraction: For each minutia at position (x, y), extract a block of size bs × bs
Compute local statistics:
- Local mean intensity
- Local standard deviation
Compare to ideal values:
- Ideal mean: 127 (balanced ridge/valley)
- Ideal standard deviation: 64 (good contrast)
Quality score: Based on deviation from ideal values

Orientation Certainty (OCL) Assessment

Block extraction: Extract block at minutia position
OCL computation: Use OCLHistogram::getOCLValueOfBlock()
Threshold application: Check if OCL > 0.80
Aggregation: Count percentage exceeding threshold

Relationship to NFIQ 2.0

The FJFX Minutiae Quality features are integrated into the overall NFIQ 2.0 score:

Minutiae count and quality are strong predictors of matching performance
Spatial distribution affects the discriminative power of the fingerprint
Local quality at minutiae impacts feature extraction reliability

FingerJetFX - Minutiae extraction engine
OCL Histogram - Orientation certainty computation
Mu - Contrast measurement
Quality Module Base - Base class for all quality modules

Core Classes

Quality Measures

Quality Modules

Supporting Types

FJFX Minutiae Quality

Overview

Class Definition

Constructor

FJFXMinutiaeQuality()

Methods

getName()

getNativeQualityMeasureIDs()

getMinutiaData()

Quality Features

FJFXPos_Mu_MinutiaeQuality_2

FJFXPos_OCL_MinutiaeQuality_80

Supporting Data Structures

MinutiaData

Private Methods

computeMuMinQuality()

computeOCLMinQuality()

computeMMBBasedOnCOM()

Usage Example

Integration Example

Algorithm Details

Contrast Quality (Mu) Assessment

Orientation Certainty (OCL) Assessment

Relationship to NFIQ 2.0

Build docs developers (and LLMs) love

Core Classes

Quality Measures

Quality Modules

Supporting Types

Documentation Index

​Overview

​Class Definition

​Constructor

​FJFXMinutiaeQuality()

​Methods

​getName()

​getNativeQualityMeasureIDs()

​getMinutiaData()

​Quality Features

​FJFXPos_Mu_MinutiaeQuality_2

​FJFXPos_OCL_MinutiaeQuality_80

​Supporting Data Structures

​MinutiaData

​Private Methods

​computeMuMinQuality()

​computeOCLMinQuality()

​computeMMBBasedOnCOM()

​Usage Example

​Integration Example

​Algorithm Details

​Contrast Quality (Mu) Assessment

​Orientation Certainty (OCL) Assessment

​Relationship to NFIQ 2.0

​Related Modules

Build docs developers (and LLMs) love

Overview

Class Definition

Constructor

FJFXMinutiaeQuality()

Methods

getName()

getNativeQualityMeasureIDs()

getMinutiaData()

Quality Features

FJFXPos_Mu_MinutiaeQuality_2

FJFXPos_OCL_MinutiaeQuality_80

Supporting Data Structures

MinutiaData

Private Methods

computeMuMinQuality()

computeOCLMinQuality()

computeMMBBasedOnCOM()

Usage Example

Integration Example

Algorithm Details

Contrast Quality (Mu) Assessment

Orientation Certainty (OCL) Assessment

Relationship to NFIQ 2.0

Related Modules