Automatic Location and Extraction of Palmprint Contour from Grayscale Image Student Yu Yue Tutor Shi Guangshun.

Automatic Location and Extractionof Palmprint Contourfrom Grayscale Image

Student Yu Yue Tutor Shi Guangshun

Contents

• Introduction• Contour Extraction

Model• Model Realization• Experimental Results

Introduction: Biometrics

Why we need Biometrics?

Introduction: Biometrics

Knowledge-basedToken-based

• Traditional Personal Identification Method

When you Forget the password

OR

Lost the card……

Introduction: Biometrics

Biometrics

based on one’s physiological or behavioral characteristics

• Unlosableness

• Universality

• Uniqueness

• Permanence

Introduction: Biometrics

Biometric Technologies

FingerprintFaceHand GeometryPalmprintIrisVoice

Introduction: Palmprint

Palmprint

• Large Quantity of Information

• Permanence & Uniqueness

• Low Cost

• Multi-Biometrics

Both in Identification and Verification

Introduction: Palmprint

Sampling

FeatureExtraction

FeatureMatch

Identification / verification

Preparation

Application

Recognition

• Location• Contour Extraction• … …

Introduction: Palmprint Contour

• Locate the Palm Area

• A Feature in Match

The First Step of Palmprint Recognition.

Importance:

Contents

• Introduction• Contour Extraction

Model• Model Realization• Experimental Results

Fill Palm Area

Select Connected Component

Extract Contour

Median Filter

Thresholding

Preprocessing

ContourExtraction

Optimazation

Contour Extraction Model

Contour Extraction Model: Median Filter

Median Filter• Erase the Noise• Keep the Edge• Felt Ridges

Palmprint (part)

Filter

Contour Extraction Model: Thresholding

Thresholding

the key point:

choose the Threshold Value

Suppose the palm area is nearly fixed.

Contour Extraction Model: Select CC

Fill Palm Area &

Select Connected Component

the key point:

the Arithimetic of

Detecting Connected Component

Contour Extraction Model: Select CC

Step 1:mark the outer back

Step 2:fill the palm area

Step 3:choose a CC

Contour Extraction Model: Extract Contour

Extract Contour

Object

Filter

Contents

• Introduction• Contour Extraction

Model• Model Realization• Experimental Results

Model Realization

Model Realization

(Demo)

Contents

• Introduction• Contour Extraction

Model• Model Realization• Experimental Results

Experimental Results

Total Smples 40

Processing Results AmountPercenta

geExact Palm Contour 31 77.5%

Palm Area Expanded 8 20%

Palm Area Lost 1 2.5%

Experimental Results

Experimental Results

Conclusion

An effective model when the image is legible enough.

Thanks

Thanks !

Thanks all my friends,Thank you for your concern and hel

p.

YuYueMay. 2004