Presentation is loading. Please wait.

Presentation is loading. Please wait.

Fingerprint Image Enhancement 程广权. Introduction Problems – Image contrast – Adverse physical factors Minimize the undesired effects Some intermediate.

Similar presentations


Presentation on theme: "Fingerprint Image Enhancement 程广权. Introduction Problems – Image contrast – Adverse physical factors Minimize the undesired effects Some intermediate."— Presentation transcript:

1 Fingerprint Image Enhancement 程广权

2 Introduction Problems – Image contrast – Adverse physical factors Minimize the undesired effects Some intermediate steps

3 Fingerprint Image Digitalization and Density Resolution: – 500 dpi ( dots per inch ) – 8bits depth ( i.e. 256 gray level ) Some feathers – Pores – Valleys – Incipient Ridges – Ridges

4 Recognition of Fingerprint Ridge

5 Enhancement result

6 Intermediate Steps in Fingerprint Image Enhancement Contrast enhancement or normalization Pore and incipient ridge removal Ridge orientation estimation Frequency estimation Foreground segmentation Ridge enhancement filtering

7 Contrast enhancement

8

9 Pore and incipient Ridge Removal Pore and incipient ridge are obstacles in frequency estimation

10 Pore and incipient Ridge Removal Methods : – Remove the minutiae originated by pores – Pores are enclosed by darker pixels(easy to confuse with some type of ridges) These two factors have not yet been fully explored, and plays an important role.

11 Ridge Orientation Estimation Estimation Methods: – Slits – Two-dimensional gradient – Two-dimensional Fourier transform Ridge orientation smoothing Suggestion: use global pattern type and prior knowledge of ridge flow

12 Frequency Estimation Frequency: the number of ridges per unit length Frequency can be rather unstable, hard to estimate.(unlike orientation) Methods: – Two-dimensional Fourier transform – Peak intervals orthogonal to the ridge orientation

13 Frequency Estimation

14 Foreground Segmentation Distinguish the fingerprint ridge region from the background Methods: – Base on the confidence of the orientation – Base on the gray-level analysis(not fit for low quality images)

15 Ridge Enhancement Filtering Methods: – Use filtering mask with fixed sizeor predetermined variable frequency – Two-dimensional Fourier transform(Gabor and wavelet filtering) Mimutia – Stable – Unstable

16 Ridge Enhancement Filtering

17 Strength of the enhancement – Law enforcement, for accuracy, not strong – Non-law enforcement, automatic, strong enhancement is OK Strong enhancement is beneficial to extract stable minutiae even from poor quality images

18 Ridge Enhancement Filtering

19 After the enhancement Fingerprint binarization Fingerprint skeletonization

20 thanks Q&A


Download ppt "Fingerprint Image Enhancement 程广权. Introduction Problems – Image contrast – Adverse physical factors Minimize the undesired effects Some intermediate."

Similar presentations


Ads by Google