CLIAMS:We Claim:

1. A method for standardization of fingerprint image to be used authentication, the method comprising steps of:
canonical representation of the fingerprint image by using the contourlets transform;
decoding directional information form the fingerprint image;
decomposing the image fingerprint images by using pyramidal directional filter bank;
directional mapping of the fingerprint image by using standard zoneplate thereby obtaining standard canonical representation of the fingerprint image.

2. The method as claimed in claim 1, wherein the decomposing is a multiscale and directional decomposition.

,TagSPECI:FORM 2
THE PATENT ACT 1970
&
THE PATENTS RULES, 2003
COMPLETE SPECIFICATION
(SEE SECTION 10 AND RULE 13)
1. TITLE OF THE INVENTION:
“Method for Standardizing Fingerprint Images used in Biometric Authentication Systems”
2. APPLICANT(s):

(a) NAME: Bydesign India Pvt. Ltd.
(b) NATIONALITY: Indian Company
(c) ADDRESS: 43 Electronics City Hosur Road,
Bangalore - 560100. Karnataka.

3. PREAMBLE TO THE DESCRIPTION:
PROVISIONAL
The following specification describes the invention. COMPLETE
The following specification particularly describes the invention and the manner in which it is to be performed

Field of the invention

The present invention relates to fingerprint imaging for biometric authentication. More particularly, the present invention relates to method for standardizing fingerprint images generated from different fingerprint scanners hardware as well as those images having non-linear elastic distortions, wherein the standardized images are compatible with all existing biometric authentication systems.

Background of the invention

Fingerprints are always considered as an authenticate identification mean due to some important properties such as every human being has a unique pattern of fingerprints which does not match with other beings pattern, no change in pattern even if the time passes. Fingerprints are considered as individual’s evidence.

A fingerprint is an impression left by the friction ridges of a human finger. Fingerprint features are unique to each individual and they remain constant for entire lifetime. Also, fingerprints of two persons never show similar pattern. Further, fingerprints are considered more reliable than traditional token based (ID card) systems or passwords. Due to this fingerprint identification is considered as authentic mode for identification of the person. Also, fingerprint impression is useful in forensic identification and considered as strong evidence in criminal cases.
Nowadays, Automatic Fingerprint Identification Systems (herein after referred as “AFIS”) has become an essential component of effective personal identification. As the fingerprint features are easy to use, difficult to share and cannot be misplaced or handed over to others. The fingerprints intrinsically represent bodily identity of an individual, therefore the fingerprints are considered more reliable than traditional token based (ID card) systems or passwords. Every individual human being has unique features on the fingerprint and they remain constant for the entire lifetime of the individual. This is a reason, why use of the AFIS is growing day by day.

Performance of the AFIS at present degrades because of following two problems:
• Interoperability issue, and
• Non-linear elastic distortion.

These issues limit the large scale deployment AFIS in a distributive manner.

Interoperability Issue in AFIS is as follows:

Biometric sensor interoperability refers to the ability of a system to compensate for the variability introduced in the raw biometric data of an individual due to deployment of different sensors, here biometric data refers to digital copy of fingerprint images. It is observed that performance of most of the AFIS drops when two different sensors/devices/scanners are used for enrollment and verification. Most fingerprint matchers are therefore restricted in their ability to compare fingerprints originating from two different sensors/devices resulting in poor inter-sensor performance.

Issues relating to non-linear elastic distortions of fingerprint image:

Human finger skin is elastic in nature. Generation of fingerprint image is a process of mapping 3D ridge structure from the fingertip on the plane surface of the fingerprint sensor. Due to different elasticity and the pressure conditions some non-linear distortions are introduced in the fingerprint image. These non linear distortions are difficult to model priory due to inconsistency of elasticity due to varied pressure. Failure to handle these distortions makes AFIS frustrating for un-necessary false rejections or security may be compromised to necessitates its operation with such a high false accept rate.

Prior Related Work (Prior Art)

Interoperability issue in the AFIS was firstly discussed by A. Jain in their Case Study on “Biometric Sensor Interoperability”. A. Ross proposed a “Thin Plate Spline Model” in his case study to compensate effect of different sensors/devices. A. Jain and A. Ross compared one to one correspondence between minutiae pairs in fingerprint image of same person taken from two different fingerprint sensors/devices. These studies are set between given two sensors and hence it is sensor dependent. Every time when new hardware/device is used again new model need to be established with all existing sensors/devices.

If fingerprint image has large non-linear distortions then average inter-ridge frequency of the two images of same person may differ, which may leads to false non-matches. Also, this method does not remove distortions along the ridges. Many researchers tried to compensate non-liner distortions at matcher level. S. Chikerrur proposed graph based fingerprint representation and matching algorithm, which improves the AIFS performance compared to NIST’s BOZORTH3 fingerprint matching algorithm. This approach also used Euclidean distance between neighboring minutiae points as a measure to find minutiae matching. T.Bhavani addressed interoperability issue at a matcher level and used ratios of the relative Euclidean distances instead of direct distance. But none of these methods are effective.

Therefore, there is a need to provide a method for standardizing fingerprint images used in biometric authentication systems, which overcomes all the drawbacks of the prior art.

Objects of the present invention

Object of the present invention is to provide a method for standardization of fingerprint images.

Another object of the present invention is to provide a method for standardization of fingerprint images, which is compatible with all existing Automatic Fingerprint Identification Systems (AFIS).

Yet another object of the present invention is to provide a method for standardization of fingerprint images, which can authenticate the fingerprint image even in case of elastic distortion due to uneven pressure, applied during capturing fingerprint image, due to elasticity of the skin or tilted finger during scanning.

Summary of the invention

According to the present invention there is provided a method for standardization of fingerprint image to be used authentication. The method having a step canonical representation of the fingerprint image by using the contourlets transform. Thereafter, decoding directional information form the fingerprint image. Further, decomposing the image fingerprint images by using pyramidal directional filter bank. At last directional mapping of the fingerprint image by using standard zoneplate thereby obtaining standard canonical representation of the fingerprint image.

Brief description of the invention

The advantages and features of the present invention will become better understood with reference to the following detailed description and claims taken in conjunction with the accompanying drawings, wherein like elements are identified with like symbols, and in which:

Figure 1 shows a flow chart of a method for standardization of fingerprint images.

Figure 2 shows directional filter bank as in contourlets used for decoding;

Figure 3 shows tight frame bound with directional criticalities are used for sub-band canonical representation of the fingerprints;

Figure 4 shows directional mapping of the fingerprint image by using standard zoneplate;

Figure 5 shows standard canonical representation of the fingerprint image, and

Figure 6 shows ROC tests results.

Details description of the invention

For a thorough understanding of the present invention, reference is to be made to the following detailed description, including the appended claims, in connection with the above-described drawings. Although the present invention is described in connection with exemplary embodiments, the present invention is not intended to be limited to the specific forms set forth herein. It is understood that various omissions and substitutions of equivalents are contemplated as circumstances may suggest or render expedient, but these are intended to cover the application or implementation without departing from the spirit or scope of the claims of the present invention. Also, it is to be understood that the phraseology and terminology used herein is for the purpose of description and should not be regarded as limiting.

The term “a” and “an” herein do not denote a limitation of quantity, but rather denote the presence of at least one of the referenced item.

The present invention provides a method for standardization of fingerprint images. Further, the method provides a fingerprint image which is is compatible with all existing Automatic Fingerprint Identification Systems (AFIS). Furthermore, the method provides a fingerprint image, which can be used for authenticate the fingerprint even in case of elastic distortion due to uneven pressure, applied during capturing fingerprint image, due to elasticity of the skin or tilted finger during scanning.

Referring now to figure 1, a flow chart of a method 100 for standardization of fingerprint images in accordance with the present invention is shown.

The method starts at step 10.

Further at step 20, the canonical representation of the fingerprint image by using the contourlets transform is done.

Further, at step 30, the fingerprint image is decoded for obtaining directional information therefrom. The fingerprint images are highly oriented and therefore need neat decoding of directional information. Directional filter bank as in contourlets is used for decoding as shown in figure 2.

Specifically, Pyramidal Directional Filter Banks as proposed in “M. N. Do and M. Vetterli, The contourlet transform” is used. This is helpful to use high pass bands of the fingerprint images effectively and also does not ignore energy efficient low pass sub-bands. In the present invention, multirate identities are used, thus transforming a l-level tree structure DBF into a parallel structure 2 raised to l channel with equivalent sampling matrices preserving underlying filter structure.
The oversampling matrices take following form:

Following family gets obtained by translating the impulse responses of synthesis filter over sampling lattices,

This provides multiscale and multidirectional basis which is capable of capturing fingerprint information efficiently to bring out potential coefficients to be standardized. The sequences of nested subspaces, which are required for MRA (Multi Resolution Analysis), satisfy following invariance properties:
Shift Invariance:

Scale Invariance:

The directional information is captured using following orthonormal basis of underlying corresponding directional sub-space:

The prototype function used is a linear combination of scaling function and is as shown below:

At step 40, the figureprint image is decomposed by using pyramidal directional filter bank. Specifically, multiscale and directional decomposition of fingerprint images is achieved through pyramidal directional filter bank (PDFB) using the basis function mentioned above. Tight frame bound with directional criticalities are used for sub-band canonical representation of the fingerprint images as shown in figure 3.

Thereafter, at step 50, directional mapping of the fingerprint image is done by using standard zoneplate thereby obtaining standard canonical representation of the fingerprint image. Specifically, directional mapping of the fingerprint image is done by using standard zoneplate as shown in figure 4. Figure 5 shows standard canonical representation of the fingerprint image.

The method stops at step 60.

Experimental Evaluation
For fingerprint image test data management is presented in a table below:
Table 1
Sensor No of Classes No of session Numbers per class Inter class combinations Intra Class combinations
L1 ID 52 5 10 63,700 6630000
Cross Match 52 5 10 63,700 6630000
Secugen (Hamster II) 52 5 10 63,700 6630000

1) In the above tables, inta class combinations reflect the tests carried between the same classes for different images (captured at same or different sessions). Inter class combinations essentially suggest combinations across classes.
2) As an example: for two different classes (people) we might have image 1_1 and 1_2 for class1 and 2_1 and 2_2 for class 2. Combinations 1_1 vs 1_2 and 2_1 vs 2_2 are intra class and combinations 1_1 vs 2_1, 1_1 vs 2_2, 1_2 vs 2_1, 1_2 vs 2_2, 2_1 vs 1_1, 2_1 vs 1_2, 2_2 vs 1_1, 2_2 vs 1_2 are inter class.
3) These do not reflect combinations across different sensors/scanners/vendors. So, above combinations do not reflect interoperable combination.

For fingerprint interoperability the combinations are worked out as under:
Table 2
Sensor No of Classes No of session Numbers per class Inter class combinations Intra Class combinations
L1 ID (L1) 52 5 10 63,700 (L1 x L1)
63,700 (L1 x CM)
63,700 (L1 x SH) 6630000 (L1 x L1)
6630000 (L1 x CM)
6630000 (L1 x SH)
Cross Match (CM) 52 5 10 63,700 (CM x L1)
63,700 (CM x CM)
63,700 (CM x SH) 6630000 (CM x L1)
6630000 (CM x CM)
6630000 (CM x SH)
Secugen (Hamster II) (SH) 52 5 10 63,700 (SH x L1)
63,700 (SH x CM)
63,700 (SH x SH) 6630000 (SH x L1)
6630000 (SH x CM)
6630000 (SH x SH)
L1=db1, CM=db2, SH=db3
Methodology:
The tests were calculated using minutiae algorithm developed for fingerprint recognition system. The results were obtained without modifying any of the parameters. The tweaking of the parameters (in optimum way) still remains an area to be explored.

Results: fingerprint
? For any biometric system the system performance can be measured by looking at EER (Equal Error Rate) obtained out of ROC (Receiver Operating Characteristics).
? Initially tests were carried without interoperability module used for pre-processing the data.
? The same tests were repeated after pre-processing the data by introducing interoperability module and formulating ‘canonical’ representations for every piece of information to be processed.
? The above ROC tests results are shown in figure 6.
? Please note L1=db1, CM=db2, SH=db3, to compare results in the ROC with the data presented in Table 3.
? Red lines show comparison of the tests between db1 vs db2. Without interoperability the EER is 23.41%, and with interoperability it is 1.44%. Although this is improvement by about 20 percentile, further improvement is desired and can be obtained by tweaking parameters. Similarly other lines can be read and understood.

Therefore, the present method 100 of the present invention is provides solutions to the problem of i) Interoperability issue and ii) Non-linear elastic distortions to the large extent. Further, for the excremental results it is clear that for fingerprint images, for the first cross test, EER was brought down from 23.4% to 1.4%. Again, these tests were carried out without tweaking parameters and also without pruning any of the bad quality images.

The foregoing descriptions of specific embodiments of the present invention have been presented for purposes of illustration and description. They are not intended to be exhaustive or to limit the present invention to the precise forms disclosed, and obviously many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the present invention and its practical application, and to thereby enable others skilled in the art to best utilize the present invention and various embodiments with various modifications as are suited to the particular use contemplated. It is understood that various omissions and substitutions of equivalents are contemplated as circumstances may suggest or render expedient, but such omissions and substitutions are intended to cover the application or implementation without departing from the spirit or scope of the claims of the present invention.