Method and system for enhancing ridges of fingerprint images

Abstract

A fingerprint processing system includes an input unit, a calculation unit and an output unit. The input unit is applied to input an original fingerprint image. The calculation unit is applied to decompose the original fingerprint image to a decomposed image by singular value decomposition (SVD) and the decomposed image is transformed into a plurality of sub-band images by discrete wavelet transformation (DWT) with a template. A plurality of compensation weight coefficients of DWT are calculated to compensate the sub-band images to generate a plurality of compensated sub-band images which are rebuilt by an inverse DWT. After rebuilding the compensated sub-band images, the output unit is applied to output an enhanced fingerprint image.

Claims

1. A fingerprint processing method comprising: decomposing a fingerprint image to obtain a decomposed matrix of the fingerprint image by a singular value decomposition method, with the decomposed matrix having a M×N SVD matrix of the fingerprint image with (M≦N) which is calculated as A=UΣV.sup.T, wherein U=[u.sub.1, u.sub.2, . . . , u.sub.m] and V=[v.sub.1, v.sub.2, . . . , v.sub.n] are orthogonal matrixes, with a diagonal of Σ=[D,O] including a plurality of singular values, where D is diag (λ.sub.1, λ.sub.2, . . . , λ.sub.k), O is a zero matrix and Σ is intensity information of the fingerprint image; transforming the M×N SVD matrix of the decomposed matrix of the fingerprint image to a plurality of sub-band images by a discrete wavelet transformation method with a predetermined template to obtain a plurality of decomposed sub-band images, with calculating the M×N SVD matrix with A=[a.sub.m, n], wherein Aε{LL, HL, LH, HH} are four decomposed sub-bands, aε{ll, hl, lh, hh} are wavelet coefficients of the four decomposed sub-bands, and m=0, 1, . . . , M/2−1; n=0, 1, 2, . . . , N/2−1 are values of the wavelet coefficients; calculating a plurality of wavelet-transformed compensation coefficients according to the plurality of decomposed sub-band images; compensating the plurality of decomposed sub-band images with the plurality of wavelet-transformed compensation coefficients to obtain a plurality of decomposed and compensated sub-band images; and rebuilding the plurality of decomposed and compensated sub-band images by an inverse discrete wavelet transformation method to obtain an enhanced fingerprint image.

2. The fingerprint processing method as defined in claim 1, wherein transforming comprises transforming the decomposed matrix of the fingerprint image by a 2D discrete wavelet transformation.

3. The fingerprint processing method as defined in claim 1, wherein transforming to obtain the plurality of sub-band images is at a first level.

4. The fingerprint processing method as defined in claim 1, wherein the plurality of sub-band images includes at least one low-frequency sub-band image, at least one middle-frequency sub-band image and at least one high-frequency sub-band image.

5. The fingerprint processing method as defined in claim 1, wherein the wavelet-transformed compensation coefficients are compensation weight coefficients.

6. The fingerprint processing method as defined in claim 5, further comprising calculating the compensation weight coefficients by ratios of mean values to a maximum mean value.

7. The fingerprint processing method as defined in claim 1, wherein the wavelet-transformed compensation coefficients are positive compensation coefficients.

8. The fingerprint processing method as defined in claim 1, wherein rebuilding comprises rebuilding the plurality of decomposed and compensated sub-band images by a 2D inverse discrete wavelet transformation.

9. The fingerprint processing method as defined in claim 1, wherein the predetermined template is a Gaussian template.

10. A fingerprint processing system comprising: an input unit connecting with a fingerprint image source to input a fingerprint image therefrom; a calculation unit connecting with the input unit and decomposing the fingerprint image to obtain a decomposed matrix of the fingerprint image by a singular value decomposition method, with the decomposed matrix of the fingerprint having a M×N SVD matrix with (M≧N) which is calculated as A=UΣV.sup.T, wherein U=[u.sub.1, u.sub.2, . . . , u.sub.m] and V=[v.sub.1, v.sub.2, . . . , v.sub.n] are orthogonal matrixes, with a diagonal of Σ=[D, O] including a plurality of singular values, where D is diag (λ.sub.1, λ.sub.2, . . . , λ.sub.k), O is a zero matrix and Σ is intensity information of the fingerprint image, with the M×N SVD matrix of the decomposed matrix of the fingerprint image transformed into a plurality of sub-band images by a discrete wavelet transformation method with a predetermined template to obtain a plurality of decomposed sub-band images, with the M×N SVD matrix calculated with A=[a.sub.m,n] wherein Aε{LL, HL, LH, HH} are four decomposed sub-bands, aε{ll, hl, lh, hh} are wavelet coefficients of the four decomposed sub-bands, and m=0, 1,. . . , M/2−1; n=0, 1, 2, . . . , N/2−1 are values of the wavelet coefficients, with a plurality of wavelet-transformed compensation coefficients calculated according to the plurality of decomposed sub-band images, with the decomposed sub-band images compensated with the plurality of wavelet-transformed compensation coefficients to obtain a plurality of decomposed and compensated sub-band images, with the plurality of decomposed and compensated sub-band images rebuilt by an inverse discrete wavelet transformation method to obtain an enhanced fingerprint image; and an output unit connecting with the calculation unit for outputting the enhanced fingerprint image.

11. The fingerprint processing system as defined in claim 10, wherein the discrete wavelet transformation method is a 2D discrete wavelet transformation.

12. The fingerprint processing system as defined in claim 10, wherein the decomposed matrix of the fingerprint image is transformed into the plurality of sub-band images at a first level.

13. The fingerprint processing system as defined in claim 10, wherein the plurality of sub-band images includes at least one low-frequency sub-band image, at least one middle-frequency sub-band image and at least one high-frequency sub-band image.

14. The fingerprint processing system as defined in claim 10, wherein the wavelet-transformed compensation coefficients are compensation weight coefficients.

15. The fingerprint processing system as defined in claim 14, wherein the compensation weight coefficients are calculated by ratios of mean values to a maximum mean value.

16. The fingerprint processing system as defined in claim 10, wherein the wavelet-transformed compensation coefficients are positive compensation coefficients.

17. The fingerprint processing system as defined in claim 10, wherein the inverse discrete wavelet transformation method is a 2D inverse discrete wavelet transformation.

18. The fingerprint processing system as defined in claim 10, wherein the predetermined template is a Gaussian template.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) The present invention will become more fully understood from the detailed description given hereinbelow and the accompanying drawings which are given by way of illustration only, and thus are not limitative of the present invention, and wherein:

(2) FIG. 1 is a block diagram of a fingerprint processing method for enhancing ridges of fingerprint images in accordance with a preferred embodiment of the present invention.

(3) FIG. 2 is a block diagram of a fingerprint processing system for enhancing ridges of fingerprint images in accordance with a preferred embodiment of the present invention.

(4) FIG. 3(a) is an original fingerprint image prior to processing with the fingerprint processing method in accordance with the preferred embodiment of the present invention.

(5) FIG. 3(b) is a series of four sub-band images decomposed and wavelet-transformed by the fingerprint processing method in accordance with the preferred embodiment of the present invention.

(6) FIG. 3(c) is a series of four compensated fingerprint images processed by the fingerprint processing method in accordance with the preferred embodiment of the present invention.

(7) FIG. 3(d) is a rebuilt fingerprint image processed by the fingerprint processing method in accordance with the preferred embodiment of the present invention.

(8) FIGS. 4(a) and 4(b) are schematic diagrams of ridge distribution data of the original fingerprint image and the enhanced fingerprint image processed by the fingerprint processing method in accordance with the preferred embodiment of the present invention.

(9) FIG. 5 is a set (a) of five original fingerprint images prior to processing with the fingerprint processing method in accordance with the preferred embodiment of the present invention; a set (b) of five enhanced fingerprint images processed by a method of Histogram Equalization; a set (c) of five enhanced fingerprint images processed by a fingerprint image contrast enhancement method proposed by Bennet & Perumal; and a set (d) of five enhanced fingerprint images processed by the fingerprint processing method in accordance with the preferred embodiment of the present invention.

DETAILED DESCRIPTION OF THE INVENTION

(10) It is noted that a fingerprint processing method and system for enhancing ridges of fingerprint images in accordance with the preferred embodiment of the present invention can be applicable to various fingerprint recognition devices and related equipment, for example: including various computer-related systems, various security systems, fingerprint collection, classification or recognition systems for crime scene investigation and other fingerprint-related system (e.g. a constellation-predicting system using fingerprint and blood types disclosed in U.S. Pat. No. 8,520,910), which are not limitative of the present invention.

(11) FIG. 1 shows a block diagram of a fingerprint processing method in accordance with a preferred embodiment of the present invention and FIG. 2 shows a block diagram of a fingerprint processing system in accordance with a preferred embodiment of the present invention suitable for the method shown in FIG. 1. Referring to FIGS. 1 and 2, the fingerprint processing system mainly includes an input unit (e.g. computer-readable data input unit) 1, a calculation unit (e.g. data-processing unit) 2 and an output unit (e.g. computer-readable data output unit) 3 which are suitably connected and arranged in a fingerprint image processing device or the like.

(12) Referring to FIG. 2, the input unit 1 selectively connects with a fingerprint image source (e.g. fingerprint database or fingerprint sensing unit) for retrieving an original fingerprint image therefrom. The calculation unit 2 further connects between the input unit 1 and the output unit 3 and is operated to process the original fingerprint image with a mixture of singular value decomposition and discrete wavelet transformation for generating a fingerprint enhanced image. The output unit 3 also selectively connects with a fingerprint storage unit (e.g. fingerprint database) or a fingerprint recognition unit for supplying the fingerprint enhanced image thereto.

(13) Referring again to FIGS. 1 and 2, the fingerprint processing method of the preferred embodiment of the present invention includes step S1: decomposing the original fingerprint image by a singular value decomposition (SVD) method to obtain a decomposed matrix of the fingerprint image and thus transforming the decomposed matrix of the fingerprint image into a plurality of sub-band images by a discrete wavelet transformation (DWT) method with a predetermined template to obtain a plurality of decomposed sub-band images. In a preferred embodiment, the predetermined template is a Gaussian template and the discrete wavelet transformation method is a 2D discrete wavelet transformation method.

(14) By way of example, FIG. 3(a) shows an original fingerprint image prior to processing with the fingerprint processing method in accordance with the preferred embodiment of the present invention. FIG. 3(b) shows a series of four sub-band images generated from the original fingerprint image (shown in FIG. 3(a)) which is decomposed and wavelet-transformed by the fingerprint processing method in accordance with the preferred embodiment of the present invention.

(15) The fingerprint processing method of the present invention utilizes the SVD method to decompose the fingerprint image A to obtain a decomposed component image which has a M×N SVD matrix A with (M≧N),
A=UΣV.sup.T,

(16) wherein U=[u.sub.1, u.sub.2, . . . , u.sub.m] and V=[v.sub.1, v.sub.2, . . . , v.sub.n] are orthogonal matrixes, a diagonal of Σ=[D, O] includes a plurality of singular values, where D is diag (λ.sub.1, λ.sub.2, . . . , λ.sub.k) having a non-increasing sequence with the singular values of λ.sub.i, i=1, . . . , k, O is a zero matrix (null matrix) of N×(M−N) and Σ is intensity information of the fingerprint image.

(17) Next, the M×N SVD matrix of the fingerprint image f is further wavelet-transformed by the 2D DWT method with the Gaussian template Ga. In the first level, four sub-band images of the fingerprint image shall be obtained and LL, HL, LH and HH represent four 2D sub-band matrices. The M×N SVD matrix A is further decomposed to A=[a.sub.m, n],

(18) wherein Aε{LL, HL, LH, HH} are four decomposed sub-bands,

(19) aε{ll, hl, lh, hh} are wavelet coefficients of sub-bands, and

(20) m=0, 1, . . . , M/2−1; n=0, 1, 2, . . . , N/2−1 are values of wavelet coefficients.

(21) Referring back to FIGS. 1 and 2, the fingerprint processing method of the preferred embodiment of the present invention includes step S2: calculating a plurality of wavelet-transformed compensation coefficients according to the decomposed sub-band images. In a preferred embodiment, the wavelet-transformed compensation coefficients are selected from positive compensation coefficients which are performed as compensation weight coefficients to thereby calculate pre-compensation mean values. A reference value is selected from a maximum mean value of the wavelet-transformed compensation coefficient of the decomposed sub-band image while the mean values of the wavelet-transformed compensation coefficients of the three remained sub-band images are adaptively calculated by ratios of mean values to the maximum mean value. The mean of all of the positive compensation coefficients of each sub-band are calculated by

(22) ${\mu }_A=\frac{4}{\mathrm{MN}}\underset{m=0}{\overset{M/z-1}{.Math.}}\underset{n=0}{\overset{N/z-1}{.Math.}}{a}_{m,n},\mathrm{where}$$a_{m,n}\ge 0,$

(23) wherein a maximum value of μ.sub.A is calculated by

(24) ${\mu }_{\max }=\underset{A\in \left\{\mathrm{LL},\mathrm{HL},\mathrm{LH},\mathrm{HH}\right\}}{\max }{\mu }_A.$

(25) The Gaussian template is calculated by Ga.sub.A=U.sub.Ga.sub._.sub.AΣ.sub.Ga.sub._.sub.AV.sub.Ga.sub.A.sup.T.

(26) In order to enhance the fingerprint image, each singular value of the sub-band images must be reviewed since the singular values contain plural foreground and intensity information. Generally, the low-frequency sub-band contains most of the dominant information, the middle-frequency sub-bands contain most of the ridge information and the high-frequency sub-band may contain noise The SVD of each matrix of the sub-band coefficients of the fingerprint image with the Gaussian template is calculated by A=U.sub.AΣ.sub.AV.sub.A.sup.T.

(27) FIG. 3(c) shows a series of four compensated fingerprint images generated from the four sub-band images (shown in FIG. 3(b)) which are compensated by the fingerprint processing method in accordance with the preferred embodiment of the present invention. Referring to FIGS. 1, 2 and 3(c), the fingerprint processing method of the preferred embodiment of the present invention includes step S3: compensating the decomposed sub-band images with the wavelet-transformed compensation coefficients to obtain a plurality of decomposed and compensated sub-band images, as best shown in FIG. 3(c). In a preferred embodiment, each of the wavelet-transformed compensation weight coefficients for the decomposed sub-band images is calculated by

(28) ${\xi }_{\mathrm{LL}}=\frac{{\mu }_{\max }}{{\mu }_{\mathrm{LL}}}*\frac{\max \left({\Sigma }_{\mathrm{Ga}\_\mathrm{LL}}\right)}{\max \left({\Sigma }_{\mathrm{LL}}\right)}$${\xi }_{\mathrm{HL}}=\sqrt[6]{\frac{{\mu }_{\max }}{{\mu }_{\mathrm{HL}}}*\frac{\max \left({\Sigma }_{\mathrm{Ga}\_\mathrm{HL}}\right)}{\max \left({\Sigma }_{\mathrm{HL}}\right)}}$${\xi }_{\mathrm{LH}}=\sqrt[6]{\frac{{\mu }_{\max }}{{\mu }_{\mathrm{LH}}}*\frac{\max \left({\Sigma }_{\mathrm{Ga}\_\mathrm{LH}}\right)}{\max \left({\Sigma }_{\mathrm{LH}}\right)}}$${\xi }_{\mathrm{HH}}=\sqrt[8]{\frac{{\mu }_{\max }}{{\mu }_{\mathrm{HH}}}*\frac{\max \left({\Sigma }_{\mathrm{Ga}\_\mathrm{HH}}\right)}{\max \left({\Sigma }_{\mathrm{HH}}\right)}}.$

(29) Accordingly, each matrix of the sub-band coefficients of the fingerprint image multiplied by corresponding wavelet-transformed compensation weight coefficients is calculated as A=U.sub.A(ξ.sub.A*Σ.sub.A)V.sub.A.sup.T.

(30) FIG. 3(d) shows a rebuilt fingerprint image generated from the four compensated fingerprint images (shown in FIG. 3(c)) which are recombined by the fingerprint processing method in accordance with the preferred embodiment of the present invention. Referring to FIGS. 1, 2 and 3(d), the fingerprint processing method of the preferred embodiment of the present invention includes step S4: rebuilding the decomposed and compensated sub-band images by an inverse discrete wavelet transformation method to obtain an enhanced fingerprint image, as best shown in FIG. 3(d).

(31) FIGS. 4(a) and 4(b) show schematic diagrams of ridge distribution data (i.e. number of pixels in relation to gray levels) of the original fingerprint image, as shown in FIG. 3(a), and the enhanced fingerprint image, as shown in FIG. 3(d), processed by the fingerprint processing method in accordance with the preferred embodiment of the present invention. Referring now to FIGS. 3(a) and 4(a), the original fingerprint image has two peaks of the ridge distribution data distributed at opposite sides of the diagram. As best shown in FIG. 4(a), the ridge distribution data of the original fingerprint image include information of background and blur regions of the fingerprint image.

(32) Turning now to FIGS. 3(d) and 4(b), the enhanced fingerprint image generated from the original fingerprint image, as shown in FIG. 3(a), processed by the fingerprint processing method of the present invention has a high degree of enhancement. As best shown in FIG. 4(b), the ridge distribution data of the enhanced fingerprint image is a normal distribution concentrated with a single peak and information of background and blur regions of the fingerprint image are removed.

(33) FIG. 5(a) shows a first set of five original fingerprint images prior to processing with the fingerprint processing method in accordance with the preferred embodiment of the present invention. Referring to FIG. 5(a), first to fourth original fingerprint images are retrieved from the database systems of FVC2002 DB1-A, FVC2002 DB2-A, FVC2002 DB3-A and FVC2002 DB4-A, respectively and the fifth original fingerprint image is retrieved from the database system NIST-4.

(34) FIG. 5(b) shows a second set of five enhanced fingerprint images generated from the original fingerprint images, as shown in FIG. 5(a), which are processed by a method of Histogram Equalization. FIG. 5(c) shows a third set of five enhanced fingerprint images generated from the original fingerprint images, as shown in FIG. 5(a), which are processed by a fingerprint image contrast enhancement method proposed by Bennet & Perumal. FIG. 5(d) shows a fourth set of five enhanced fingerprint images generated from the original fingerprint images, as shown in FIG. 5(a), which are processed by the fingerprint processing method in accordance with the preferred embodiment of the present invention. Referring to FIGS. 5(b), 5(c) and 5(d), the fourth set of five enhanced fingerprint images, as best shown in FIG. 5(d), processed by the fingerprint processing method of the present invention has a quality of contrast much better than those of the second and third sets of the enhanced fingerprint images, as best shown in FIGS. 5(b) and 5(c), processed by the fingerprint processing method of Histogram Equalization and Bennet & Perumal.

(35) Although the invention has been described in detail with reference to its presently preferred embodiment, it will be understood by one of ordinary skill in the art that various modifications can be made without departing from the spirit and the scope of the invention, as set forth in the appended claims.