JP2006500662A - Palmprint authentication method and apparatus - Google Patents

Abstract

【Task】
A palm print authentication method includes a step of analyzing a certain area of a palm image and acquiring texture data for the skin surface of the area. The texture data is compared with reference information in the database to determine the individual's identity. An apparatus for capturing an image of a palm includes a container having a window therein, and an image sensor and a light source disposed within the container and disposed for capturing the image through the window. A protrusion is provided on the surface. The protrusion is placed in a known juxtaposition with respect to the hand properly placed in the window for capturing an image containing the palm region of the hand.

Description

  The present invention relates to a method for analyzing palm prints for biometric authentication, in particular personal authentication. The present invention also relates to an apparatus for capturing a palmprint image for personal authentication.

  Using palmprint recognition as a method for personal authentication is a new biometric technology that replaces fingerprints. Known methods include the analysis of palm prints to identify singularities, detail points, and wrinkles in the palm print image. These known methods require high resolution images as shown in FIG. This is obtained by ink palm print. However, these are messy and real-time authentication cannot be obtained passively.

  Some companies have developed high resolution palmprint scanners and authentication systems to solve the problem of ink palmprints. However, these devices that capture high-resolution images are expensive and depend on high-performance computers to meet the requirements of real-time authentication.

  One solution to the above problem appears to be the use of low resolution images. FIG. 2 shows a low resolution image corresponding to FIG. However, singular points and detailed points cannot be easily observed in a low-resolution palm print image, and thus wrinkles that can be more easily identified must play an important role in authentication. However, as can be seen from FIG. 2, only a small part of the heel is quite clear, but it is questionable whether they provide sufficient discrimination to reliably authenticate individuals from a large population.

  The object of the present invention relates to a biometric authentication method, particularly a method for analyzing a palmprint for personal authentication that overcomes or improves the problems associated with prior art methods. It is a further object of the present invention to provide an apparatus for capturing palmprint images that overcomes or improves the shortcomings of prior art devices, or at least as an alternative useful to the public.

According to the first aspect of the invention,
Obtaining an image of an area of the skin surface from an individual;
Analyzing the image to extract texture features of the skin surface area;
A biometric authentication method is provided that includes comparing texture features with reference information in a database.

According to a second aspect of the invention,
Obtaining an image of a portion of the inner surface of a person's hand;
Obtaining a sub-image of the skin surface within a predetermined area of the inner surface of the hand;
Analyzing the sub-image to obtain texture data for the skin surface;
A biometric authentication method is provided that includes comparing the texture data with reference information in a database.

  Preferably, the predetermined area depends on one or more characteristics of the hand.

  Preferably, the one or more characteristics are areas between fingers of the hand.

Preferably, the sub-image is
Identifying at least two points representative of the area between the fingers of the hand;
Determining a coordinate system having first and second axes, the first and second axes having the two points located on the first axis and equidistant from the second axis When,
Determining the parameters of the sub-image in the coordinate system using the distance between the two points.

  Preferably, the parameters of the sub-image are (0.25D, 0.5D), (1.25D, 0.5D), (0.25D, -0.5D) and (1.25D, -0.5D). (Where D is the distance between two points) and includes a point in the coordinate system.

  Preferably there is a further step of normalizing the sub-image.

  Preferably, the step of analyzing the sub-image includes using a Gabor filter.

  Preferably, the step of analyzing the sub-image includes a step of dividing the layer of the low-resolution sub-image using Gabor analysis.

  Preferably, the sub-image is divided into two parts, a real part and an imaginary part, each part being stored as a vector.

  Preferably, the comparison between the texture data and the reference information in the database is based on the following form of Hamming distance.

_However, P _{R (Q} R) and _{P I (Q} I) is a real part and an imaginary part.

According to a third aspect of the invention,
A container having a window therein;
An image sensor disposed within the container and disposed to capture an image through the window;
A light source arranged to illuminate the window;
A palm print comprising at least one protrusion adjacent to the window, the protrusion being arranged in a known juxtaposition with respect to a hand appropriately placed on the window for capturing an image including a palm region of the hand An image capture device is provided.

According to a fourth aspect of the invention,
A container having a window therein;
An image sensor disposed within the container and disposed to capture an image through the window;
A light source arranged to illuminate the window;
A control device operable to control the image sensor and the light source to capture an image;
There is provided a palmprint image capture device comprising at least one protrusion, the protrusion disposed in a known juxtaposition with respect to a hand appropriately placed on the window for capturing an image including a palm region. .

  Preferably, the protrusion is a peg or pin disposed between two or more fingers of a hand properly placed on the window.

  Preferably, the light source is annular and has an image sensor at its center.

  Preferably, the image sensor is a charge coupled device (CCD) or a complementary metal oxide semiconductor (CMOS) sensor.

  Further aspects of the present invention will become apparent from the following description, given by way of example only.

  In the following, embodiments of the present invention will be described with reference to the accompanying drawings.

  The palmprint authentication method according to the present invention includes 1) a step of acquiring an individual's palmprint image, 2) a step of analyzing skin texture data from the image, and 3) collating the skin texture data with information stored in a database. Includes three parts of the step of: These parts are described in more detail below.

1) Acquisition of Individual Palmprint Image Referring to FIG. 3, a low resolution image of a portion of the inner surface of the hand is acquired by a known method using a CCD camera. In order to extract the authentication data from the image, a reproducible sub-image of the palm area must be identified using hand characteristics. In the preferred embodiment, the holes between the fingers are identified and used as parameters to create a coordinate system in which the parameters defining the sub-image are found. The preferred embodiment has six main steps given below.

Referring to FIG. 4, in the first step, a low-pass filter L (u, v) such as a Gaussian filter is applied to the original image O (x, y). A threshold value T _p is then used to convert the superimposed image into a binary image B (x, y).

Referring to FIG. 5, the second step is to obtain the boundary (F _i x _j , F _i y _j ) (where i = 1, 2) between the fingers using a boundary tracking algorithm. . The boundary of the hole between the ring finger and the middle finger is not extracted because it is not useful for the following processing.

Referring to FIG. 6, the third step is to calculate the tangent of the hole (F _i x _j , F _i y _j ). If (x ₁ , y ₁ ) and (x ₂ , y ₂ ) are two points on (F ₁ x _j , F ₁ y _j ) and (F ₂ x _j , F ₂ y _j ), respectively, A line passing through two points (y = mx + c) satisfies the inequality F _i y _i ≦ mF _i x _j + c for all i and j. The line (y = mx + c) is the tangent of the two holes. This line represented by the numeral 2 in FIG. 6 is the Y axis of the coordinate system that determines the position of the sub-image 1.

  The fourth step is to determine the X axis and the origin of the coordinate system by obtaining a line 3 that is perpendicular to the line 2 and passes through the midpoint of the two points. The two points are on the Y axis and equidistant from the X axis.

The fifth step is to extract the variable-size sub-image 1 based on the coordinate system. The size and position of the sub-image 1 are based on the Euclidean distance (D) between the two points (x ₁ , y ₁ ) and (x ₂ , y ₂ ). Points 4, 5, 6, and 7 representing the corners of the sub-image 1 in the coordinate system are (0.25D, 0.5D), (1.25D, 0.5D), and (0.25D, -0.5D), respectively. And (1.25D, -0.5D). Therefore, the sub-image 1 is a square that is symmetric with respect to the X-axis 3 with the distance along each side equal to the Euclidean distance. Since the sub-image is based on the characteristics of the hand (the area between the fingers), it can be reproduced for each individual hand.

FIG. 7 shows the x-axis 3 and the y-axis 2 of the coordinate system and the sub-image 1 superimposed on the unprocessed image of FIG.

  The sixth step extracts the sub-image 1 using bilinear interpolation for future extraction and normalizes it to the standard size. FIG. 8 shows the sub-image 1 extracted and normalized.

  Once the palm sub-image 1 is acquired, the next part of the method is performed.

2) Analysis of skin texture of image Circular Gabor filter is an effective tool for texture analysis and has the following general shape

  However, i = √−1, u is the sine wave frequency, θ controls the direction of the function, and σ is the standard deviation of the Gaussian envelope. Gabor filters are widely used for texture analysis, and thus skilled readers will be familiar with using them for such purposes.

  In order to make the texture analysis more robust to image brightness variations, the discrete Gabor filter G [x, y, θ, u, σ] is changed to zero DC by applying the following formula:

However, (2n + 1) ² is the size of the filter. In fact, the imaginary part of the Gabor filter is automatically zero DC due to odd symmetry. The use of an adjusted Gabor filter is to filter the preprocessed image. The phase information is then encoded by the following inequality.

However, I (x, y) is a preprocessed image, and (x ₀ , y ₀ ) is the center of filtering.

  Referring to FIGS. 9 and 10, since it is expected that some people do not place their hands correctly, some non-palm images are included in the palm sub-image. A mask is created to indicate the location of pixels that are not palm prints. Since the original image is considered a semi-closed environment, pixels that are not palmprints arise from the black border of the image background. Thus, a threshold can be used to divide pixels that are not palm prints. Typically, the feature size including mask and palmprint features is 384 bytes.

  FIG. 11 shows the preprocessed image, FIG. 12 shows the real part of the corresponding texture feature, FIG. 13 shows the imaginary part of the corresponding texture feature, and FIG. 14 shows the corresponding mask.

Useful discussions regarding the use of Gabor filters for texture analysis can be found in the following two publications. That is, A. Jain and G. Healey, "A multiscale
representation including opponent color features for texture recognition ",
IEEE Transactions on Image Processing, vol. 7, no. 1, pp. 124-128,
1998, and D. Dunn
and WE Higgins, "Optimal Gabor filters for
texture segmentation, "IEEE Transactions on
Image Processing, vol. 4, no. 4, pp. 947-964, 1995.

3) Palmprint matching Real and imaginary features are represented as vectors, which are compared to the stored palmprint data vector. Palmprint matching is based on normalized Hamming distance. For example, let P and Q be two palmprint feature matrices. The normalized Hamming distance is described as follows. That is,

_{However, P R (Q R),} the real part of P _{I (Q} I) and _{P M (Q} M), respectively P (Q), the imaginary part and the mask. If the two bits PR _(I) (i, j) are equal to QR _(I) (i, j) and only then the result of the Boolean operator "x" is equal to zero. ∩ represents an AND operator, and the size of the feature matrix is N × N. It is noted that D ₀ is between 1 and 0. For perfect matching, the matching score is zero. Due to incomplete preprocessing, the features need to be translated vertically and horizontally and rechecked. At this time, the range of vertical and horizontal translation is -2 to 2. The minimum value of D ₀ obtained by translation verification is regarded as the final verification score.

  The following experimental results illustrate the effectiveness of the system according to the present invention.

  Palmprint images were collected from 154 subjects using a palmprint scanner. Approximately 65% of the subjects were male. The age distribution of the subjects is shown in Table 1 below.

  Each subject provided two image groups. Each group includes 10 images for the left palm and 10 images for the right palm. Overall, each subject provided 40 images to create an image database containing 6191 images from 308 different palms. The average time difference between the collection of the first and second image groups from each subject was 57 days. The maximum and minimum time differences were 90 days and 4 days, respectively. After completion of the first acquisition, the light source was changed to resemble the image acquisition by two different palmprint scanners and the focus was adjusted with respect to the CCD camera. Figures 15 and 16 show corresponding hand images captured in the first and second groups for one subject. The collected images were of two sizes, 384x284 and 768x568. The larger image was reduced to 384x284. As a result, the size of all test images in the following experiment is 384 × 284 and the resolution is 75 dpi.

  In order to obtain verification accuracy of the palmprint system, each palmprint image was checked against all palmprint images in the database. We referred to collation of two palmprint images from the same palm of the same subject as correct collation. The total number of comparisons was 19,161,145. The number of correct matches was 59,176.

  The probability distributions for the principal and others are estimated by correct matching and incorrect matching, respectively, and are shown in FIG. FIG. 18 shows the corresponding receiver action curve (ROC), which is a plot of the person acceptance rate against the other person acceptance rate for all operating point candidates. From FIG. 18, it is estimated that the method according to the present invention operates at a personal acceptance rate of 96% and a foreign acceptance rate of 0.1%. The corresponding threshold is 0.35. This result is comparable to prior art palmprint methods and other hand biometric techniques including hand shape and fingerprint verification.

  The method according to the present invention uses low resolution images and is low in computational cost. It can be seen that the verification accuracy is comparable to known high performance methods using high resolution images.

  This method can be used for access management, ATM and various security systems.

19 and 20 illustrate a palmprint image capture device according to the present invention. The device includes a housing 8 having a flat top surface 9 on which a hand is placed with the palm down to capture a palmprint image. The surface 9 is opaque and has a window 15 through which an image is captured. In the preferred embodiment, the window 15 contains a glass panel. In alternative embodiments, the window 15 may include other transparent covers, lenses, or nothing (ie, an open window).

An image sensor 11 such as a charge coupled device (CCD) is mounted in the housing 8 . The lens 12 is screwed to the CCD. The opening of the lens 12 faces the window 15 in the surface 9 .

An annular light source 13 is mounted around the lens 12 to illuminate the image in the window 15 . A mounting arm 14 supports the annular light source 13 and a screw 16 is used to secure the CCD to the mounting arm 14 . A palm print image is formed through this optical plane from the lens 12 to the CCD 11 and then the digitized image data is transferred to an external processing device such as a personal computer (not shown) for processing or manipulation. .

FIG. 21 shows a plan view of the lens 12 and the light source 13 passing through the section AA ′ of FIG. The lens 12 is at the center of the annular light source 13 . The lens 12 is attached to the top of the CCD 11 .

Adjacent the window 15 on the surface 9 are a plurality of protrusions in the form of pegs 10 that are used to correctly position the hand on the surface 9 in the palm area on the window 15 . In use, a person places his hand on the surface 9 and places the peg 10 between the thumb and other fingers of the hand. This ensures that the hand is correctly placed on the device to capture the optimal area of the palm through the window 15 .

FIG. 22 shows an image of the target palm print area captured through the window 15 . As can be seen at a glance, the use of the opaque surface 9 with the target window 15 ensures that the problem area can be obtained from the palm accordingly. This image is captured by the personal computer from the CCD 11 for further processing.

  Palmprints acquired by the device are suitable for use in biometric authentication. Palmprint features and characteristics are acquired and then compared to a database record to authenticate the individual. Many techniques can be used to determine the characteristics of the palm in the image. One suitable technique is texture analysis. Texture analysis is suitable because it provides a high level of accuracy in low resolution images.

  In the embodiment described above, a CCD image sensor is used. In an alternative embodiment, a complementary metal oxide semiconductor (CMOS) sensor is used. The lower the cost of a CMOS sensor, the lower its resolution. However, this can be improved by using texture analysis.

In the preferred embodiment, the protrusion adjacent to the window 15 is a peg 10 . In an alternative embodiment, the surface 9 with the window 15 has a depression or recess in which the hand is placed with the palm down.

  The device is used to capture an image for use in the described method.

  Where reference is made in the above description to completeness or elements having known equivalents, those equivalents are included as if individually set forth herein. While the embodiments of the invention have been described, it will be appreciated that changes, improvements or modifications may be made without departing from the spirit of the invention or the appended claims.

It is a figure which shows a typical high resolution palm print image. It is a figure which shows a typical low-resolution power palm print image. It is FIG. (1) explaining the pre-processing of the image inside a hand. It is FIG. (2) explaining the pre-processing of the image inside a hand. FIG. 10 is a third diagram illustrating the preprocessing of the image inside the hand. It is FIG. (4) explaining the pre-processing of the image inside a hand. It is FIG. (5) explaining the pre-processing of the image inside a hand. It is FIG. (6) explaining the pre-processing of the image inside a hand. It is a figure which shows incorrect arrangement | positioning of a hand on a palm reading apparatus. It is a figure which shows the pre-processing image corresponding to arrangement | positioning of the incorrect hand on a palm reading apparatus. It is a figure which shows the pre-processing image which concerns on embodiment of this invention. It is a figure which shows the real part of the texture characteristic corresponding to a pre-processing image. It is a figure which shows the imaginary part of the texture characteristic corresponding to a pre-processing image. It is a figure which shows the mask of the texture characteristic corresponding to a pre-processing image. It is FIG. (1) explaining the difference of the image quality between the 1st and 2nd acquired images. It is FIG. (2) explaining the difference of the image quality between the 1st and 2nd acquired images. It is a figure which shows the result of the verification test with respect to the method by this invention. It is a figure which shows the result of the verification test with respect to the method by this invention. 1 is a schematic diagram of a palmprint image capturing apparatus according to the present invention. It is a top view of the image capture surface of this apparatus. FIG. 20 is a cross-sectional view taken along A-A ′ of FIG. 19, in which the CCD camera is surrounded by a circular light source. It is a figure explaining the unprocessed palm image captured by the apparatus.

Explanation of symbols

1 Sub-image
2 Y axis
3 X axis
4-7 Corner of sub-image
8 Housing
9 Housing top
10 pegs
11 CCD
12 lenses
13 Light source
14 Mounting arm
15 windows
16 screw

Claims (19)

Obtaining an image of an area of the skin surface from an individual;
Analyzing the image to extract texture features of the skin surface area;
Biometric authentication method comprising comparing texture features with reference information in a database. Obtaining an image of a portion of the inner surface of a person's hand;
Obtaining a sub-image of the skin surface within a predetermined area of the inner surface of the hand;
Analyzing the sub-image to obtain texture data for the skin surface;
Comparing the texture data with reference information in a database.   The method of claim 2, wherein the predetermined area depends on one or more characteristics of the hand.   4. A method according to claim 2 or 3, wherein the one or more characteristics are areas between fingers of a hand. Identifying at least two points where the sub-image represents a region between fingers of a hand;
Determining a coordinate system having first and second axes, the first and second axes having the two points located on the first axis and equidistant from the second axis When,
A method according to any of the preceding claims, comprising the step of determining a parameter of a sub-image in a coordinate system using the distance between the two points.   The parameters of the sub-image are (0.25D, 0.5D), (1.25D, 0.5D), (0.25D, -0.5D) and (1.25D, -0.5D) (where 6. The method of claim 5, comprising a point in the coordinate system represented by D is a distance between two points).   The method according to claim 5 or 6, further comprising the step of normalizing the sub-image.   A method according to any preceding claim, wherein the step of analyzing the sub-image comprises using a Gabor filter.   The method according to any of the preceding claims, wherein the step of analyzing the sub-image comprises the step of dividing the layer of the low-resolution sub-image using Gabor analysis.   A method according to any preceding claim, wherein the sub-image is divided into two parts, a real part and an imaginary part, each part being stored as a vector. The method according to claim 11, wherein the comparison of the texture data with the reference information in the database is based on a Hamming distance of the form: That is,
_However, P _{R (Q} R) and _{P I (Q} I) is a real part and an imaginary part. A container having a window therein;
An image sensor disposed within the container and disposed to capture an image through the window;
A light source arranged to illuminate the window;
A palm print comprising at least one protrusion adjacent to the window, the protrusion being arranged in a known juxtaposition with respect to a hand appropriately placed on the window for capturing an image including a palm region of the hand Image capture device.   14. The device of claim 13, wherein the protrusion is a peg or pin disposed between fingers of two or more hands appropriately placed on the window.   The apparatus according to claim 13 or 14, wherein the light source is annular and an image sensor is in the center.   16. The apparatus according to claim 13, wherein the image sensor is a CCD or CMOS sensor. A container having a window therein;
An image sensor disposed within the container and disposed to capture an image through the window;
A light source arranged to illuminate the window;
A control device operable to control the image sensor and the light source to capture an image;
A palmprint image capture device comprising at least one projection, a projection arranged in a known juxtaposition with respect to a hand appropriately placed on the window for capturing an image including a palm region.   18. The device of claim 17, wherein the protrusion is a peg or pin disposed between fingers of two or more hands appropriately placed on the window.   The apparatus according to claim 17 or 18, wherein the light source is annular and has an image sensor in the center thereof. 20. An apparatus according to any one of claims 17 to 19, wherein the image sensor is a CCD or CMOS sensor.