JP2004288221A - Iris identification system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an iris identification system and method for identifying the identity of a living animal by iris identification. <P>SOLUTION: The iris identification system comprises a control unit 70 interfaced with a data processing unit 60 for preliminarily processing an input image signal to a process data. The processed data shows one of a plurality of parameters for iris identification selected from a group including (1) density and texture pattern of iris fibrous structure using frequency conversion method, (2) pupillary reaction, (3) shape of pupil, (4) autonomic nerve ring reaction, (5) shape of autonomic nerve ring, (6) presence of recess, (7) position of recess, and (8) shape of recess. The control unit compares the processed data with the parameters preliminarily stored in the database to recognize the identity. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to an iris identification system and a method for confirming the identity of a living body by using an iris image. More specifically, the present invention measures the iris fiber structure, the shape of the autonomic nerve ring (ANW) and the pupil and its spontaneous response to light, and one or more identifying parameters relating to the presence, shape and location of indentations from the iris image. Iris identification method.

Known iris identification systems for identifying particular humans use an iris code from a database of iris image signals taken from the human eye to allow or reject particular humans according to the results of the comparison. Compare with the searched corresponding iris information. This conventional iris identification system, however, does not achieve an acceptable high level of identification accuracy (see US Pat. No. 6,037,059).
JP-A-11-15972

  In addition, because various iris identification systems have limitations in identifying whether or not a living human is observing the true iris, the high risk of falsely identifying a forged iris image may result. It cannot be used safely with systems such as banking systems, electronic settlement systems, and the like.

  It is an object of the present invention to identify individuals using an iris recognition system having a high level of accuracy. Another object is to achieve a high level of identification accuracy by analyzing a number of properties associated with the iris.

  Thus, one object of the present invention is to provide an iris image, the response of the pupil and the autonomic nerve ring caused by light, the shape of the autonomic nerve ring and the pupil, and the iris fibers obtainable from the presence, location and shape of the depression. It is to quickly and unambiguously identify a living iris by measuring a plurality of iris identification parameters according to the structure.

  A system for confirming the identity of a living animal by iris scanning according to the present invention relates to a control unit for receiving identification information identifying a living animal, and an identifiable organism for comparison with the identification information. A data storage unit including a database accessed by the control unit for containing predetermined personal information. An iris image pickup unit including a camera is operated by a control unit to first capture an iris image and generate an input image signal when the identification information corresponds to predetermined information. The data processing unit pre-processes the input image signal into processed data. For each identifiable living body, the storage unit comprises: (1) density and texture morphology of the iris fiber structure using the frequency conversion method; (2) pupil response; (3) pupil shape; , (5) the shape of the autonomic nerve ring, (6) the presence of the depression, (7) the position of the depression, and (8) the shape of the depression. At least one of the following parameters: The control unit may be operable to compare the processed data with the parameters to determine whether there is a match indicating that the identity has been confirmed.

  According to a preferred aspect of the present invention, the camera is configured to image both eyes of a living animal. The data processing unit separately processes the image input signal into processing data representing each eye. In this way, a higher level of identification accuracy is obtained by iris scanning to confirm the identity of the surviving animal.

  To access the identification system, the organism to be identified first enters identification information into the system. In some embodiments, the identification information data can be entered using a card reading unit, key entry unit, etc., that accesses the system using a PIN (Personal Identification Number) or some other unique identification information. . Also, a voice activation system can be used.

  The iris image pickup unit is preferably operated by the control unit in conjunction with the camera and includes a light source for capturing a plurality of iris images in succession and generating an input image signal. By illuminating the eye, it is noted that the iris contracts, and therefore, a continuous image allows the system to capture the iris contraction and, if desired, the subsequent iris dilation. The light source is preferably an infrared light source.

  In accordance with another unique feature of the present invention, a system is provided for controlling a camera based on accessing predetermined personal information in a database including characteristics of a user's physique translated into positioning data. And an interface between the iris image pickup unit and the control unit to enable automatic alignment.

  As described above, the system preferably has the ability to analyze different parameters that can be attributed to iris characteristics. However, the present invention can be operated to confirm the identity of the user by screening and analyzing only one of the above described process parameters or a subset of the process parameters below the entire parameter set.

  The unique methods used to analyze the density and texture morphology of the iris fiber structure preferably include frequency conversion methods using the Haar transform. According to a presently preferred aspect, the present invention provides a variable multi-sector system for spectral transformation using a Haar transform function. According to this unique method, a selected one of the iris images is divided into a plurality of sectors and a number of Haar function coefficients are calculated for each sector. This multi-sector system of spectral analysis advantageously makes it possible to eliminate sectors with irises that are distorted by interference caused by the presence of defects or interferences, for example by eyelids, eyelashes and the like. These deficiencies are clarified by the generation of high frequency coefficients and predetermined sharp variations between adjacent sector coefficients. Therefore, the analysis for each adjacent sector makes it possible to apply the Haar transform to the iris recognition technology.

  Using the Haar transform, it is possible to produce an iris reference record with an unprecedented high level of recognition accuracy, preferably in combination with selecting Haar coefficients representing the selected low frequency zone.

  The present invention, in one aspect, uses a series of iris images to generate a pupil histogram or pupilogram represented by curves representing iris contraction and dilation over time. The present invention provides a novel technique for analyzing pupilograms by analyzing the gradient change as a further means to improve identification accuracy. The invention is also characterized by analyzing the pupil boundary, shape and position as a further means for improving identification accuracy.

  The present invention also processes information about the presence, location and shape of the indentation as a further means to improve the shape and response of the autonomic nervous ring and, if present, the identification accuracy.

The present invention also contemplates and enables the analysis of the iris identification parameters described above for the system, using it for other than iris identification. For example, it is within the scope of the present invention to form a system for testing an individual for the presence or use of a drug or alcohol. That is, the present invention includes the following aspects.
1. A system for testing for the presence of a drug or alcohol in humans, comprising:
(A) a control unit for receiving identification information identifying living animals;
(B) an iris image pickup unit including a camera operated by a control unit for capturing an iris image and generating an input image signal;
(C) a data processing unit for pre-processing the input image signal into processed data, the control unit responding the processed data obtained as a result of processing the input image signal to a light stimulus per time. The pupilogram curve is calculated to represent the mean contraction and dilation of the iris pupil, thereby comparing a person who is addicted to alcohol or drugs with a person who does not contain alcohol or drugs. Said system operable to determine whether the processed data is indicative of present or past addiction due to the different shape of the.
2. A method for testing for the presence of at least one drug or alcohol in the human body, comprising:
(A) obtaining a plurality of images of substantially the same portion of the human iris and pupil at different times;
(B) processing the plurality of images into a pupil histogram;
(C) calculating characteristics of the pupil histogram processed in step (b); and (d) confirming the presence of at least one of a drug or alcohol based on the comparing step (c). The method as described above.
3. The method of claim 2 wherein the flatness characteristic of the curve associated with the pupil histogram is indicative of at least one of a drug and an alcohol. In this type of system, a control unit for receiving identification information identifying a living animal, and a control unit for including predetermined personal information relating to an organism that can be identified for comparison with the identification information. A data storage unit is provided that includes a database accessed by the unit. The iris image pickup unit includes a camera scanned by a control unit for capturing an iris image and generating an input image signal. The data processing unit pre-processes the input image signal into processed data. The control unit processes the data resulting from processing of the input image signal representing the average contraction and dilation of the iris pupil in response to light stimuli per hour to determine whether the processed data indicates current or past addiction. Measure. This system is based on the finding that the shape of the curve of the pupilogram is different when a person is addicted to alcohol or drugs compared to the shape of a curve when a person does not contain alcohol or drugs.

  Methods for confirming the identity of living animals or testing for the presence of at least one of drugs and alcohol in the human body are also based on the particular selection of iris identification parameters as described above. , Provided by the present invention. Still other objects and advantages of the present invention will become apparent to those skilled in the art from the following detailed description. In the following description, only the preferred embodiments of the invention are shown and described merely to show the best mode intended for practicing the invention. As will be realized, the invention is capable of other different aspects, and its several details are capable of modifications in various obvious features, all without departing from the invention. Accordingly, the drawings and description should be regarded as illustrative and not restrictive.

  FIG. 1A is a block diagram of a system 14 for identifying a living iris according to the present invention. Living organisms include living animals having an iris for identification, including humans and animals. The system 14 preferably includes a card reading unit 40 for reading and identifying information recorded on the card storing the personal card number by accessing the information recording medium of the card. The key input unit 50 having a plurality of keys for inputting a PIN generates an electric signal corresponding to each key input. The iris image pickup unit 20 automatically and continuously flashes in accordance with a predetermined order of a specific control signal, and the camera having the auto-focus function operates for several seconds (for example, 25 frames or more per second). It has multiple light sources that allow it to capture a dynamic image of the iris (at the rate at which it is generated). The driving unit 30 supplies power for illuminating and adjusting the camera position to the iris image pickup unit 20 by supplying specific control signals using a monitor.

  The data processing unit 60 pre-processes the input image signal captured by the iris image pickup unit 20. The data storage unit 80 comprises: (1) autonomous responsive to light; (1) the density and texture form of the iris fiber structure that can be extracted from the individual iris image signals, each associated with a respective human personal card number or PIN. (3) pupil movement in response to light; (4) ANW shape; (5) pupil shape; (6) presence, location and shape of depressions. Form a database for storing parameters. The control unit 70 searches the database for the card number or PIN entered or read by the card reading unit or key input unit, obtains a dynamic iris image, and automatically positions the camera based on personal information. Identify an iris for a particular human by adjusting and verifying the identity of the user by measuring a plurality of iris identification parameters to determine the identity of the user from the image signal. In this case, not only the iris information for a particular person, but also additional information about each person (e.g., directly related to the camera height when capturing the iris image or the most appropriate camera height for each person). Individual height information) can be stored in the data storage unit 80, allowing the unit 80 to automatically adjust the camera height by reading additional information.

  FIG. 1B is a side perspective view of a binocular camera and a portion of an illumination system showing an example of the iris image pickup unit 20 of FIG. 1A. The iris image pickup unit 20 has a camera 16 and a plurality of first and second lenses 11 and 12 at positions required to capture an image of the iris. It is preferable to use a camera 16 having a variable capture zone (flex zone) for autofocus. Further, it is preferable to use a second lens 12 having a depth of focus greater than 5 mm to prevent the autofocus function from deteriorating as a result of user eyelid and eyelash movement and blinking. The first and second lenses 11, 12 are preferably provided for each eye.

  The iris image pickup unit 20 further includes a visual field guidance unit for preventing the iris from being biased to one side by fixing the user's eye to the central portion of the camera in connection with the iris image pickup unit 20 of FIG. 1B. 18 (see FIGS. 2 to 4). Further, in some embodiments, a side light source for initial autofocus and brightness adjustment can be used. FIG. 2 is a reference diagram from FIG. 1B to show the view pointer, and FIG. 3 shows the view pointer of FIG. 2 in detail. FIG. 4 is a reference diagram for explaining the operation of the visual field pointer in FIG. 2 for reflecting light from the visual field guiding light source L4 to the user's eyes.

  As shown in FIGS. 2-4, the field-of-view guidance unit can be comprised of a guidance light source L4 (also indicated by reference numeral 17) that emits light blue or red light to guide the user's field of view. The reflection unit 18b reflects the light from the light source 17 from the central portion of the camera 16, and the support unit 18a aligns and supports the reflection unit at the center of the lens. In this case, the support unit 18a is preferably made of a transparent glass-like material so as to prevent affecting the iris image even when located at the center of the lens. The reflection unit 18b has a reflection surface having a certain inclination so as to reflect light from the visual field guide light source L4 to the user's eyes.

  FIG. 5 is a front view of the second lens 12 of FIG. 1B. In one arrangement of the front surface 12 of the second lens, a flash light source L3 (12b) is provided that stimulates pupil and ANW movement (ie, contraction and dilation). A plurality of infrared light sources L2 (12a) are arranged in an annular or annular arrangement along the periphery of the lens 12, so that the software allows the camera 16 to capture an iris image and a clear pupil image. Is controlled. Flash light source 12b is preferably a blue diode that is controlled by software to automatically emit light in short cycles.

  FIG. 6 is a flowchart of an example of capturing an image signal by the iris image pickup unit 20 performed by computer control. In adjustment step S61, the camera position (height) is adjusted by using personal information retrieved from a database based on the card number or PIN, and the plurality of light sources (12a and 17) are simultaneously set for autofocus. Switched to ON, attracts the user's eyes to the objective lens. In step S62, the user's eyes are guided to the objective lens by using the light source 17 that emits light. Automatic focusing of the camera is performed in step S63. In step S67, a dynamic image of the moving ANW and pupil is acquired for a few seconds by the light source 12b emitting light (S66) in a short cycle to induce pupil and ANW movement of the iris. In step S68, after the above-described steps S61 to S68, all light emitting sources are turned off.

  FIG. 7 is a detailed flowchart of the adjustment of the camera height in step S61 of FIG. The flowchart includes the following steps: inputting a card number or PIN (S71); acquiring information on the camera height of the corresponding individual by searching a database for the input card number or PIN (S72). ); After comparing the acquired information of the camera height with the information of the current height of the camera, adjust the camera height suitable for the corresponding individual by rotating the motor forward or backward (S73). I do.

  FIG. 8 is an example of an iris image to illustrate selected zones (x) of input iris signals required to perform iris analysis according to one aspect of the present invention. The zone to be selected for analyzing the iris must include at least a portion of the ANW 10b and the entire pupil 10a. Further, the selection zone (x) is preferably a portion that is clearly visible and is not affected by eyelids, eyelashes, and the like, and is set within a range of more than １ ／ and less than の of the iris diameter.

  FIG. 16 is an illustration of iris image analysis using a variable multi-sector system of spectral analysis in a preferred embodiment of the present invention, as described in more detail below. The variable multi-sector system of spectral analysis allows the selection of the visible portion of the iris to be distorted by interference for analysis, thereby improving the reliability of the identification process. Anatomical features of the eye caliber and eyelid bulge can be identified and eliminated. Edge defects or interferences, or glare, which are almost impossible in advance, can be identified and eliminated.

  FIG. 9 is a high-level flowchart for explaining the whole iris identification method of the present invention, and FIGS. 10A to 10C are detailed series of flowcharts of the steps of the high-level flowchart of FIG.

  As shown in FIGS. 10A-10C, in the iris identification method according to the present invention, various aspects can be recognized by including one or more of the following processes. First, the identity of a user is confirmed by searching and acquiring personal information according to a corresponding number from a database by a card number or PIN confirmation process S101 to S103. After inputting the card number or PIN, the dynamic iris image is acquired by adjusting the camera position according to the user's personal information obtained during the confirmation process and controlling a plurality of light sources by the image acquisition processes S104 to S105. Is done.

  The different identification parameters that can be obtained and analyzed according to the invention are described below.

  After measuring the identification parameters using the iris fiber structure confirmation process S106-S110 by transforming the image signal of the zone suitable for iris analysis by Wavelet transform of the selected still iris image acquired during the above process The identity of the user is ascertained by the corresponding identification parameters. The transformation can be applied to the zones of FIG. 8 or to the individual sectors of FIG. 16, as described below with respect to the variable multi-sector system of spectral analysis.

  In one aspect of the present invention, the iris structure confirmation process includes pre-processing the dynamic signal of the iris image in steps S106 and S107 (FIG. 10A), thereby forming a zone suitable for iris image analysis (selection zone x; substantially (Including a part of the ANW and the entire pupil and within a range of more than 1/3 and less than 1/2 of the iris diameter); in step S108, a two-dimensional Wavelet transform (for example, such as a Haar transform) is performed. After transforming the selected still image signal of the zone selected by the Wavelet transform), the Wavelet transform coefficient identification parameters representing specialized information on iris fiber structure density and texture morphology are measured; By searching for the corresponding parameter from To confirm the identity of the Za; it can be comprised of steps.

  In general, the density and texture morphology of the iris fiber structure varies from person to person. When a Wavelet transform (particularly, a Wavelet transform such as a Haar transform) mainly including information on a density in a low-frequency zone is performed, particularly, a low-frequency component of a two-dimensional Haar transform coefficient includes most information of an iris fiber density and a texture form. Having.

  The low frequency component is obtained by performing Wavelet transform on the iris image signal selected as described above, measuring the frequency characteristic, and determining the selected low frequency coefficient of the two-dimensional Haar transform indicating the frequency characteristic as an identification parameter. , Iris fiber density and texture morphology. The Haar transform so applied in the present invention uses the following Haar function, which is symmetric, separable, unitary and varies with size (scale) and position.

  A further Haar function is defined as:

スペクトル分析アルゴリズムを簡単にするために、好ましい態様は、標準化ファクター２−ｍ／２を無視する。したがって、上記関数の定義は次のようになる。

To simplify the spectral analysis algorithm, the preferred embodiment ignores the normalization factor 2-m / 2. Therefore, the definition of the above function is as follows.

この関数は、また、シグナムＨａａｒ関数と呼ばれる。関数の二進数がリスティングにおける関数の次数を決定する。本発明において、実際のＨａａｒ変換を測定する場合には、大きなサイズのイメージのために膨大な計算となる。しかしながら、コンピュータ計算時間は、変換フローグラフのバタフライ構造によって大きく向上する。

This function is also called the Signum Haar function. The binary of the function determines the order of the function in the listing. In the present invention, when measuring the actual Haar transform, an enormous calculation is required for a large-sized image. However, the computation time is greatly improved by the butterfly structure of the transformation flow graph.

  FIG. 15 shows a graph of a rapid spectrum conversion algorithm using the Signum Haar function (here, n = 18). This graph represents the butterfly structure of the conversion flow graph. Along the vertical axis of the graph, P0 to P7 indicate pixel values to be converted.

  In this flow graph, the value for each successive level that progresses to the right depends on the previous value based on the following equation:

式中、ｍ＝ｎ＋１であり、Ｘ［ｎ］はピクセルｎの明度である。

Where m = n + 1 and X [n] is the brightness of pixel n.

  As a result of using the butterfly structure of the conversion flow graph, the spectrum conversion program with the Haar function operates 200 times faster than the conventional method.

  Variable Multi-Sector Analysis A variable multi-sector system for spectral analysis using the Haar function is described below. The variable multi-sector system is preferably implemented throughout the zone selection method described above with respect to FIG. Occasionally, glare formation, interference at the eyelids and eyelashes may occur within the section of the iris, or a large visible portion of the iris may not be available for analysis. The interference or unavailability of the iris portion reduces the number of sectors available for comparison, thereby reducing the reliability of human identification.

  Image analysis begins by dividing the iris image also into an inner annular zone that includes outer ring 100 and pupil 120, as shown in FIG. The boundary dividing the outer inner and outer rings 100 from the annular zone is set at a distance of 50 pixels from the origin of the iris. Values larger or smaller than 50 pixels can be used in further aspects. The outer border of the outer annulus begins approximately at the sclera-iris border. The diameter of the outer annulus can vary due to many factors, such as different iris sizes between individuals, different ranges for obtaining iris images, and the like. Since the portion of the iris corresponding to the outer ring 100 and the region 110 between the pupil 120 and the outer ring 100 contract and expand at different rates, different normalization factors are used for the inner and outer rings.

  In a currently preferred embodiment, each region 100, 110 is then further divided into 16 sectors, preferably of equal size, ie, a total of 32 sectors, indicated by the reference signs I0-I31. For each of the 32 sectors I0 to I31, a 1,024 Haar function coefficient is calculated using Wavelet transform as described above to form a 32 × 32 matrix. However, it is within the scope of the invention that the total number of sectors in all zones (n + p), the number of sectors in each individual zone (n, p), the size of the sectors and the number of Haar function coefficients may vary. Is within. That is, each zone may have a different number of sectors (n, p), ie, the region 110 can be radially divided into ten sectors (p = 10), while the outer ring 100 Is radially divided into 16 sectors (n = 16).

  Then, generally, sectors that are obstructed or interfered by eyelids or eyelashes, indicated by reference symbols 13A and 13B with respect to the lower and upper eyelids, respectively, are rejected by adjacent sector coefficient comparison. If a sharp variation in the high frequency coefficient is observed by comparing adjacent sector coefficients, the sector is rejected as a defect. Referring to FIG. 16, the sector comparison is performed on a group of four sectors beginning at the horizontal line dividing sector 0 from 15 and sector 7 from 8, proceeding clockwise or counterclockwise for adjacent sectors. I do. In this method, sectors I0-I3, I7-I4, I8-I11 and I15-I12 are compared.

  For example, the high frequency coefficients in sectors I0 and I1 are compared, and if the difference does not exceed a predetermined sector high frequency coefficient threshold, sector I1 is recognized as good. Sector I0 is recognized as good or rejected during pupil analysis if closed by eyelids. Next, the coefficients of sectors I1 and I2 are compared, and the border of the eyelid is located at sector I2, as shown in FIG. A boundary in sector I2 causes a difference that exceeds the sector high frequency coefficient threshold, and sector I2 is rejected. After rejecting sector I2, sector I3 is similarly rejected. A similar process is repeated for the remaining quadrant sectors.

  After removing the rejected sectors from the analysis, a subset of the complete 1,024 Haar coefficients is selected. The number of coefficients selected is determined by various factors. If the coefficients are unnecessarily large, the size of the database increases, and if the coefficients are too small, the quality of recognition decreases. In addition, some coefficients are not selected because they change with changes in image brightness, and some high frequency coefficients contain too much noise. As a result of experiments, the preferred embodiment uses 31 coefficients selected from a 32 × 32 matrix of 1,024 Haar coefficients. The specific matrix positions of the selected coefficients are: (0,1), (1,0), (1,1), (0,2), (0,3), (2, 0), (3,0), (1,2), (1,3), (2,1), (3,1), (0,4), (0,5), (0,6) , (0,7), (4,0), (5,0), (6,0), (7,0), (1,4), (1,5), (1,6), ( (1,7), (4,1), (5,1), (6,1), (7,1), (2,2), (3,2), (2,3) and (3,1) 3). More or less coefficients may be used in different embodiments.

  By selecting 31 coefficients from each of the 32 sectors I0-I31, an iris reference record approximately 1 kilobyte in size can be generated. The first 32 bits of the record contain the result of the sector rejection analysis.

  The new iris image that is identified is processed in a manner similar to the reference image above. The resulting characteristic record is compared to all iris reference records recorded in the database. The sum of the difference between the input and the reference iris coefficient is calculated for each sector. The value of the sum is in the range of 0 to 2 for coefficient normalization. A total value of 1 represents an absolutely gray image, 0 represents a perfect match of the sector spectrum, and 2 represents a spectrum of equivalent modules but of opposite sign.

  Direct experimentation has shown that some sectors differ from the reference sector due to iris peristalsis (rapid unconscious movement of some areas of the iris). Therefore, only sectors with a total value less than 1 are preferably used in iris image analysis. In this way, the sectors perturbed by peristalsis are added to the rejected sectors and are excluded from the recognition task. The minimum number of sectors used in iris image analysis is 10 for each eye. Fewer sectors can be used if the quality of perception is correspondingly reduced. Multiply the results of the sector coefficient comparisons to emphasize the differences. Therefore, if the image to be identified matches the reference image, the value obtained will be close to zero. In contrast, the values derived by different images are in the range of hundreds to thousands.

  Since the low frequency spectral coefficients contain much information about the density and texture morphology of the iris fiber structure, the coefficients of the low frequency zones experimentally selected from the results of the Haar transform are used for identification.

  By detecting the pupil and the ANW from the acquired dynamic image of the iris by the pupil and ANW reaction confirmation processes S111 to S116 of FIG. 10B, the identity of the user is confirmed by the corresponding identification parameters. Next, the identification parameters are measured using the dynamic response (contraction and dilation) of the pupil and the ANW from the detected dynamic image signal.

  Further, in one aspect of the present invention, the pupil and ANW response confirmation process may comprise the following steps: detecting the pupil zone by computer processing a central portion of the acquired dynamic image of the iris. (S111, S112 in FIG. 10B); Computer processing of the ANW zone (S113); Recognition of the surviving pupil and the ANW, respectively, after calculating the reaction (expansion or contraction) time of the moving pupil and the ANW in the detected zone. An identification parameter for the user is measured (S114); the identity of the user is confirmed by searching a corresponding parameter from a database.

  FIG. 11 is an exemplary image of the iris 10c for explaining the movement of the pupil 10a and the ANW 10b at a constant angle θ, and FIG. 12 is a diagram illustrating an average pupil radius R1 converted from one-dimensional data with time variation. Is a graph showing a pupil radius R2 at an angle θ and a ANW radius R3 at a constant angle θ. In the graph, the first time “t1” represents the operation time of the flash light source, the second time “t2” represents the start time of the contraction of R1, and the third time “t3” represents the time when the average pupil radius is the smallest. The fourth time “t4” represents the minimum time at a constant pupil radius θ, and the fifth time “t5” represents the minimum time at a constant ANW radius angle. Therefore, as a result of comparing these dynamic images, it is possible to obtain respective identification parameters through the pupil and the ANW that respond unconsciously to light.

  For this purpose, the present invention provides a predetermined reference value that is considered to represent a living body if the movement of the pupil radius exceeds a certain percentage (for example, 5% or more) when the pupil contracts. Used. Pupil edges must also be detected from the captured iris dynamic image in order to observe pupil movement. At present, it is possible to detect the pupil edge after determining the approximate center of the iris by using a symmetric center search algorithm.

  According to this method, it is possible to identify the iris without causing an error even if the iris image cannot be captured at the center thereof and is somewhat shifted right or left.

  If the iris image is too biased for identification, it can be captured again. In addition, when another image is captured instead of the iris image, a true image and a false image can be distinguished in many cases.

  In the symmetry center search algorithm described above, the following function F (i) is obtained for the horizontal and vertical lines of the image.

ここで、

here,

およびＮはイメージ線の長さであり、ｘ（ｉ）は水平線又は垂直線のｉ番目のピクセルの輝度であり、ｉ≦０の場合にはｘ（ｉ）＝ｘ（０）であり、ｉ≧０の場合にはｘ（ｉ）＝ｘ（Ｎ）である。

And N is the length of the image line, x (i) is the luminance of the i-th pixel of the horizontal or vertical line, and if i ≦ 0, x (i) = x (0), i When ≧ 0, x (i) = x (N).

  In such a case, the definition domain (i) that minimizes the absolute value of the function F (i) exists at the center of symmetry. Applying these methods to the horizontal and vertical lines, the point of intersection of the definition domain (i) that minimizes the absolute value of the function F (i) is set as the center of symmetry. If the domain (i) of the definition of horizontal and vertical lines scatters without traversing, especially if they deviate from a predetermined range, this means that the captured image is not an iris image or an iris image Indicates that the iris image is excessively biased to the right or left, and this iris image must be recaptured before further identification can be performed.

  If a dynamic image is captured for the iris, the pupil and the ANW are contracted and expanded by the flash light source 12b. Pupil and ANW movements do not occur at the same time, but are of different types for each person. The parameters from the pupil and the ANW response (FIGS. 11 and 12) resulting from such non-simultaneous movements are used for identification.

  Referring to FIG. 17, the pupilogram or pupil histogram shows the average pupil contraction and dilation in response to a flash of light at time T0. The vertical axis indicates the size or radius of the pupil, and the horizontal axis indicates the time beginning with T0 or the time of the light flash. TL is the time at which the pupil begins to contract in response to a flash of light; TL-TO is the latency time of the pupil response. AL is the average radius of the pupil before the flash of light. Tm is the time when the pupil contracts to the minimum radius Am. The value obtained by subtracting the latent time from the time for the pupil to contract is Tm-TL or TP. T1 and T2 are the 40th and 70th times of the dynamic iris image, respectively, and A1 and A2 are the corresponding pupil radii at times T1 and T2. SP is the rate of pupil contraction between TL and Tm. DAB is the linear distance along the pupil contraction curve between TL and Tm. % A1 is the ratio of the pupil dilation distance or A1-Am at T1 to the average pupil contraction distance or AL-Am. % A2 is the ratio of the pupil dilation distance or A2-Am at T2 to the average pupil contraction distance or AL-Am.

  Pupilograms can be used to detect drug and / or alcohol use by individuals, as human pupylograms using drugs or alcohol are different from human pupylograms without drugs or alcohol. it can. In a human pupilogram using a drug, the time Tm at which the pupil contracts to the minimum radius becomes slower than the standard time. Pupilograms for humans using alcohol are flatter, ie, the DAB is smaller for alcohol users, compared to humans who do not use alcohol. Also,% A1 and% A2 are smaller for alcohol users.

  Next, the ANW and pupil shape confirmation processes S117 to S119 detect the ANW and the pupil from the selected iris image, and measure the identification parameters according to the detected shapes of the ANW and the pupil to confirm the identity of the user. . The ANW shape is uniquely measured in the present invention by using a central axis transform applied to the selected iris image signal. Pupil shape is uniquely measured in the present invention by using edge detection and curve matching algorithms, as is known in the art.

  Furthermore, in each aspect of the present invention, the ANW and pupil identification process may comprise the following steps: Step A117 (FIGS. 10B and 10C) detects ANW and pupil from the selected iris image signal. In step S118, the detected ANW two-dimensional image signal is converted by the center axis conversion, and the pupil shape is converted by the edge detection and curve matching algorithm as described above, thereby identifying based on the shape of the ANW. The parameters are computed by computer; steps S119 and S120 confirm the identity of the user by searching the database for the corresponding identification parameters.

  FIG. 13 is a view showing an example of the ANW 10b and the depression 10d representing the edges detected from the iris image. FIG. 14 is a diagram showing the ANW 10b and the depression 10d shown in FIG. FIG. The central axis transformation of an object having a specific shape is, for example, the position of the center of a circle that touches the boundary of the object at two or more points. In other words, this is the position of the point closest to the target boundary. Therefore, since it is possible to convert a two-dimensional object into one-dimensional data by means of a central axis transformation, this can be applied to identify the ANW shape and the shape and position of the pupil.

  FIG. 18 is an exemplary diagram of a pupil and an iris showing a shape, a position, and identification parameters. Pupil shape is uniquely measured in the present invention using edge detection algorithms known in the art. The pupil end is generally indicated by reference numeral 10a in FIG. Further, the centers of the iris and pupil, indicated by reference symbols 10i and 10h, respectively, in FIG. 18, are measured by geometry and used as identification parameters. The separation direction and distance between the center of the iris and the center of the pupil, generally indicated by reference symbol 10j, are used as identification parameters.

  Furthermore, many pupils are not perfectly circular in shape, but many are ellipses that deviate in the direction of the major axis of the ellipse. In a preferred embodiment, the orientation and magnitude of the pupil ellipticity is uniquely measured in the present invention using curve and shape matching algorithms known in the art. Referring to FIG. 18, the major and minor axes of the pupil ellipse are indicated by reference signs 10m and 10n, respectively. The pupil ellipticity is expressed as the ratio between the short axis 10n and the long axis 10m. Recognize and characterize the flat portions along the periphery of the pupil, generally indicated by reference numeral 10f in FIG. 18, and the areas where the pupil is depressed or bulging, indicated by 10g and 10e, respectively, and recognize them as further identification parameters. Quality can be improved.

  Further, it is determined whether or not a dent exists using the dent confirmation processes S121 to S127. If the dent exists, the zone is detected. Then, the position and shape of the detected dent are determined. The identity of the user is ascertained by measuring the identification parameters based on

  Further, in the iris identification method according to the present invention, the indentation confirmation process can be applied to the above-described embodiment by including at least one of the following multiple steps shown in FIG. 10C. That is, a first dimple confirmation process S121 is performed, and then, according to steps S125 to S129, the identity of the user is determined by determining whether a dimple exists based on the selected still image of the acquired iris. Then, if a dent exists, the image signal of the dent is converted by central axis conversion, and the identification parameter is measured based on the position and shape of the dent from the conversion result. The second indentation confirmation processes S122 to S124 are performed when no indentation exists as a result of the first indentation confirmation process. After it is determined again whether or not the indentation exists in the reference iris, the second indentation confirmation process S122 to S124 is performed. If there are no depressions, the user is allowed; conversely, if there is a depression in the reference iris, the user is rejected.

  Further, in each aspect of the present invention, the first dip identification process may comprise the following steps: Step S121 (FIG. 10C), from the acquired iris still image signal, obtains the edge of the iris image. Is detected to determine whether or not a dent exists; if a dent is present as a result of the determination step, the dent zone is detected in step S125; and the detected dent zone is determined in step S126. After the two-dimensional image signal is converted by the central axis conversion, the identification parameters are computer-calculated based on the shape and position of the hollow; in steps S127 to S129, the identity of the user is confirmed by searching the database for the corresponding parameters. .

  Further, in each aspect of the present invention, the second indentation confirmation process may comprise the following steps: Step S121 determines whether an indentation is present in the input image; Step S122. Then, if there is no depression in the input image, it is determined again whether or not there is a depression in the reference iris; in step S123, if the depression is not present in the reference iris, permission is given. If there is a depression in the reference iris in step S124, the rejection is made.

  The operation of the present invention configured as described above and the effects achieved by the present invention will be described below using examples.

  First, the card number is read by the card reading unit 40 or the key input unit 50, or the user's PIN is input by keying. A card number and PIN verification process is performed to determine whether the corresponding number exists by searching for the card number and PIN in the data storage unit 80 in the control unit 70.

  If there is no corresponding number, the user is rejected as an unidentified user. If a corresponding number is found, then the control unit 70 allows the user to be identified as an identifiable user and then, after reading further personal information corresponding to the card number or PIN, the control signal To the motor drive unit 30 to adjust the camera height to suit the user. At this point, the control signal compares the height difference between the current camera position and the camera position appropriate for the user and then adjusts the camera height by driving the motor forward or backward. Adjust automatically. Further, the width between the two eyes can also be adjusted for each user.

  Next, the control unit 70 controls the iris image pickup unit 20 to turn on the infrared light source 12a, thereby performing the first autofocus adjustment. The user is guided to the visual field pointer 18 by turning on the visual field guiding light source 17, thereby preventing the iris image from being biased in the position direction. Therefore, when capturing the iris image, the field of view of the captured human iris image is fixed to the central portion of the lens by the guide light source 17 (that is, the light blue light source).

  The iris dynamic image is taken for a few seconds while operating the flash light source 12b in a short cycle. Then all light sources are turned off. The iris image must be taken to show the changing shape of the pupil and the ANW by the flash light source 12b, which is controlled in a short time by predetermined software. For this purpose, the present invention takes more than 25 iris images per second.

  The continuous image of the iris captured as described above is pre-processed in the data processing unit 60. Next, after selecting a zone suitable for iris analysis, an image signal of the corresponding zone is selected. In a preferred embodiment, the analysis is performed on a sector-by-sector basis, according to the variable multi-sector system of spectral transformation described in detail above. In another embodiment, the zone used to analyze the iris is a horizontal strip-like area that includes the entire pupil shown in FIG. 8, ie, a clearly visible portion. This is because it is not affected by the identified human eyelashes, eyelids, etc.

  The control unit 70 performs an iris fiber structure verification process to verify the identity of the user by using only the selected low frequency components, as described in detail above.

  Next, in the control unit 70, a pupil and ANW response confirmation process is performed, and after analyzing the pupil and ANW response from the pupil dynamic image, if the pupil and ANW contraction and expansion are caused by the flash light source 12b. The image captured as representing a living body is permitted, and if there is no movement, the image captured as not representing a living body is rejected.

  Subsequently, an ANW and pupil shape confirmation process is performed by the control unit 70 to convert the two-dimensional image signal of the ANW zone having different characteristics in each human from the selected still iris image into the above-described central axis transformation (or glass). By converting the data into one-dimensional data by the fire method, the identity of the individual is confirmed based on the ANW and the pupil shape. Certain features of the pupil as described above are also identified and compared.

  Next, the first and second depression confirmation processes are performed by the control unit 70, and the identity of the user is determined by an identification method that identifies whether the depression is present and, if so, its position and shape. Is finally confirmed. As shown in the result of the median axis transformation of the ANW, the dimples were extracted from the selected iris still images of FIGS. 13 and 14 by an image processing process, and if the dimple shape was clearly recognized by edge detection, By expressing the shape and position of the one-dimensional data by central axis conversion, the shape and position of the depression are used as identification parameters.

  Accordingly, the present invention provides a method for identifying multiple irises using iris fiber structure and texture morphology, pupil and ANW response to light, ANW and pupil shape, the presence, shape and location of dimples obtained from iris dynamic images. Banking and electronic processing / closing systems by providing a method of identifying specific humans by measuring the parameters for the identification of surviving human irises quickly and accurately by preventing counterfeiting. It has a number of advantages in that it can prevent financial accidents when applied and can eliminate security-related incidents when applied to access control systems.

FIG. 1A is a block diagram of an exemplary system for identifying a living iris according to the present invention. FIG. 1B is a side perspective view of the camera showing the embodiment of FIG. 1A and one embodiment of the iris image pickup unit in the lighting system. FIG. 2 is a reference diagram partially taken from FIG. 1B to illustrate a field pointer used for the iris image pickup unit of FIG. 1B. FIG. 3 is a detailed view of the field pointer of FIG. FIG. 4 is a reference diagram illustrating the view pointer of FIG. 2 that reflects the view guide light source to the user's eyes. FIG. 5 is a front view of the second lens in the system of FIG. 1B. FIG. 6 is a flowchart of an example in which an iris image pickup unit, which is controlled by a computer, takes in an image signal. FIG. 7 is a flowchart of an example of camera height adjustment in FIG. FIG. 8 is an exemplary diagram of an iris image illustrating a selected area (x) of an input image signal required for iris analysis. FIG. 9 is a high-level flowchart illustrating all operations of the iris identification method of the present invention. 10A to 10C are detailed continuous flowcharts of the flowchart of FIG. FIG. 11 is an exemplary diagram of an iris image for explaining dynamic movement of the iris and the autonomic nerve ring at a certain angle. FIG. 12 is a corresponding graph showing the pupil radius of FIG. 11 in the form of a pupil histogram or pupilogram and showing the autonomic nerve ring converted to one-dimensional data due to time variation. FIG. 13 is an exemplary diagram of an autonomic nerve ring and a hollow from which an edge of an iris image is extracted. FIG. 14 is a view showing an example of the autonomic nerve ring and the hollow of FIG. 13 converted into one-dimensional data by central axis conversion. FIG. 15 is a diagram of a butterfly structure of a Haar transform flow graph. FIG. 16 is an exemplary view of an iris image for explaining a variable multi-sector system of spectral analysis. FIG. 17 is a graph of another Pupilogram of the average radius of the pupil per hour. FIG. 18 is an exemplary diagram of the pupil including the extracted pupil characteristic information.

Claims (2)

A system for confirming the identity of living animals by iris scanning, comprising the following elements:
(A) a control unit for receiving identification information identifying a living animal, and access by the control unit for including predetermined personal information relating to an identifiable organism for comparison with the identification information; Data storage unit containing the database to be created;
(B) an iris image pickup unit including a camera operated by a control unit to capture an iris image and generate an input image signal when the identification information corresponds to predetermined information; and (c) the input image A data processing unit for pre-processing the signal into processed data;
The storage unit comprises: (1) density and texture of the iris fiber structure using a frequency conversion method, (2) pupil reaction, (3) pupil shape, (4) autonomic nervous ring reaction, (5) autonomic nervous ring. A plurality of pre-stored iris identifications for an identifiable organism selected from the group consisting of: (6) the presence of a depression, (7) the position of the depression, and (8) the shape of the depression. Including at least one of the parameters related to the shape of the pupil,
The control unit may be operable to compare the processed data to the parameter to determine if there is a match indicating that identity has been confirmed; and
The iris identification system, wherein the data processing unit generates the processing data of the parameters related to the shape of the pupil, and determines the approximate center of the iris and then measures the iris using an algorithm for detecting the pupil end. A system for confirming the identity of living animals by iris scanning, comprising the following elements:
(A) a control unit for receiving identification information identifying a living animal, and access by the control unit for including predetermined personal information relating to an identifiable organism for comparison with the identification information; Data storage unit containing the database to be created;
(B) an iris image pickup unit including a camera operated by a control unit to capture an iris image and generate an input image signal when the identification information corresponds to predetermined information; and (c) the input image A data processing unit for pre-processing the signal into processed data;
The storage unit comprises: (1) density and texture of the iris fiber structure using a frequency conversion method, (2) pupil reaction, (3) pupil shape, (4) autonomic nervous ring reaction, (5) autonomic nervous ring. A plurality of pre-stored iris identifications for an identifiable organism selected from the group consisting of: (6) the presence of a depression, (7) the position of the depression, and (8) the shape of the depression. Including at least one of the parameters,
The control unit may be operable to compare the processed data to the parameter to determine if there is a match indicating that identity has been confirmed; and
The data processing unit generates processing data related to the parameter,
Regarding the parameters regarding the density and texture form of the iris fiber structure using the frequency conversion method, a zone suitable for iris image analysis is selected, and a selected still image signal of the selected zone is converted by a two-dimensional Wavelet transform. And measure,
For the parameters of the pupil reaction and the autonomic nervous ring reaction, the diameter of the pupil and the diameter of the autonomic nervous ring with time variation when light is applied to the pupil,
For parameters about the shape of the pupil, measured using an algorithm that detects the pupil edge after determining the approximate center of the iris,
For the parameters about the shape of the autonomic nervous ring, the two-dimensional image signal of the detected autonomic nervous ring is converted and measured by central axis conversion,
Regarding the parameters for the presence of the depression and the position of the depression, the zone in which the depression is present is detected.For the parameter for the shape of the depression, the two-dimensional image signal of the detected depression zone is transformed by central axis transformation. An iris identification system characterized by measuring.

Images