JP2017078937A - Personal authentication device, personal authentication method, and personal authentication program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a personal authentication method higher in practicality and flexibility than conventional.SOLUTION: The current invention is applied to equipment provided with a touch screen 20, for example a smart phone 10. According to the smart phone 10, in unlocking it, a flick request screen requesting flick operation is displayed to the touch screen 20. On the flick request screen, when an arrow 30 displayed in the center is flicked, a feature quantity showing a feature of an operator related to the flick operation is extracted. Then, with the feature quantity as an authentication key, authentication about whether the operator is a claimant having a legitimate right to operate the smart phone 10. Here, flick operation frequency necessary for authentication is appropriately selected, so that practicality including improvement of authentication accuracy is improved. Also, there is a possibility of application to equipment whose touch screen 20 is not a multi-touch screen, that is, there is high flexibility.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a personal authentication device, a personal authentication method, and a personal authentication program. In particular, the present invention is applied to a device equipped with a touch screen such as a smartphone or a tablet, and an operator who operates the device is authorized to operate the device. The present invention relates to a personal authentication device, a personal authentication method, and a personal authentication program for authenticating whether a right holder has a proper right.

  Some devices equipped with a touch screen such as a smartphone or a tablet are provided with a lock function that invalidates operations other than predetermined operations in order to prevent unauthorized operations by a third party. In order to unlock the lock function, the operator who operates the device needs to be authenticated that he / she has the right to operate the device (usually the device owner) As the authentication method, for example, there is a method of inputting a personal identification number (PIN), a password or a pattern. However, in these authentication methods, it is necessary to remember information that serves as an authentication key such as a personal identification number, password, or pattern registered by the right holder, that is, management of the authentication key. Further, when this authentication key is input, there is a possibility that it is stolen from a third party. Furthermore, there is a possibility that the authentication key may be overlooked by the trace of operation such as sebum remaining on the touch screen by the input of the authentication key. In addition, it has the disadvantage of being vulnerable to brute force attacks.

  In order to eliminate these drawbacks, a so-called biometric authentication method focusing on human physical characteristics has been put into practical use. One example is a fingerprint authentication method using, for example, a fingerprint as an authentication key. However, in this fingerprint authentication method, a special sensor for reading a fingerprint is necessary, and the price of the entire device including the sensor is increased accordingly. From another point of view, it cannot be applied to a device not equipped with the sensor, and in particular, cannot be additionally applied to an existing device not equipped with the sensor. Further, when the reading part of the sensor is dirty, when the finger to be read is dirty, or when the finger is excessively dry or dull, the authentication accuracy decreases. In addition, there is a case in which authentication (spoofing authentication) succeeds with a dummy fingerprint copied using a seal or an adhesive. In addition, since the physical characteristics such as fingerprints cannot be changed arbitrarily (or many times) unlike the above-described personal identification number, the authentication key based on such physical characteristics is copied even once. If this is the case, there is a fatal drawback that the safety (security) after that is not guaranteed.

  In addition to the fingerprint authentication method, for example, there is a face authentication method using a face as an authentication key. According to this face authentication method, especially when this is applied to a device equipped with a camera such as a smartphone, the camera is used as a sensor for copying the face. There is an advantage that it is not necessary to provide a special sensor. However, in this face authentication method, there is a possibility that authentication will be successful because the face picture of the right holder is directed to the camera. As a countermeasure, for example, there is a technique for requesting a predetermined operation such as blinking at the time of authentication, but in this case as well, authentication is successful because the moving image of the right holder who performs the predetermined operation is directed to the camera. there is a possibility. Furthermore, this face authentication method has a drawback that the accuracy of authentication decreases when the right holder wears sunglasses or the like, or when the surroundings are excessively dark or bright.

  In addition to this face authentication method, there is also an iris authentication method using an iris in the pupil as an authentication key. According to this iris authentication method, the time required for authentication can be shortened, and there is an advantage that it is difficult to duplicate the authentication key. However, this iris authentication method requires a special sensor (generally an infrared camera) for photographing the iris, as in the above-described fingerprint authentication method. Also, as with the face authentication method, it depends on the influence of the surrounding environment. There is a drawback that the authentication accuracy may be lowered.

  There is also a voiceprint authentication method using a voiceprint as an authentication key. According to this voiceprint authentication method, particularly when this voiceprint authentication method is applied to a device equipped with a microphone such as a smartphone, the microphone is used for detecting a voiceprint, so there is no need to provide a special sensor. . However, in this voiceprint authentication method, it is possible to duplicate the authentication key by recording the voice of the right holder, and the accuracy of the authentication may be lowered due to the influence of ambient noise or the condition of the right holder's throat. There is a drawback of being.

  By the way, as a biometric authentication method, in addition to the above-described physical features such as fingerprints, there are methods that focus on human behavioral features, that is, wrinkles. As a biometric authentication method paying attention to this human behavioral feature, there is a conventional one disclosed in Patent Document 1, for example. According to this prior art, a gesture operation is performed using at least three fingers on a touch screen (touch panel). Then, time-series position information of each touch point by this gesture operation is acquired from the touch screen. Further, based on the acquired position information, a feature amount pattern of an operator (unspecified user), that is, the feature amount pattern representing a feature related to a gesture operation using at least three fingers is calculated. . Then, authentication is performed based on whether or not the calculated feature amount pattern and the feature amount pattern of the right holder (user) registered in advance match each other.

  According to this prior art, since a feature related to a relatively simple action such as a gesture operation using at least three fingers is used as an authentication key, the burden on the right holder for storing the authentication key Is reduced. In addition, even if the authentication key is entered by a third party, it is very unlikely that the behavioral characteristics of the right holder included in the authentication key will be imitated. It is extremely difficult. Further, since the authentication key is input by a touch screen mounted on the device, it is not necessary to provide a special means for inputting the authentication key. In addition, the surrounding environment basically does not affect the authentication accuracy (so-called normal time). Therefore, according to this prior art, further improvement in safety is expected. In addition, this conventional technique can be easily applied to existing devices, and the application cost is low.

Japanese Patent Laying-Open No. 2015-118663

  However, in the above-described conventional technology, even if it is the right holder, gesture operation using at least three (in other words, three or more) fingers including the feature as an authentication key is always stable (feature). It is not easy to do this (so that the quantity patterns are substantially the same), which leads to an increase in the probability that the right holder himself will fail to authenticate, the so-called false rejection rate (FRR). And this increase in the rejection rate of the person is impeding practicality. In the prior art, the touch screen needs to be a so-called multi-touch screen capable of detecting each touch point by at least three fingers. Therefore, there is a problem that it cannot be applied to a device equipped with a touch screen other than a multi-touch screen, that is, the flexibility is low.

  Accordingly, an object of the present invention is to provide a novel personal authentication device, personal authentication method, and personal authentication program that are more practical and flexible than ever before.

  In order to achieve this object, the present invention includes a first invention related to a personal authentication device, a second invention related to a personal authentication method, and a third invention related to a personal authentication program.

  The first invention is a personal authentication device that is applied to a device equipped with a touch screen and authenticates whether an operator who operates the device is a right holder who has a right to operate the device. , Operation request means, feature quantity extraction means, and authentication means. Specifically, the operation request unit displays operation request information for requesting a predetermined operation using the touch screen on the touch screen. Then, the feature amount extraction unit performs predetermined processing by the operator based on operation information output from the touch screen in response to a predetermined operation performed by the operator according to the operation request information displayed on the touch screen. A feature amount representing a feature related to the operation is extracted. Further, the authentication means is based on the operator feature quantity that is the feature quantity extracted by the feature quantity extraction means and the right holder feature quantity that is a feature quantity related to a predetermined operation by the right holder registered in advance. Authenticate.

  According to the first aspect of the invention, the operation request information requesting to perform a predetermined operation using the touch screen is displayed on the touch screen by the operation request unit. In response to the display of the operation request information, when the operator performs a predetermined operation according to the operation request information, the operation information is output from the touch screen in response thereto. Then, the feature amount extraction unit extracts a feature amount representing a feature related to a predetermined operation by the operator based on the operation information. Further, the authentication means includes an operator feature amount that is a feature amount related to a predetermined operation by the operator extracted by the feature amount extraction means, and a feature amount that is related to the predetermined operation by a right holder registered in advance. Authentication is performed based on a certain right holder characteristic amount.

  That is, according to the first aspect of the present invention, the human behavioral feature related to a predetermined operation on the touch screen is used as the authentication key. Since operation request information requesting a predetermined operation for obtaining the authentication key is displayed on the touch screen, the right holder can determine what the predetermined operation is based on the operation request information displayed on the touch screen. Therefore, it is not necessary to memorize what the predetermined operation is. Further, the predetermined operation does not have to be a gesture operation using at least three fingers as in the above-described prior art. Therefore, for example, by adopting a simpler operation as the predetermined operation, the right holder can always perform the predetermined operation stably (in a substantially identical manner). Thereby, reduction of the person rejection rate is achieved. In addition, the probability that another person succeeds in authentication by adopting an appropriate operation as the predetermined operation, including the balance with the degree of simplicity of the predetermined operation, the so-called false acceptance rate (FAR). Is also reduced. In addition, unlike the prior art, the touch screen need not be a multi-touch screen.

  Note that the predetermined operation referred to here may be, for example, a flick operation (operation to quickly touch the finger after touching the finger on the touch screen). That is, the operation request information may be information requesting to perform a flick operation as the predetermined operation. This flick operation is an extremely simple operation (particularly compared to the above-described gesture operation using at least three fingers). On the other hand, the characteristics of the individual operators including the right holder are easily revealed. Therefore, it is very suitable as the predetermined operation. For example, when the device is a smartphone, the flick operation can be performed by one-hand input (an input mode in which the hand of the device and the hand used for the operation are the same) and two-hand input (the device handle and operation). Any input mode that is different from the hand provided) can be implemented, so that it is possible to select an input mode that is easy for the right holder to handle. This greatly contributes to the improvement of operability and also greatly improves the accuracy of authentication including the distinction between the right holder and the third party. Incidentally, in the above-described conventional technology, a gesture operation using at least three fingers is required at the time of authentication. However, such a gesture operation cannot be performed with one-hand input and is limited to two-hand input. For this reason, the operability is low, and the authentication accuracy cannot be improved due to the difference in input mode between the one-hand input and the two-hand input.

  The flick operation as the predetermined operation may be sequentially performed in a plurality of different directions. That is, the operation request information may be information requesting to sequentially perform flick operations in a plurality of different directions as a predetermined operation. Thus, when a flick operation in a plurality of directions, that is, various directions, is a predetermined operation, for example, when the predetermined operation is only one flick operation in a certain direction. In comparison, the characteristics as an authentication key are accurately captured, and as a result, the authentication accuracy is improved. The plurality of directions referred to here include, for example, upper, lower, left, and right directions of the touch screen. Of course, it is not limited to this.

  Further, the order of the flick operations in the plurality of directions may be random (irregular). Thus, by making the order of the flick operations in a plurality of directions random, for example, compared with the case where the order is fixed (regular), the adaptability of the operator including the right holder to the order is improved. In addition to the feature as an authentication key, this also contributes to the improvement of authentication accuracy. Of course, the order of the flick operations may be constant, but in this case, it is considered that the authentication accuracy is somewhat lowered as compared with the case where the order is random.

  In addition, the authentication unit may perform authentication using a statistical method. As this statistical method, there are a standardized Euclidean distance method and a spectrum filtering method, which will be described later.

  Then, when the operator is identified (determined) as the right holder by the authentication by the authenticating means, that is, when the authentication is successful, the right holder feature quantity updating means for adding the operator feature quantity as the right holder feature quantity , May be further provided. According to this configuration, the right holder feature quantity, which is a database of the right holder's feature quantity, is updated by the newly obtained right holder's feature quantity called the operator feature quantity. As a result, the database of rights holder features is enriched, which also greatly contributes to the improvement of authentication accuracy.

  The second invention relates to the personal authentication method as described above, and is a method invention corresponding to the first invention. That is, the second invention is applied to a device equipped with a touch screen, and is a personal authentication method for authenticating whether an operator who operates this device is a right holder who has a right to operate the device. , An operation request process, a feature quantity extraction process, and an authentication process. Specifically, in the operation request process, operation request information for requesting a predetermined operation using the touch screen is displayed on the touch screen. In the feature amount extraction process, a predetermined operation by the operator is performed based on operation information output from the touch screen in response to a predetermined operation performed by the operator according to the operation request information displayed on the touch screen. A feature amount representing a feature related to the operation is extracted. Further, in the authentication process, based on the operator feature quantity that is the feature quantity extracted in the feature quantity extraction process and the right holder feature quantity that is a feature quantity related to a predetermined operation by a right holder registered in advance. Authenticate.

  The third invention relates to the personal authentication program as described above, and is a program invention corresponding to the first invention. That is, the third invention is applied to a device equipped with a touch screen, and in a personal authentication program for authenticating whether an operator who operates this device is a right holder who has a right to operate the device. , Causing the computer to execute an operation request procedure, a feature amount extraction procedure, and an authentication procedure. Specifically, in the operation request procedure, operation request information for requesting a predetermined operation using the touch screen is displayed on the touch screen. In the feature amount extraction procedure, a predetermined operation by the operator based on operation information output from the touch screen in response to a predetermined operation performed by the operator according to the operation request information displayed on the touch screen. A feature amount representing a feature related to the operation is extracted. Further, in the authentication procedure, based on the operator feature amount that is the feature amount extracted by the feature amount extraction procedure and the right holder feature amount that is a feature amount related to a predetermined operation by the right holder registered in advance. Authenticate.

  As described above, according to the present invention, a human behavioral feature related to a predetermined operation using a touch screen is used as an authentication key. By adopting an appropriate operation as the predetermined operation, the rejection rate of the person can be reduced and the acceptance rate of others can be reduced. Furthermore, the touch screen does not have to be a multi-touch screen, unlike the prior art. Therefore, it is possible to provide a personal authentication method that is more practical and flexible than in the past.

It is the schematic external view which looked at the said smart phone including the flick request | requirement screen displayed on the touch screen of the smart phone which concerns on one Embodiment of this invention from the front. It is an illustration figure which shows the example of the flick request | requirement screen in the embodiment. It is an illustration figure which shows the example of the motion of the finger | toe used for the flick operation on the same flick request | requirement screen. It is a block diagram which shows schematic structure of the flick authentication apparatus in the embodiment. It is an illustration figure which shows the example of the flick data in the same embodiment. It is an illustration figure which shows the coordinate and determination frame which were set to the flick request | requirement screen in the embodiment. It is a figure which shows the list of the feature-values as an authentication key in the embodiment. It is an illustration figure for demonstrating the calculation point of the flick length as one of the same feature-value. It is an illustration figure for demonstrating the calculation point of the flick angle as one of the feature-values. It is a figure which shows an example of the calculation result of the feature-value. It is a flowchart which shows the operation | movement at the time of the registration mode of the flick authentication apparatus in the embodiment. It is a flowchart which shows the operation | movement at the time of the authentication mode of the same flick authentication apparatus. It is a graph which shows the experimental result about the principal rejection rate and others acceptance rate in the embodiment. It is a standard normal distribution graph for demonstrating another example in the embodiment. It is a graph which shows the experimental result by another example in the embodiment.

  One embodiment when the present invention is applied to a smartphone will be described with reference to FIGS.

  The smartphone 10 according to the present embodiment has the lock function described above. When the lock by the lock function is applied, an authentication flick request screen as shown in FIG. 1 for releasing the lock is displayed on the touch screen 20. In FIG. 1, a band-like portion 22 with a shaded pattern at the top of the touch screen 20 is a status bar for displaying the current time, the remaining battery capacity, and the like. A band-like portion 24 with a hatched pattern at the bottom of the touch screen 20 is a navigation bar on which appropriate software buttons such as a home button are arranged. Then, the flick request screen is displayed in a content area (also referred to as an effective view area) 26 occupying most of the touch screen 20 between the status bar 22 and the navigation bar 24.

  In the flick request screen shown in FIG. 1, an arrow 30 pointing in one of up, down, left and right directions is displayed at the center. In FIG. 1, a state in which the center arrow 30 points upward is illustrated. Then, another arrow 32 pointing upward is displayed above the central arrow 30 and in the vicinity of the upper edge of the content area 26. Further, another arrow 34 pointing downward is displayed below the central arrow 30 and in the vicinity of the lower edge of the content area 26. In addition, another arrow 36 pointing to the left is displayed on the left side of the center arrow 30 and in the vicinity of the left side edge of the content area 26. Further, another arrow 38 pointing to the right side is displayed on the right side of the center arrow 30 and in the vicinity of the right side edge of the content area 26.

  As described above, the center arrow 30 indicates one of up, down, left, and right directions. Strictly speaking, one of the four types of flick request screens shown in (a) to (d) of FIG. 2 is displayed on the touch screen 20 (content area 26) immediately after activation of the smartphone 10 or immediately after cancellation of the sleep state. Is displayed. Note that the flick request screen shown in FIG. 2A is the same screen as shown in FIG. In these flick request screens, the center arrow 30 is appropriately modified for differentiation from the other arrows 32 to 38. Further, the modification to the center arrow 30 differs depending on the direction indicated by the center arrow 30. For example, as shown in FIG. 2A (and FIG. 1), when the central arrow 30 points upward, the central arrow 30 is displayed in red. Then, as shown in FIG. 2B, when the center arrow 30 points downward, the center arrow 30 is displayed in blue. Further, as shown in FIG. 2C, when the center arrow 30 points leftward, the center arrow 30 is displayed in green. Then, as shown in FIG. 2D, when the center arrow 30 points to the right, the center arrow 30 is displayed in yellow. However, in the drawing, instead of the color of the center arrow 30, an appropriate halftone pattern is given.

  As described above, this flick request screen is for releasing the lock by the lock function, and more specifically, requests a flick operation in the direction indicated by the center arrow 30. For example, the flick request screen shown in FIG. 2A requests to flick the center arrow 30 upward, and at that time, the arrow 32 above becomes a kind of mark. Then, the flick request screen shown in FIG. 2B requests that the center arrow 30 be flicked downward, and at that time, the arrow 34 below serves as a mark. In addition, the flick request screen shown in FIG. 2C requests to flick the center arrow 30 to the left, and the arrow 36 on the left serves as a mark. Then, the flick request screen shown in FIG. 2D requests to flick the center arrow 30 to the right, and the arrow 38 on the right serves as a mark.

  In response to this so-called flick request, the operator flicks the center arrow 30 in the direction that it points. This flick operation is performed N (N: an integer of 1 or more) times in each of the up, down, left, and right directions. In other words, the flick request screens shown in FIGS. 2A to 2D are sequentially displayed N times.

  Note that the value of N is, for example, N = 4. Therefore, the flick operation is performed four times in the four directions of up, down, left, and right, that is, the flick operation is repeatedly performed 16 times in total. The direction of the flick operation over 16 times in total is random, and is determined based on, for example, a random number (pseudo-random number). In other words, each flick request screen shown in FIGS. 2A to 2D is displayed four times in a random order.

  Here, FIG. 3 shows the movement (trajectory) of a finger used for the flick operation when a total of 16 flick operations are performed by two operators as described above. In FIG. 3, a curved line 42 indicated by a thick solid line indicates a finger movement during an upward flick operation, and a curved line 44 indicated by a thick short broken line indicates a finger movement during a downward flick operation. A curve 46 indicated by a thick middle broken line indicates the finger movement during the flicking operation to the left, and a curve 48 indicated by a thick long broken line indicates the movement of the finger during the flicking operation to the right. As can be seen from FIG. 3, the movement of the finger during the flick operation is different for each operator. On the other hand, the finger movements of the same operator are similar in the vertical and horizontal directions. That is, there is a flaw in the movement of the finger during the flick operation for each operator. In FIG. 3, a square frame 40 indicated by a thin broken line is a determination frame for determining whether or not the flick operation is successful. Specifically, at the time of a certain flick operation, when the touch position of the first finger is within the determination frame 40 and the position where the finger is separated is outside the determination frame 40, the flick is performed. The operation is considered successful.

  In this embodiment, paying attention to the fact that there is a flaw in the movement of the finger during the flick operation for each operator as described above, the information related to the flaw in the movement of the finger is used as an authentication key, and the lock is released. Authentication is performed.

  In order to realize this so-called flick authentication, in this embodiment, first, registration of information relating to the movement of the finger used for the flick operation of the right holder who has the right to operate the smartphone 10, that is, profile data, is performed. Is called. Specifically, a registration flick request screen similar to that shown in FIG. 2 is displayed on the touch screen 20 in the registration mode. Then, on this registration flick request screen, a total of 16 (= 4N) flick operations are performed in the same manner as described above, and further, the total of 16 flick operations are α (α; an integer of 1 or more) ) Repeatedly. In other words, the total of 16 flick operations are set as one set, and this one set of flick operations is repeated α times. Here, the number of repetitions α is, for example, α = 10. That is, one set of flick operations is repeatedly performed 10 times, that is, a total of 160 (= 4N × α) flick operations are performed. Then, based on data output from the touch screen 20 (strictly speaking, an output circuit (not shown) of the touch screen 20) during the flick operation for a total of 160 times, that is, flick data, the right person's finger related to the flick operation is referred to. A feature value obtained by quantifying the motion feature is extracted. The feature amount data, that is, feature data is registered as profile data.

  In this way, the flick authentication is performed after the profile data is registered. Specifically, the authentication flick request screen shown in FIG. 2 is displayed in the authentication mode. Then, a total of 16 flick operations are performed on the authentication flick request screen as described above, that is, one set of flick operations is performed. Then, based on the flick data obtained from the touch screen 20 at the time of the one set of flick operations, a feature value obtained by digitizing the feature of the finger movement related to the flick operation of the operator who is the person to be authenticated is extracted. . The feature amount data, that is, feature data is used as test data, and authentication is performed based on the test data and the profile data of the right holder registered in advance. For example, when the operator is identified as the right holder by this authentication, that is, when the authentication is successful, the lock is released. If not, that is, if authentication fails, the lock is not released and the display of the authentication flick request screen is maintained. It is important that the right holder performs the flick operation with the same input mode (one-hand input or two-hand input) and the same finger in the registration of the profile data in the registration mode and in the authentication in the authentication mode.

  This flick authentication is realized by starting a flick authentication program as application software installed in the smartphone 10. Conceptually, when this flick authentication program is activated, a flick authentication device 50 as shown in FIG. 4 is configured.

  As shown in FIG. 4, the flick authentication device 50 includes a data collection unit 52. The data collection unit 52 is linked to the touch screen 20 and displays the above-described flick request screen for registration on the touch screen 20 in the registration mode. When a total of 160 (= 4N × α) flick operations, that is, 10 (= α) flick operations described above are performed on the registration flick request screen, the data collection unit 52 Flick data output from the touch screen 20 in response to the flick operation is acquired and input to the feature quantity extraction unit 54 as a feature quantity extraction unit. Based on the flick data corresponding to a total of 160 times (for 10 sets) of flick operations input from the data collection unit 52, the feature amount extraction unit 54 described above related to the right holder who is the current operator. Extract the amount. The profile data, which is the extracted feature value data, is stored in the database 56 as storage means, that is, registered.

  On the other hand, in the authentication mode, the data collection unit 52 causes the touch screen 20 to display the above-described authentication flick request screen. Then, when a total of 16 (= 4N) flick operations described above, that is, one set of flick operations are performed on the authentication flick request screen, the data collection unit 52 responds to the flick operation. Flick data output from the touch screen 20 is acquired and input to the feature amount extraction unit 54. Based on the flick data, the feature amount extraction unit 54 extracts a feature amount relating to the current operator who is the person to be authenticated. The test data, which is the extracted feature quantity data, is input to the scoring unit 58 as a scoring means. The scoring unit 58 includes the test data and profile data stored in the above-described database 56, specifically, a profile data group consisting of profile data for M (M; integers greater than α) pairs described later. Based on this, scoring (scoring) how close the test data is to the profile data group, so to speak, the operator's rightness is numerically expressed. The digitized data, that is, the score value, is input to the authentication unit 60 as authentication execution means. The authentication unit 60 performs authentication based on the score value input from the scoring unit 58. In this authorization, for example, when the current operator is identified as the right holder, the fact that the authentication has succeeded is notified to the lock release unit 62 as the lock release means. In response to this, the lock release unit 62 releases the lock. On the other hand, when the current operator is not identified as the right holder, the data collection unit 52 is notified that the authorization has failed. In response to this, the data collection unit 52 maintains the display of the flick request screen for authentication. In other words, the locked state is maintained without being released.

  More specifically, as shown in FIG. 5, flick data (so-called raw data) output from the touch screen 20 is indicated by UNIX (registered trademark) time, the number of successful flick operations, and a flick request. Flick direction, finger touch state on the touch screen 20, x-coordinate value of the finger touch position on the touch screen 20 (content area 26), y-coordinate value of the touch position, finger pressing pressure at the touch position, It contains a total of 8 types of information separated by commas, that is, the finger contact area. These pieces of information are all obtained from the smartphone 10 as standard. The flick data shown in FIG. 5 is for a smartphone “Xperia (registered trademark) _Z_SO-02E” manufactured by Sony Mobile Communications Inc., and Android (registered trademark) manufactured by Google, which is the OS (Operating System) thereof. ) Is obtained at a time interval of about 20 ms according to the specification (however, this time interval slightly increases or decreases due to the influence of interrupt processing, etc.). Data obtained at each time interval, that is, data for one row in FIG. 5 is called event data.

  Of the flick data (event data), especially in the flick direction, “u”, “d”, “l” representing upper (up), lower (down), left (left), and right (right). And the symbol “r” is selectively applied. For the touch state, the symbols “p”, “m”, and “r” representing the start (press), the movement of the touch position (move), and the end of the touch (release) are selectively applied. . Event data from the touch state from “p” to “m” through “m” is flick data corresponding to one flick operation. For example, a total of seven event data in the first to seventh rows in FIG. 5 is the flick data corresponding to the first flick operation, and a total of five event data in the eighth to twelfth rows is the second. This is flick data corresponding to the flick operation. Further, the x-coordinate value and the y-coordinate value follow xy orthogonal coordinates set in the content area 26 as shown in FIG. In the xy orthogonal coordinates, the upper left corner of the content area 26 is the origin o. The horizontal axis passing through the origin o is the x axis, and the vertical axis passing through the origin o is the y axis. In addition, the coordinate value in the xy orthogonal coordinates follows the number of pixels in the content area 26. Here, the number of pixels W in the horizontal direction of the content area 26 is W = 1080, and the number of pixels H in the vertical direction of the content area 26 is H = 1700. Incidentally, the dimension of one side of the determination frame 40 described above is set to about ¼ to ３ of the number of pixels W in the horizontal direction of the content area 26.

  As described above, the flick data is input from the data collection unit 52 to the feature amount extraction unit 54. Based on this flick data, the feature quantity extraction unit 54 uses the seven feature quantities X [dir (i)], Y [dir (i)], L [ dir (i)], V [dir (i)], θ [dir (i)], P [dir (i)] and S [dir (i)] are calculated. Of these, the feature quantity X [dir (i)] is the position on the x coordinate of the first touch position in each flick operation, strictly speaking, from the center of the first touch position on the x coordinate. Deviation (deviation). The X coordinate X [dir (i)] is obtained by the following equation (1). Here, dir represents the flick direction indicated by the flick request, and this dir has the symbols “u”, “d”, “l” and “r” as in the flick direction shown in FIG. Is applied selectively. Dir (i) represents the i (i; integer of 1 to N (= 4)) flick operation to dir among the above-described one set (16 times in total) of flick operations. Further, x [dir (i), j] is the x coordinate value of the j (j; integer of 1 to m [dir (i)])-th event data in the flick operation of dir (i). Note that m [dir (i)] is the total number of event data in the flick operation of dir (i).

  According to this equation 1, for example, when the value of the X coordinate X [dir (i)] based on the equation 1 is a negative number (X [dir (i)] <0), the first touch position is on the x coordinate. Means to the left of its center in When the value of the X coordinate X [dir (i)] based on Equation 1 is 0 (X [dir (i)] = 0), the first touch position is at the center on the x coordinate. Means. Further, when the value of the X coordinate X [dir (i)] based on the number 1 is a positive number (X [dir (i)]> 0), the first touch position is more than the center on the x coordinate. Means to the right.

  The feature quantity Y [dir (i)] in FIG. 7 is the position on the y coordinate of the first touch position in each flick operation, strictly speaking, the center of the first touch position on the y coordinate. It is a deviation from. The Y coordinate Y [dir (i)] is obtained by the following equation 2. In Equation 2, y [dir (i), j] is the y coordinate value of the jth event data in the fir operation of dir (i).

  According to Equation 2, when the value of Y coordinate Y [dir (i)] based on Equation 2 is a negative number (Y [dir (i)] <0), the first touch position is on the y coordinate. Means above the center of When the value of the Y coordinate Y [dir (i)] based on Equation 2 is 0 (Y [dir (i)] = 0), the first touch position is at the center on the y coordinate. Means. Further, when the value of the Y coordinate Y [dir (i)] based on the equation 2 is a positive number (Y [dir (i)]> 0), the first touch position is more than the center on the y coordinate. Means that it is below.

  Further, the feature quantity L [dir (i)] in FIG. 7 is the flick length, and more specifically, the length of the locus of the touch position from the start to the end of each flick operation. The flick length L [dir (i)] is obtained by the following equation (3).

  According to Equation 3, as shown in FIG. 8, the total Euclidean distance between the two adjacent touch positions 70 and 70 among the touch positions 70, 70,. The flick length is L [dir (i)].

  The feature quantity V [dir (i)] in FIG. 7 is the flick speed, and more specifically, the moving speed of the touch position from the start to the end of each flick operation. The flick speed V [dir (i)] is obtained by the following equation (4). Here, t [dir (i), j] is the UNIX (registered trademark) time of the jth event data in the flick operation of dir (i).

  Further, the feature quantity θ [dir (i)] in FIG. 7 is a flick angle, and more specifically, an angle formed by the direction of the actual flick operation based on the flick direction dir instructed by each flick request ( Direction), and the calculation formula differs depending on the flick direction dir. For example, when the flick direction dir is upward, the flick angle θ [u (i)] is calculated by Equation 5. When the flick direction dir is downward, the flick angle θ [d (i)] is calculated by Equation 6. Further, when the flick direction dir is to the left, the flick angle θ [l (i)] is calculated by Equation 7. When the flick direction dir is rightward, the flick angle θ [r (i)] is calculated by Equation 8.

  According to these equations 5 to 8, the flick angles θ [u (i)], θ [d (i)], θ [l (i)] and θ [r (i) based on the equations 5 to 8 are used. )]) Takes a positive value in the clockwise direction with each flick direction dir as a reference, and takes a negative value in a counterclockwise direction with the flick direction dir as a reference, as shown in FIG.

  Further, the feature quantity P [dir (i)] in FIG. 7 is an average value of the finger pressing pressure from the start to the end of each flick operation, and is obtained by the following equation (9). Here, p [dir (i), j] is the pressing pressure of the jth event data in the flick operation of dir (i).

  The feature quantity S [dir (i)] in FIG. 7 is an average value of the finger contact area from the start to the end of each flick operation, and is obtained by the following equation (10). Here, s [dir (i), j] is a contact area of the jth event data in the flick operation of dir (i).

  As an example, each feature amount X [u (1)], which is calculated based on the flick data corresponding to the first flick operation composed of a total of seven event data in the first to seventh rows in FIG. Y [u (1)], L [u (1)], V [u (1)], θ [u (1)], P [u (1)] and S [u (1)] 10 shows.

  Further, the feature quantity extraction unit 54 includes each feature quantity X [dir (i)], Y [dir (i)], L [dir (i)], V [dir (i)], θ [dir (i). ], P [dir (i)], and S [dir (i)], the symmetry of the flick operation in the vertical direction is digitized, and the symmetry of the flick operation in the horizontal direction is digitized. For example, each feature amount X [dir (i)], Y [dir (i)], L [dir (i)], V [dir (i)], θ using a code F [dir (i)] If [dir (i)], P [dir (i)] and S [dir (i)] are expressed individually, the symmetry F [ud (i)] of the flick operation in the vertical direction is given by 11 is obtained. Then, the symmetry F [l-r (i)] of the flick operation in the left-right direction is obtained by Expression 12.

  Then, each of the above-described seven feature amounts X [dir (i)], Y [including the symmetry F [ud (i)] and F [1-r (i)] based on the equations 11 and 12. dir (i)], L [dir (i)], V [dir (i)], θ [dir (i)], P [dir (i)] and S [dir (i)], ie, features The data is stored in the database 56 as profile data as described above, or input to the scoring unit 58 as test data.

  The scoring unit 58 performs the following processing for each set described above for the set of profile data stored in the database unit 56, that is, the profile data group. Note that the number M of sets of profile data constituting the profile data group is M = α (= 10) immediately after the registration of the profile data in the registration mode described above.

  That is, the scoring unit 58 includes the feature amounts X [dir (i)], Y [dir (i)], L [dir (i)], V [dir (i)], θ [ For each of dir (i)], P [dir (i)] and S [dir (i)], the averaging process is performed according to the following equation (13). Here, k is the number of the profile data set in the profile data group (1 ≦ k ≦ M). Also in this equation 13, each feature quantity X [dir (i)], Y [dir (i)], L [dir (i)], V [dir () is used by using the code F [dir, k]. i)], θ [dir (i)], P [dir (i)] and S [dir (i)] are averaged X [dir, k], Y [dir, k], L [dir, k ], V [dir, k], θ [dir, k], P [dir, k] and S [dir, k] are individually expressed.

  Further, the scoring unit 58 performs a normalization process according to the following expression 14 on all the feature values F [dir, k] after the averaging process according to the expression 13, that is, normalization (non-dimensionalization). Here, max {F [dir, q]} is the maximum value of the feature amount F [dir, k] in all the sets q (1 ≦ q ≦ M). Min {F [dir, q]} is the minimum value of the feature value F [dir, k] in all the sets q.

  The feature quantity nF [dir, k] after the normalization processing according to the equation 14 includes seven types (nX [dir, k], nY [dir, k], nL [dir, k], nV [dir, k], nθ [dir, k], nP [dir, k] and nS [dir, k]), and the flick direction dir includes six directions (u, d, l, r, ud and 1−r), one set of profile data is a 42 (= 7 × 6) -dimensional vector vF expressed by the following equation (15).

  In the same manner as this, the scoring unit 58 also performs the averaging process according to the above-described equation 13 for the test data, and further performs the normalization process according to the equation 14. The test data after this normalization process is also a 42-dimensional vector vF expressed by Equation 15.

  After that, the scoring unit 58 sets the vector of the test data as vFt and the vector of the profile data of the kth set as vFp [k], and based on both of these vectors vFt and vFp [k ], The normalized (normalized) Euclidean distance NED (vFt, vFp [k]) is obtained.

  The standardized Euclidean distance NED (vFt, vFp [k]) based on this equation 16 is small when the operator providing the test data and the right holder providing the profile data are the same person, and otherwise. Tend to grow.

  The scoring unit 58 uses the standardized Euclidean distance NED (based on the equation 16 between the vector vFp [k] of all sets (that is, M) of profile data and the vector vFt of the test data. vFt, vFp [k]). Then, a test average distance NEDt that is an average of them is obtained. This test average distance NEDt is input to the authentication unit 60 as the above-described score value.

  At the same time, the scoring unit 58 combines the vectors vFp [k] of all sets of profile data by two pieces in total, and based on the following Equation 17 based on Equation 16, the two vectors vFp [ The normalized Euclidean distance NED (vFp [k], vFp [k ′]) between k] and vFp [k ′] is obtained. Note that k ′ represents a set different from k (k ′ ≠ k).

Then, the scoring unit 58 obtains a profile average distance NEDp that is an average of the _{M 2} standardized Euclidean distances NED (vFp [k], vFp [k ′]) based on the equation (17). This profile average distance NEDp is also input to the authentication unit 60, as is the test average distance NEDt described above.

  The authentication unit 60 includes a test average distance NEDt as a score value input from the scoring unit 58, a profile average distance NEDp as a reference value input from the scoring unit 58, and a preset threshold value Th [NED] is used to determine whether the following equation 18 is satisfied.

  Here, for example, when the number 18 is satisfied, the authentication unit 60 identifies the operator as the right holder, that is, indicates that the authentication has been successful, and notifies the lock release unit 62 accordingly. In response to this, the lock release unit 62 releases the lock. On the other hand, if the number 18 is not satisfied, the authentication unit 62 determines that the operator is a third party different from the right holder, that is, the authentication has failed, and the data collection unit Tell 52. Thereby, the display of the above-mentioned flick request screen is maintained without unlocking.

  If authentication is successful, the test data at that time is stored in the database 56 as new profile data. That is, the profile data group stored in the database 56 is updated. By updating the profile data group, the total number M of sets of profile data constituting the profile data group is increased. When the total number M of the set of profile data reaches a predetermined maximum value M ′, the oldest profile data is discarded from the profile data group (database 56) every time new profile data is added thereafter. . In this way, the profile data group in the database 56 is sequentially updated, thereby enriching the profile data group. This greatly contributes to the improvement of authentication accuracy.

  When this series of processing is viewed as processing of the entire flick authentication device 50, the flick authentication device 50 operates according to the procedure shown in the flowchart of FIG.

  That is, when a predetermined operation for entering the registration mode is performed, the flick authentication device 50 proceeds to step S1 and determines a direction for requesting the flick operation. This flick direction is determined based on random numbers as described above.

  After executing step S1, the flick authentication device 50 proceeds to step S3, and displays a registration flick request screen for requesting a flick operation in the direction determined in step S1 on the touch screen 20. Then, the flick authentication device 50 proceeds to step S5.

  In step S5, the flick authentication device 50 determines whether a flick operation has been performed on the display of the flick request screen. When the flick operation is performed, the flick authentication device 50 proceeds to step S7.

  In step S7, the flick authentication device 50 determines whether the flick operation in the previous step S5 is successful. This determination is performed using the determination frame 40 as described above. Here, for example, when the flick operation is successful, the flick authentication device 50 proceeds to step S9, and when the flick operation is unsuccessful, the process returns to step S3.

  In step S9, the flick authentication device 50 temporarily stores flick data obtained from the touch screen 20 in response to the flick operation. In order to temporarily store the flick data, a temporary storage unit such as a memory (not shown) is provided. Then, the flick authentication device 50 proceeds to step S11, and whether or not the conditions necessary for registration of the profile data in this registration mode are satisfied, specifically, the above-mentioned 160 (= 4N × α) flick operations are performed. It is determined whether or not the flick data corresponding to the 160 flick operations has been obtained. Here, for example, when the necessary condition is not satisfied, the process returns to step S1. On the other hand, if the necessary condition is satisfied, the process proceeds to step S13.

  In step S13, the flick authentication device 50 extracts the feature amount of the right holder who is the current operator based on the flick data temporarily stored in the previous step S9. In step S15, the feature amount data extracted in step S13 is registered as profile data. Then, the process proceeds to step S17, and a predetermined message (not shown) indicating that the registration is completed is displayed on the touch screen 20. At the same time, the touch screen 20 displays an operable button on which, for example, “OK” is written to prompt the right holder, who is the current operator, to confirm that the registration has been completed. The

  After executing step S17, the flick authentication device 50 proceeds to step S19, and determines whether or not the above-mentioned “OK” button displayed on the touch screen 20 in step S17 has been tapped (touched). When the tap operation on the “OK” button is performed, the flick authentication device 50 ends the registration mode.

  In contrast to the operation in the registration mode, the flick authentication device 50 operates in the authentication mode according to the procedure shown in the flowchart of FIG.

  That is, when the smartphone 10 is activated or the sleep state is released, the flick authentication device 50 proceeds to step S101 and determines a direction in which a flick operation is requested. This flick direction is determined based on random numbers as described above.

  After executing step S101, the flick authentication device 50 proceeds to step S103, and causes the touch screen 20 to display an authentication flick request screen for requesting a flick operation in the direction determined in step S101. Then, the flick authentication device 50 proceeds to step S105.

  In step S105, the flick authentication device 50 determines whether a flick operation has been performed on the display of the flick request screen. Here, for example, when the flick operation is not performed, the flick authentication device 50 proceeds to step S107 and determines whether or not a predetermined time as a waiting time has elapsed. And when this predetermined time has not passed, it returns to step S105.

  On the other hand, when the flick operation is performed in step S105, the flick authentication device 50 proceeds to step S109 and determines whether or not the flick operation is successful. Here, for example, when the flick operation is successful, the flick authentication device 50 proceeds to step S111, and when the flick operation is unsuccessful, the process returns to step S103.

  In step S111, the flick authentication device 50 temporarily stores flick data obtained from the touch screen 20 in response to the flick operation. Then, the flick authentication device 50 proceeds to step S113, and whether or not the conditions necessary for the registration of the profile data in this registration mode are satisfied, specifically, the total 16 (= 4N × α) flick operations described above are performed. It is determined whether it is successful, that is, whether flick data corresponding to the 16 flick operations has been obtained. Here, for example, when the necessary condition is not satisfied, the process returns to step S101. On the other hand, if the necessary condition is satisfied, the process proceeds to step S115.

  In step S115, the flick authentication device 50 extracts the feature amount of the person to be authenticated who is the current operator based on the flick data temporarily stored in the previous step S111. The extracted feature data is used as test data.

  Further, the flick authentication device 50 proceeds to step S117, performs scoring based on the test data and pre-registered profile data, that is, calculates a score value. Then, the flick authentication device 50 proceeds to step S119 and performs authentication based on the score value. Then, the flick authentication device 50 proceeds to step S121 and determines the result of the authentication.

  In this step S121, for example, when it is determined (identified) that the operator is the right holder, that is, when the authentication is successful, the flick authentication device 50 proceeds to step S123 and indicates that the authentication has succeeded (not shown). Is displayed on the touch screen 20. The display time of this authentication success message is, for example, about several seconds.

  Then, the flick authentication device 50 proceeds to step S125, adds test data as new profile data, that is, updates the profile data group. Then, the flick authentication device 50 proceeds to step S127, releases the lock, and ends this authentication mode.

  If the predetermined time has elapsed in step S107 described above, the flick authentication device 50 proceeds from step S107 to step S129. Further, when it is determined in step S121 described above that the operator is not the right holder, that is, when the authentication fails, the flick authentication device 50 proceeds from step S121 to step S129. In step S129, a predetermined message (not shown) indicating that the authentication has failed is displayed on the touch screen 20, and the authentication mode is terminated. The display time of the authentication failure message in step S129 is also appropriate, for example, about several seconds.

  In order to confirm the effectiveness of the flick authentication method according to the present embodiment, the following experiment was performed.

  That is, for 10 subjects (male; 8 people, 2 women), the above-mentioned flick data for one set (16 (= 4N) times in total) per person, that is, a total of 160 (= 4N × α) Collect in batches. When collecting the flick data, two input modes, one-handed input and two-handed input, are envisaged, but they were arbitrary for each subject (one-handed input; eight persons, two-handed input: two persons). Then, the rejection rate (FRR) and the acceptance rate (FAR) of others were evaluated by using a known leave-one-out cross-validation (LOOCV) method. The identity rejection rate is obtained by the following equation 19, and the other person acceptance rate is obtained by the equation 20.

  The results of this experiment are shown graphically in FIG. As can be seen from the graph of FIG. 13, the smaller the above-mentioned threshold Th [NED] used for authentication, the lower the acceptance rate of others, that is, the prevention of intrusion of others, but on the other hand, the rejection rate of identity is high. In other words, there is a high possibility that the person will fail to authenticate. In contrast, the greater the threshold Th [NED], the less likely the person will fail to authenticate, but on the other hand, it is easier to allow others to enter. That is, the principal rejection rate and the stranger acceptance rate vary depending on the threshold value Th [NED] and have a so-called trade-off relationship. This is a tendency commonly seen in biometric authentication methods. In addition, when an equivalent error rate (EER) in which the principal rejection rate and the other person acceptance rate are equivalent by changing the threshold Th [NED] is seen, this equivalent error rate is about 1.5% (Th [NED] ≈1.3), confirming high authentication accuracy. This equivalent error rate of about 1.5% is a value that can be sufficiently put into practical use as a so-called security measure.

  As described above, according to the present embodiment, it is possible to take security measures that can be sufficiently put into practical use. Further, the touch screen 10 does not need to be a multi-touch screen unlike the above-described conventional technology, and therefore can be applied to existing devices with higher flexibility than the conventional technology. In addition, since the flick operation used for authentication may be either one-handed input or two-handed input, the operability at the time of authentication is improved as compared with the conventional technology limited to two-handed input, It is also possible to improve the authentication accuracy including differentiation between the right holder and the third party.

  Note that this embodiment is one specific example of the present invention and does not limit the scope of the present invention. That is, the present invention may be realized by forms other than the present embodiment.

  For example, the so-called standardized Euclidean distance method using the standardized Euclidean distance has been adopted as the scoring method described above, but is not limited thereto. For example, the spectrum filtering method described below may be employed together.

  In this spectrum filtering method, if the same person, the feature amounts X [dir (i)], Y [dir (i)], L [dir () included in the above-described feature data (profile data and test data) are used. i)], V [dir (i)], θ [dir (i)], P [dir (i)] and S [dir (i)] are assumed to follow a normal distribution. Under this assumption, using the profile data group that is a set of profile data, each feature amount X [dir (i)], Y [dir (i)], L [dir (i)], V [ Standard normal distributions of dir (i)], θ [dir (i)], P [dir (i)], and S [dir (i)] are created.

  Specifically, first, among the profile data constituting the profile data group, a certain feature amount corresponding to the i-th flick operation included in the k-th set of profile data is represented by F [dir (i), k]. Then, an average value μ [F [dir]] of the feature amount F [dir (i), k] is obtained based on the following equation (21).

  Then, based on the following equation 22, the standard deviation σ [F [dir]] of the feature amount F [dir (i), k] is obtained.

  Further, a standardized feature quantity Z [dir (i), k] is obtained based on the following equation (23).

  The probability density A [Z [dir (i), k]] with the standardized feature quantity Z [dir (i), k] as a variable is expressed by the following equation (24). The probability density A [Z [dir (i), k]] based on the equation 24 ideally draws a graph according to the standard normal distribution shown in FIG.

  Here, it is assumed that the standardized feature quantity Z [dir (i)] is obtained for the test data in the same manner. For example, when the test data is that of the right holder, the probability density A [Z [dir (i)] having the standardized feature quantity Z [dir (i)] included in the test data as a variable is used. ] Is presumed to follow a standard normal distribution. On the other hand, if the test data is not that of the right holder, the probability density A [Z [dir (i)]] is assumed to deviate from the standard normal distribution to some extent without following the standard normal distribution. The That is, it is expected that it is possible to determine whether or not the test data belongs to the right holder by using the standard normal distribution created based on the profile data group. Since this method is an aspect of filtering test data using a standard normal distribution created based on the profile data group as a kind of filter, it is called a spectrum filtering method as described above.

  However, with respect to the standardized feature quantity Z [dir (i), k] of the profile data based on the above equation 23, it seems that some rights holders may not follow the standard normal distribution. . It is not appropriate that such a feature quantity index including the feature quantity Z [dir (i), k] is adopted as an element for discrimination (scoring) by the spectrum filtering method. The feature quantity index is excluded from the elements for the determination. For example, in the ideal standard normal distribution shown in FIG. 14, the probability that the absolute value | Z | of the variable Z is | Z | ≦ 3σ is about 99.73%. , The absolute value | Z [dir (i), k] | of the feature quantity Z [dir (i), k] based on Equation 23 is | Z [dir (i), k] | ≦ 3σ. A ratio of the number of [dir (i), k] is obtained, and only a feature quantity index having this ratio of 99.73% or more is adopted as an element for discrimination by the spectrum filtering method. Specifically, the number of feature quantities Z [dir (i), k] included in a certain feature quantity index is NM. Of the NM feature quantities Z [dir (i), k], the number D of the feature quantities Z [dir (i), k] satisfying | Z [dir (i), k] | ≦ 3σ. Is employed as an element for discrimination by the spectrum filtering method.

  In addition, the standardized feature quantity Z [dir (i)] is also obtained for the test data as described above. Then, the standardized feature quantity Z [dir (i)] of the test data follows the standard normal distribution created based on the data that satisfies | Z [dir (i)] |> 3σ, that is, the profile data group. The number Q of no data is counted. If the number of feature quantity indexes employed as an element for discrimination by the spectrum filtering method is R (0 ≦ R ≦ 42), the count value Q is 0 ≦ Q ≦ NR.

  This count value Q is a small value when the test data relates to the right holder. On the other hand, when the test data is related to a third party other than the right holder, the count value Q is a large value. Therefore, the count value Q is one of the score values, and is compared with a preset threshold value Th [SF] as shown in the following equation (25).

  If this number 25 is satisfied, it is determined that the operator as the person to be authenticated is the right holder. On the other hand, when Expression 25 is not satisfied, it is determined that the operator is a third party who is not the right holder.

  In the experiment described with reference to FIG. 13 described above, an experiment using this spectrum filtering method was further performed. The result is shown in FIG. As can be seen from FIG. 15, by using the spectrum filtering method in addition to the standardized Euclidean method, the acceptance rate (FAR) of others can be reduced. In particular, the smaller the threshold Th [SF], the lower the other person acceptance rate. In this experiment, the cross-validation method without one piece is used as described above, and the test data is included in the profile data group as the profile data. Therefore, the rejection rate (FRR) of the person is determined by the spectrum filtering method. Not affected. According to this experiment, the equivalent error rate (EER) is about 1.0% (Th [NED] ≈1.55), and it has been confirmed that the authentication accuracy can be further improved. In actual operation, which is different from this experiment, due to the influence of the spectrum filtering method, the rejection rate is reduced, and the equivalent error rate is further reduced, that is, the authentication accuracy is further improved. Expected.

  Of course, not only the spectrum filtering method but also the standardized Euclidean method, other statistical methods such as likelihood, standardized Manhattan distance, and correlation analysis may be used.

  In the present embodiment, the seven feature quantities X [dir (i)], Y [dir (i)], L [dir (i)], V [dir (i)], θ shown in FIG. Although [dir (i)], P [dir (i)] and S [dir (i)] are extracted, the present invention is not limited to this. For example, the degree of bending of the locus of the touch position is obtained, and the so-called flick curvature may be used as the feature amount. Moreover, since the smart phone 10 is also provided with an acceleration sensor, an inclination sensor, etc., a suitable feature-value may be extracted from these outputs.

  Further, the order in the flick direction is random, but may be constant. However, since the order in the flick direction is random, the adaptability of the operator including the right holder to the order is more characteristic feature X [dir (i)] than in the case where the order is constant. , Y [dir (i)], L [dir (i)], V [dir (i)], θ [dir (i)], P [dir (i)] and S [dir (i)] It is speculated that this also contributes to the improvement of authentication accuracy.

  Moreover, although this embodiment demonstrated the case where this invention was applied to the smart phone 10 which uses Android as OS, it does not restrict to this. The present invention can be applied to an Apple smartphone “iPhone” (registered trademark), and can also be applied to devices other than the smartphone such as a tablet. In short, the present invention can be applied to a device having a touch screen. It is.

  And in this embodiment, although the flick authentication method which used the behavioral characteristic which concerns on a flick operation as an authentication key was proposed, the behavioral characteristic which concerns on the said flick operation, such as a tap operation and a swipe operation (operation which traces on a screen) May be used as an authentication key. The flick operation is an extremely simple operation (particularly compared to the gesture operation using at least three fingers in the above-described prior art). On the other hand, the characteristics of the individual operator including the right holder are easily revealed. Therefore, it is extremely suitable as an operation for personal authentication. Further, these flick operations, tap operations, swipe operations, etc. may be combined as appropriate.

  In addition, the value of the number of repetitions N of the flick operations in the above-described one set of flick operations is not limited to N = 4 and may be other values. Also, the value of the number of repetitions α of the set of flick operations in the registration mode is not limited to α = 10, and may be other values. Authentication accuracy can be controlled by appropriately determining the values of the number of repetitions N and α. Furthermore, since the accuracy of authentication varies depending on the number M of sets of profile data constituting the profile data group, it is important that the number M of sets is appropriately determined according to the situation.

DESCRIPTION OF SYMBOLS 10 Modeling apparatus 20 Control part 30 Memory | storage part 100 Cell space 110 Cell 120 Partition

Claims (8)

In a personal authentication device that is applied to a device equipped with a touch screen and authenticates whether an operator operating the device is a right holder having a legitimate right to operate the device,
Operation request means for displaying operation request information for requesting to perform a predetermined operation on the touch screen on the touch screen;
In accordance with the operation request information, a feature amount representing a feature related to the predetermined operation by the operator is extracted based on operation information output from the touch screen in response to the predetermined operation performed by the operator. Feature quantity extraction means for
The authentication is performed based on the operator feature amount that is the feature amount extracted by the feature amount extraction unit and the right holder feature amount that is the feature amount related to the predetermined operation by the right holder registered in advance. Authentication means to perform;
A personal authentication device comprising: The predetermined operation includes a flick operation,
The personal authentication device according to claim 1. The flick operation is sequentially performed in a plurality of different directions.
The personal authentication device according to claim 2. The order of the flick operations in the plurality of directions is random.
The personal authentication device according to claim 3. The authentication means performs the authentication using a statistical method.
The personal authentication device according to claim 1. A right holder feature amount updating means for adding the operator feature amount as the right holder feature amount when the operator is identified as the right holder by the authentication;
The personal authentication device according to claim 1. In a personal authentication method that is applied to a device equipped with a touch screen and authenticates whether an operator operating the device is a right holder having a legitimate right to operate the device,
An operation request process for displaying operation request information for requesting to perform a predetermined operation on the touch screen on the touch screen;
In accordance with the operation request information, a feature amount representing a feature related to the predetermined operation by the operator is extracted based on operation information output from the touch screen in response to the predetermined operation performed by the operator. Feature extraction process,
The authentication is performed based on the operator feature amount that is the feature amount extracted in the feature amount extraction process and the right holder feature amount that is the feature amount related to the predetermined operation by the right holder registered in advance. The authentication process to be performed,
The personal authentication method characterized by comprising. In a personal authentication program that is applied to a device equipped with a touch screen and authenticates whether an operator operating the device is a right holder having a legitimate right to operate the device,
An operation request procedure for displaying operation request information for requesting to perform a predetermined operation on the touch screen on the touch screen;
In accordance with the operation request information, a feature amount representing a feature related to the predetermined operation by the operator is extracted based on operation information output from the touch screen in response to the predetermined operation performed by the operator. A feature extraction procedure to
The authentication is performed based on the operator feature quantity that is the feature quantity extracted by the feature quantity extraction procedure and the right holder feature quantity that is the feature quantity related to the predetermined operation by the right holder registered in advance. The authentication procedure to perform,
A personal authentication program characterized in that a computer is executed.

Images