High speed id-less collation method and system by multi-stage collation

【0001】

[Technical field to which the invention belongs]

In recent years, personal authentication devices using fingerprints have been sold. In a biometrics personal authentication system, in a method of comparing and collating input biometric information with multiple registered biometric information without identifying an individual by ID etc. (hereinafter referred to as IDless collation), if a round-robin collation is performed, the person and the person are collated. It is conceivable that a huge amount of processing time will be spent before being authenticated. As a means for avoiding this, various methods for shortening the collation time have been considered. In the above personal authentication system, the present invention narrows down a huge amount of registered data to similar data including the person by rough collation, and performs high-speed and accurate personal authentication by multi-step verification. Regarding less matching method and system.

【0002】

[Conventional technology]

Conventionally, various techniques for collating the biometric information input for personal authentication with a plurality of registered biometric information groups have been known. For example, as a fingerprint collation technique, registered data is grouped by a fingerprint pattern and a collation priority is given, and at the time of collation, the pattern is collated using the collation priority and the pattern determined from the input fingerprint is collated. There is a method of limiting the registered data (see Patent Document 1). In the following explanation, when collating the biometric information input for personal authentication with a plurality of registered biometric information groups, the method of collating without specifying a user name or ID is an IDless collation. That is. Human fingerprints can be broadly classified into four types: right-flow type in Fig. 23 (a), left-flow type in Fig. 23 (b), wave-type in Fig. 23 (c), and vortex type in Fig. 24 (d). Can be classified into. Of course, it is possible to classify the data in more detail, but if the data is classified into these four types and populated, the collation target can be narrowed down, so that the collation process can be performed efficiently without unnecessary collation. In this prior art, even if there is a mistake in the pattern classification, the pattern is prioritized in the order in which the classification error is likely to occur, and the matching is performed until the person is found. There is no reason not to do it. Since strict pattern determination processing is not required by being aware of the collation priority, even if the correct pattern cannot always be recognized, the maximum determination can be made from the state of the image. For example, if a fingerprint image having a large area, which is originally indispensable for pattern classification, cannot be collected, it is assumed that an accurate fingerprint pattern cannot be identified. However, even with such a fingerprint image, if the fingerprint wave at a specific position below the center of the fingerprint can be discriminated, it is possible to discriminate a rough pattern, so even if the pattern cannot be accurately determined, priority is given. It is possible to judge a pattern with a high degree. In summary, by sequentially collating the judged patterns with priority, it is possible to authenticate the person even if the patterns do not always match (occurrence of classification error) and minimize unnecessary verification. It becomes. In this way, even if the conventional method is used

【0003】

[Patent Document 1]

JP-A-2002-133416 [0004]

[Problems to be Solved by the Invention]

However, this prior art has the following problems. First, if there is another person's finger that first satisfies the specified match rate in the registered data group as the population due to a classification error of the collation data, even if there is a person's finger that shows a higher match rate in the correct classification, Authentication is performed with another person's finger and the process ends. In addition, if no matching data is found (no registered finger or personal authentication failure), the result is finally obtained after collating all registered data, so the same time as brute force verification. It takes. The present invention has been made to solve the above-mentioned problems of the prior art, and an object of the present invention is to perform IDless collation at high speed while maintaining the reliability of the conventional round-robin collation. The purpose is to provide a high-speed IDless verification method and system that can be performed.

【0005】

[Means for solving problems]

The registered data and collation data have a two-stage structure of coarse collation data and detailed collation data. A coarse collation function that can process at high speed is added to the collation algorithm. Since this can be a process with a low security level for both the personal acceptance rate and the acceptance rate of others, as described in the claims, the ridge width of the fingerprint, the number of feature points, pattern matching in a narrow area of the image, only a part. Use a method such as collating and interrupting the judgment. First, rough collation is performed on all registered data at a higher speed than rough collation to narrow down the data, and then detailed collation processing is performed on the narrowed down registered data, and the data having the highest matching rate is authenticated as the person. By narrowing down the data by coarse collation in this way, IDless collation can be efficiently performed without unnecessary collation. As described above, in the present invention, since the coarse collation may have a high acceptance rate of others, the data to be detailed collated is narrowed down by using a collation method in which the processing is completed at high speed with a high acceptance rate of the principal. Matching is performed only on the narrowed down data, and for registered data that should not match, matching is not performed and matching is completed at high speed, so the reliability of conventional brute force matching is improved. IDless verification can be performed at high speed while maintaining it. The present invention can also be configured as follows. (1) Using the ridge width of the fingerprint and the number of feature points of the fingerprint as the above biometric information, the ridge width for rough verification created for authentication, the number of feature points of the fingerprint, and the registered ridge for rough verification The line width and the number of feature points of the fingerprint are collated, and those within the range close to the ridge width of the fingerprint for authentication and the number of feature points are narrowed down as a population for detailed collation, and registered in association with the narrowed down biometric information for rough collation. The detailed collation biometric information is collated with the detailed collation biometric information created for authentication, and the one with the highest degree of matching is determined as the person. As a result, the rough collation process can be performed at high speed without lowering the personal acceptance rate, and the IDless collation can be performed at high speed. Further, a specific area of the fingerprint is used as the coarse collation information for authentication, and only the specific area of the fingerprint is referred to during the rough collation, and the pattern of the specific area and the above rough collation are used. Pattern matching is performed with the pattern of a specific area registered in the biometric information registration unit, and those in a close range are narrowed down as a population for detailed collation, and only a part of the detailed collation algorithm is used for rough collation. May be used to perform a certain collation and interrupt it, and based on the collation results up to that point, those in a close range may be narrowed down as a population for detailed collation. Further, it may be configured to perform rough collation and detailed collation using only biometric information. For example, the fingerprint feature score is used for rough collation and detailed collation, and in coarse collation, the biometric information created for authentication is narrowed down to a range close to the feature score, and the detailed collation biological information linked to the narrowed biometric information for rough collation. Using the information as a population, detailed collation is performed using the same fingerprint feature score, and the one with the highest degree of matching is determined as the person. As a result, rough collation and detailed collation can be performed using the same database and the same collation algorithm, and the processing content, device configuration, and the like can be simplified. (2) The above-mentioned rough collation information obtained from one biometric information among a plurality of types of biometric information (for example, fingerprint and iris) is registered in the rough collation biometric information registration unit, and among the plurality of types of biometric information. The above detailed collation information obtained from other biometric information is registered in the detailed collation biometric information registration unit, and rough collation information and detailed collation information for authentication are created from a plurality of types of biometric information, respectively, for collation. At that time, the rough collation biometric information (for example, fingerprint) having a fast collation speed is collated with the rough collation information for collation to narrow down the biometric information that may be the person himself / herself, and the biometric information for detailed collation (for example, iris) with a slow collation speed is collated. And, the detailed collation information for collation is collated, and the one with the highest degree of matching is determined as the person. In this way, by performing personal authentication using a plurality of types of biometric information, rough collation can be performed using biometric information having a high collation speed, and if the biometric information used for rough collation is appropriately selected, the whole It is possible to improve the collation processing speed as. In addition, since two types of biometric information are used, it is expected to reduce authentication errors and the like. (3) Detailed verification with the acquired biometric information for rough verification for authentication A client equipped with a means for transmitting biometric information to a server connected to a network and a means for acquiring a collation result on the server, a rough collation biometric information for authentication sent from the client, and a registered rough collation. For collating the biometric information for the server, narrowing down the biometric information that may be the person, and linking with the narrowed down biometric information for rough collation, the registered detailed biometric information information for verification and the detailed collation for authentication sent from the above client. A multi-step verification system for personal authentication is configured with a server that collates with biometric information and sends the collation result to the client. As a result, if the biometric information for rough collation and the biometric information for detailed collation are registered on the server side, it is possible for the server side to collectively perform collation processing for collation requests from a plurality of clients. , The configuration can be simplified and the cost can be reduced. Further, when the data is encrypted and transmitted at the time of transmission / reception between the client and the server, personal data can be kept secret and security can be ensured.

【0006】

BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, embodiments of the present invention will be described. First, as a first embodiment of the present invention, a multi-step collation using a fingerprint as biometric information will be described. In this embodiment, pattern matching is performed using only a part of the fingerprint image collected as the rough matching means, and high-precision fingerprint matching is performed using all of the collected images as the detailed matching means. The fingerprint matching algorithm with high accuracy includes, for example, matching by feature point matching, but various methods have already been proposed in the fingerprint matching method, and the present invention does not limit the method. Any algorithm may be used as long as the accuracy is high. FIG. 1 is a diagram showing an example of a device configuration for registering biological information in advance. As shown in FIG. 1, first, the biological information input unit 101 collects biological information. From this biometric information, the crude collation biometric information extraction unit 122 extracts the crude collation biometric information used as the rough collation. In addition, the detailed collation biometric information extraction unit 123 extracts the detailed collation biometric information to be used as the detailed collation. The extracted two types of biometric information are registered in each of the crude collation biometric information storage unit 121 and the detailed collation biometric information storage unit 124, and the process is completed. Each of the rough collation biometric information storage unit 121 and the detailed collation biometric information storage unit 124 (hereinafter, these biometric information storage units 121 and 124 are also referred to as a rough collation biometric information registration DB and a detailed collation biometric information registration DB, respectively. Needless to say, who the biometric information indicates (for example, user ID) is also stored in). The rough collation biometric information and the detailed collation biometric information registered in each of the rough collation biometric information storage unit 121 and the detailed collation biometric information storage unit 124 include, for example, a fingerprint as described above. The fingerprint image information itself indicating the shape of the fingerprint or the like may be used, or the feature information such as the feature points and the ridge width extracted from the fingerprint may be used as described later. , Called biometric information for detailed verification.

【0007】

FIG. 2 is a flowchart showing the flow of processing when registering biological information. With reference to FIG. 2, the flow of the process of registering biological information in the database in advance will be described using a fingerprint as an example. The processing flow is the same as that described in FIG. 1. First, user information and biometric information (fingerprint) are input (step S1 in FIG. 2). Then, a part of the fingerprint image is extracted from the fingerprint image input for registration (step S2 in FIG. 2). An example of this processing is shown in FIG. In the fingerprint image shown in FIG. 3, only a narrow area surrounded by a square is held in the rough matching biometric information registration DB as a rough matching area. At this time, the central part of the fingerprint is held as shown in FIG. By defining the area for pattern matching as a reference, it is easy to define the position where pattern matching is performed. In FIG. 3, a specific area is defined from the center of the fingerprint to the edge of the image in the horizontal direction, but it goes without saying that the pattern matching process can be further speeded up by further narrowing the area near the center. Then, the biometric information for detailed collation is extracted (step S3). As the biometric information for detailed collation, information necessary for the detailed collation algorithm is obtained from the entire image. The biometric information for detailed collation may be a fingerprint image (usually the entire fingerprint image in this case) as well as the biometric information for coarse collation, but as will be described later, the feature amount information extracted from the fingerprint ( For example, it may be a feature point, a ridge width, etc.). In this way, these biometric information and the user information (for example, user ID, name, etc.) for linking the individual who input the biometric information are registered in the biometric information registration DB for detailed collation. As shown in FIGS. 1 and 2, the coarse collation information and the detailed collation information may be stored separately in the database, or may be collectively stored in a single database. Needless to say, the present invention does not stick to the storage format. In this way, the biometric information of a plurality of users is registered in the database in advance by the same method. These are the data to be referred to when executing personal authentication.

【0008】

Next, the flow of the process of performing multi-step verification with the fingerprint input at the time of personal authentication and authenticating the personal person will be described. FIG. 4 is a diagram showing an example of a device configuration for performing multi-step verification for authenticating with the person when biometric information is input without an ID indicating the user for the person authentication. First, biometric information is collected by the first biometric information input unit 101. From this biometric information, the crude collation biometric information extraction unit 122 extracts the crude collation biometric information used as the rough collation. In addition, the detailed collation biometric information extraction unit 123 extracts the detailed collation biometric information used as the detailed collation. Next, the coarse collation unit 125, which performs high-speed processing, collates a plurality of coarse collation biometric information registered in advance in the coarse collation biometric information storage unit 121 with the coarse collation biometric information extracted for authentication. Narrow down only those with similarities. The detailed collation biometric information that matches the user name of the coarse collation biometric information narrowed down in this way is referred to from the detailed collation biometric information registered in advance in the detailed collation biometric information storage unit 124, and is authenticated by the detailed collation biometric information 126. Compared with the biometric information for detailed collation extracted for the purpose, the person determination unit 127 determines the person with the highest degree of agreement as the person and ends the process.

【0009】

FIG. 5 is a flowchart for performing multi-step verification for authenticating with the person, and the flow of the process for performing multi-step verification for authenticating with the person according to FIG. 5 will be described using a fingerprint as an example. When the biometric information for authentication is input (step S1), the biometric information for rough verification is extracted from the fingerprint image input for authentication (step S2). At this time, the area size may be the same as that at the time of registration, but the fingerprint position may be slightly deviated from that at the time of registration, and the area may be used as a rough collation area in a slightly wider range than at the time of registration. .. In this embodiment, the biometric information for detailed collation is extracted, but if it takes time to extract the biometric information for detailed collation, rough collation is performed first and only when there is one or more similar data. Needless to say, biometric information for detailed collation should be extracted. As a collation procedure, first, a plurality of pre-registered biometric information for rough collation is compared with the biometric information for rough collation input for authentication (step S3). The accuracy of this process may be low as long as the acceptance rate of the person is high. Some high-speed pattern matching processing that has already been proposed may be used, but the presence or absence of feature points is ignored, and the flow of ridges and how many ridges are within a specific range are determined. It doesn't matter if it's just a simple pattern matching. With such pattern matching, it is possible to narrow down some things including others without excluding the person. As described above, the registered biometric information for rough collation is compared with the input biometric information for rough verification, and if similar data cannot be extracted from the biometric information for coarse collation, step S5. Go to step S9 from, and finish the collation with no matching data. If one or more similar data are extracted, a similar data list is created and detailed collation is performed using this as a population.

【0010】

Next, the detailed collation biometric information that matches the user name (user ID) indicated by the narrowed-down rough collation biometric information is read from the detailed collation biometric information registration DB and subjected to highly accurate and detailed collation (step). S6). For this detailed collation process, a highly accurate collation process that has already been proposed may be used. For example, a method that can suppress acceptance by others, such as a method of collating using the positions of the end points and branch points of fingerprints (hereinafter referred to as a feature point matching method), may be used. As a result of detailed collation of the narrowed population, if there is only registered data that satisfies the threshold value for determining the person, the process goes from step S7 to step S8, and the user indicated by the registered data is the person. If a plurality of data match even at this point, the data with the highest matching rate should be the person himself / herself. On the contrary, if there is no data exceeding the threshold value, the process proceeds to step S9 and the process ends without authenticating with the person. The result of the determination in this way is notified to the user, and the process ends.

【0011】

Next, as a second embodiment of the present invention, a specific example of a method of narrowing down data by the thickness of a ridge will be described. However, since the flow of processing from data registration and collation to authentication with the person is the same as in the first embodiment, the description thereof will be omitted, and here, only the rough collation method will be described specifically. It is said that the ridge spacing of human fingerprints is 1 to 4 between 0.5 mm. For example, if the image collected as a fingerprint is a 15 mm square 300 pixel x 300 pixel image, 0.5 mm is 10 images. Therefore, if a person has four ridges between 10 pixels, the thickness of one ridge is 2.5 pixels. Although this definition is used in this embodiment, classification may fail if only one ridge line is processed, so the average thickness of the ridge line existing in a specific range is calculated. Let it be rough collation information. When actually performing rough collation, if the ridge width of the fingerprint input for authentication is 2.5 pixels in this way, something close to it from the rough collation biometric information storage unit 121, for example, 2 pixels to 4 It narrows down to only the pixels and targets it as a population for detailed verification.

【0012】

Next, as a third embodiment of the present invention, a specific example of a method of narrowing down data by the number of feature points will be described. However, since the flow of processing from data registration and collation to authentication with the person is the same as in the first embodiment, the description is omitted, and here, only the rough collation method will be described specifically. The feature points of human fingerprints have an end point that is the end point and a bifurcated branch point, and in feature point matching, by comparing the positions and types of these feature points in detail, highly accurate personal authentication is performed. Has been realized. In the rough collation of this embodiment, these positions and types are ignored, and only the number of feature points is used for narrowing down. However, if you narrow down by the correct number, there is a risk of lowering the acceptance rate of the person, so in consideration of counting mistakes, the population is narrowed down to about ± 5 of the feature points of the fingerprint entered for authentication as the allowable range. It is the target of detailed collation. In this embodiment, the permissible range is described as ± 5, but it goes without saying that the permissible range is not particularly limited. In addition, instead of the feature score of the entire fingerprint, for example, the feature score only near the center of the fingerprint shown in FIG. 3 described above, or the feature score of a gentle part where only the upper part of the fingerprint does not change is used as shown in FIG. In the present invention, the processing type is not limited.

【0013】

As described above, when compared with the conventional processing speed of detailed collation, it is clear that these rough collation processes are completed at high speed, but especially in the above-mentioned third embodiment, only the feature points are compared purely. Therefore, rough matching can be performed at a dramatically high speed. Although there are some fluctuations depending on the performance of the computer, the processing is completed in about microseconds (about 0.000001 seconds) if only pure numerical comparison processing is performed. If this process is used, for example, if the number of users registered in the DB in advance is 100,000, the processing speed of the conventional collation process (detailed collation) is 0.001 seconds for collation between the person and another person in one collation. If 0.01 seconds are required for the person-to-person collation, the time required for all the collations is expressed by the following formula. Since the total time for narrowing down = (other person's index x time for detailed verification of another person's finger) + verification time of the person's finger, the following formula is used. (99,999 × 0.001) + (1 × 0.01) = 100.009 seconds That is, in the past, it took about 1 minute and 40 seconds to complete the process.

【0014】

On the other hand, if the above-mentioned rough collation processing of the feature points is combined and narrowed down to 1/100 (that is, 1000) by the method of the present invention (third embodiment), the following calculation formula is obtained and the processing is completed. The collation time up to can be reduced to about 1/100. Filtering result 1/100: (100,000 × 0.000001) + (999 × 0.001) + (1 × 0.01) = 1.109 seconds Needless to say, the filtering result is the fingerprint entered for collation with the status of the registered database. Therefore, a uniform definition cannot be made. However, as in the calculation formula shown below, whether the number of narrowing down is large or small does not greatly affect the present invention. For example, if 100,000 cases are narrowed down to 1/500 cases, it is narrowed down to 200 cases, so the collation time until the processing is completed is as follows. Narrowing down result 1/500 cases: (100,000 × 0.000001) + (199 × 0.001) + (1 × 0.01) = 0.309 seconds Also, if 100,000 cases are narrowed down to 1/50 cases, it will be narrowed down to 2000 cases, so until the processing is completed. The collation time of is as follows. Narrowing down result 1/50 : (100,000 × 0.000001) + (1999 × 0.001) + (1 × 0.01) = 2.109 seconds In this way, if you combine the rough collation processing that performs high-speed processing and perform the conventional detailed collation only for the narrowed down data. , The population can be surely included, and the identity authentication by IDless verification can be completed at high speed without unnecessary verification.

【0015】

In the above embodiment, in the rough collation, the population is narrowed down by using the positions of the end points and branch points of the fingerprint, the thickness of the ridge line, the number of feature points, etc., and then the feature point matching method and the like have high accuracy. The case where the person is authenticated by performing detailed collation has been described. For example, in coarse collation, only a part of the detailed collation algorithm is used to perform a certain collation, interrupt the process, and perform the collation up to that point. Based on the result, those in a close range may be narrowed down as a population for detailed collation. For example, an algorithm that divides the input fingerprint image into partial images and collates each partial image with a collation partial image registered in the database is used, and a part of the algorithm is used to collate the input fingerprint. Rough collation is performed on a part of the image to narrow down the population, and then in detailed collation, all the input fingerprint images are collated in the database using the above algorithm . The same algorithm may be used for coarse collation and detailed collation, such as collation with data, and a part of the above algorithm may be used to perform rough collation.

【0016】

Alternatively, the only biometric information may be extracted from the biometric information input for personal authentication, and only the biometric information may be narrowed down by rough collation to detailed collation to authenticate the principal. FIG. 7 is a diagram showing an example of an apparatus configuration for performing registration processing when only biometric information is used as described above, and FIG. 8 is a diagram showing a flow of the registration processing. As shown in FIGS. 7 and 8, biological information is collected by the biological information input unit 101 (step S1 in FIG. 8). From this biometric information, the biometric information extraction unit 300 extracts the biometric information used for rough collation / detailed collation, and registers the extracted biometric information in the biometric information storage unit 301 (step S2 in FIG. 8). In the case of a fingerprint, for example, only characteristic information such as the end point and the position of a branch point (position of the characteristic point) of the fingerprint is extracted from the collected biological information and registered in the biological information storage unit 301.

【0017】

FIG. 9 is a diagram showing an example of an apparatus configuration for performing collation processing when only biometric information is used, and FIG. 10 is a diagram showing the flow of the collation processing. At the time of collation, as shown in FIGS. 9 and 10, first, the biometric information is input by the biometric information input unit 101 (step S1 in FIG. 10), and the biometric information for authentication is input by the collation biometric information extraction unit 300. Extract (step S2 in Figure 10). Then, the coarse collation unit 125 collates with the crude collation biometric information stored in the collation biometric information storage unit 301, and extracts similar data (step S3 in FIG. 10). For example, the position of the feature point is extracted from the input biometric information for authentication, and the position of the feature point of a plurality of biometric information stored in the biometric information storage unit is collated by the rough collation unit, and the position is in a close range. Is narrowed down by extracting as having similarities. Similar data is stored in the similar data list. If there is no similar data, the process goes from step S4 to step S8, and the collation is terminated with no matching data. Next, the detailed collation unit 126 performs detailed collation. In detailed collation, based on a similar data list, biometric information that matches the user name (user ID) indicated by the narrowed down biometric information is read out from the biometric information registration DB (biological information storage unit 301) and referred to with high accuracy. Perform detailed matching (step S5). That is, the position of each feature point of the registered biometric information, which is the only biometric information, is collated with the position of each feature point of the input biometric information for authentication. Then, if there is no matching data satisfying the threshold value, the process proceeds from step S6 to step S7, and the collation is terminated with no matching data. If there is matching data that satisfies the threshold value, the person determination unit 127 determines the person (step S8) and ends the process. In this way, coarse collation and detailed collation are performed using only biometric information. Since the biometric information for collation and detailed collation can be shared and the same algorithm can be used for collation, the process can be simplified.

【0018】

Next, an example of authenticating the person using two types of biological information (for example, fingerprint and iris) will be described. FIG. 11 is a diagram showing an example of a device configuration for registering two types of biological information. As shown in FIG. 11, first, biological information (for example, a fingerprint) is collected by the first biological information input unit 201. From this biometric information, the coarse collation biometric information extraction unit 122 extracts the rough collation biometric information used as the rough collation, and stores it in the rough collation biometric information storage unit 121. Next, the second biological information input unit 202 collects biological information (for example, an iris). From this biometric information, the detailed collation biometric information extraction unit 123 extracts the detailed collation biometric information used for detailed collation, registers it in the detailed collation biometric information storage unit 124, and ends the process. As described above, each of the rough collation biometric information storage unit 121 and the detailed collation biometric information storage unit 124 stores the biometric information indicating who (for example, user ID).

【0019】

FIG. 12 is a flowchart showing a processing flow when registering two types of biological information. FIG. 12 illustrates the flow of the process of registering two types of biometric information in the database. The processing flow is the same as that described in FIG. 11, and first, user information and first biometric information (for example, fingerprint) are input (step S1 in FIG. 12). Then, the crude collation biometric information is extracted from the input first biometric information (step S2), and is stored together with the user information in the rough collation biometric information registration DB as the rough collation biometric information. Next, the second biometric information (for example, the iris) is input (step S3), the detailed collation biometric information is extracted from the input second biometric information (step S4), and the user information is used as the detailed collation biometric information. , Stored in the biometric information registration DB for rough verification. As described above, the coarse collation information and the detailed collation information may be stored separately in the database, or may be collectively stored in a single database. In this way, the biometric information of a plurality of users is registered in the database in advance by the same method. These are the data to be referred to when executing personal authentication. In the above, the case where the fingerprint is used as the biometric information for rough collation and the iris is used as the biometric information for detailed collation has been described. However, for example, the iris is used as the biometric information for rough collation and the fingerprint is used as the biometric information for detailed collation. It may be used, in short, the rough collation biometric information may be extracted from the biometric information having a high collation speed, and the detailed collation biometric information may be extracted from another biometric information. Further, as described above, the accuracy of the rough collation may be low as long as the acceptance rate of the person is high. Therefore, as the biometric information for the rough collation, as described above, how many ridges flow and how many lines are within a specific range. It may be biometric information or the like for performing simple pattern matching such as determining whether or not there is a ridge.

【0020】

FIG. 13 is a diagram showing an example of a device configuration for performing multi-step verification for authenticating with the person when two types of biometric information input for authenticating with the person are input. First, the first biological information input unit 201 collects biological information (for example, a fingerprint). From this biometric information, the crude collation biometric information extraction unit 122 extracts the crude collation biometric information used as the rough collation. Further, the second biometric information input unit 202 collects biometric information (for example, an iris), and the detailed collation biometric information extraction unit 123 extracts the detailed collation biometric information used for detailed collation. Next, in the coarse collation unit 125 that performs high-speed processing, the coarse collation biometric information (for example, fingerprint) input for authentication and a plurality of coarse collation biometric information registered in advance in the coarse collation biometric information storage unit 121 Collate with biometric information and narrow down only those with similarities. The detailed collation biometric information (for example, iris) that matches the user name of the coarse collation biometric information narrowed down in this way is referred to from the detailed collation biometric feature information registered in advance in the detailed collation biometric information storage unit 124. It is compared with the biometric information for detailed verification extracted for authentication, and the person having the highest degree of matching is determined by the person determination unit 127 as the person, and the process is completed.

【0021】

FIG. 14 is a flowchart for performing multi-step verification using two types of biometric information for authenticating the person. When the first biometric information for authentication (for example, a fingerprint image) is input (step S1), the biometric information for rough matching is extracted from the first biometric information input for authentication (step S2). Then, when the second biometric information (for example, an iris image) is input (step S3), the biometric information for detailed collation is extracted from the second biometric information input for authentication (step S4). As a collation procedure, first, a plurality of coarse collation biometric information registered in advance in the coarse collation biometric information registration DB is compared with the coarse collation biometric information input for authentication (step S5). Then, if similar data cannot be extracted from the crude collation biometric information, the process proceeds from step S6 to step S10, and the collation ends with no matching data. If one or more similar data are extracted, a similar data list is created and detailed collation is performed using this as a population. That is, the detailed collation biometric information (for example, iris) that matches the user name (user ID) indicated by the narrowed-down rough collation biometric information is read from the detailed collation biometric information registration DB and referred to, and the details are highly accurate. (Step S7). As a result of detailed collation of the narrowed population, if there is only registered data that satisfies the threshold value for determining the person, the process goes from step S8 to step S9, and the user indicated by the registered data is the person. If a plurality of data match even at this point, the data with the highest matching rate should be the person himself / herself. On the contrary, if there is no data exceeding the threshold value, the process proceeds to step S10 and the process ends without authenticating with the person. The result of the determination in this way is notified to the user, and the process ends.

【0022】

FIG. 15 is a diagram showing an example of a device configuration for performing biometric information registration processing when performing multi-step collation via a network, and FIG. 16 illustrates a biometric information registration process when performing multi-step collation via a network. The processing content is the same as that of the above embodiment except that the processing is divided into a client and a server and the data is transferred as packet data. When the registration process is performed using the network, as shown in FIG. 15, the above-mentioned biometric information input unit 101, rough collation biometric information extraction unit 122, and detailed collation biometric information extraction unit 123 are provided on the client ant side. , Packet data generation unit 143, and packet data transmission unit 144 for transmitting data via a network are provided. Further, on the server side, a packet data receiving unit 145, a packet data separating unit 146, a packet data transmitting unit 144, the above-mentioned rough collation biometric information storage unit 121, and a detailed collation biometric information storage unit 124 are provided. Then, as shown in FIG. 16, the following processing is performed on the client side. First, user information and biometric information are input (step S1), and crude collation biometric information and detailed collation biometric information are extracted from the input biometric information (steps S2 and S3). Then, packet data is generated from the biometric information for coarse collation and the biometric information for detailed collation (step S4), and the packet data is transmitted to the server (step S5). The following processing is performed on the server side. Receive packet data from the client (step S6), separate the data, and store the crude collation biometric information and detailed collation biometric information together with the user information in a database (coarse collation biometric information registration DB, detailed collation biometric information registration DB). ) (Step S7).

【0023】

FIG. 17 is a diagram showing an example of a device configuration for performing biometric information collation processing when performing multi-step collation via a network, and FIG. 18 is a diagram illustrating biometric information collation processing when performing multi-step collation via a network. In the figure, the processing contents are the same as those in the above embodiment except that the processing is divided into a client and a server and the data is transferred as packet data. When collation processing is performed using a network, as shown in FIG. 17, the biometric information input unit 101, the rough collation biometric information extraction unit 122, and the detailed collation biometric information extraction unit 123 are provided on the client ant side. , Packet data generation unit 143, packet data transmission unit 144 for transmitting data via a network, collation result reception unit 136, and collation result user notification unit 132. Further, on the server side, a packet data receiving unit 145, a packet data separating unit 146, a coarse collation unit 133, a biometric information storage unit 121 for rough collation, a detailed collation unit 134, a biometric information storage unit 124 for detailed collation, and a personal determination unit 135. , The collation result notification unit 131 is provided.

【0024】

Then, as shown in FIG. 18, the following processing is performed on the client side. First, the biometric information for authentication is input (step S1), and the crude collation biometric information and the detailed collation biometric information are extracted from the input biometric information (steps S2 and S3). Then, packet data is generated (step S4), the packet data is sent to the server, and a collation request is made (step S5). Then, it waits for the notification from the server side (step S6), and if there is a notification, it goes from step S7 to step S8, notifies the user of the collation result, and ends the process. The following processing is performed on the server side. Receive the packet data from the client, receive the verification request (step S9), separate the packet data (step S10), refer to the biometric information registration DB for coarse collation, and refer to the received biometric information and coarse for authentication. Matching is performed and similar data is extracted (step S11). Similar data is stored in the similar data list. If there is no similar data, the process proceeds from step S12 to step 17, and the process ends with no matching data. If there is matching data, detailed collation is performed based on the narrowing result by rough collation as described above (step S13). Then, if there is no matching data that satisfies the threshold value, the process proceeds from step S14 to step S17, the process is terminated with no matching data, and if there is matching data, the data having the highest degree of matching is determined as the person (step S15). Notify the client of the matching result (step S16).

【0025】

FIG. 19 is a diagram showing an example of a device configuration when encrypted packet data is transmitted via a network and biometric information is registered. In FIG. 19, in order to encrypt packet data, a packet data encryption unit 141 is provided on the client side in the one shown in FIG. 15, and the encrypted packet data received on the server side is decrypted. The packet data decoding unit 142 is provided, and other configurations are the same as those shown in FIG. FIG. 20 is a diagram illustrating a process in which encrypted packet data is transmitted via a network and biometric information is registered. The processing content is the same as that in FIG. 16 except that the processing for encrypting the packet data (step S21) and the processing for decrypting the packet data (step S22) are added. On the client side, user information and biometric information are input (step S1), and crude collation biometric information and detailed collation biometric information are extracted from the input biometric information (steps S2 and S3). Then, packet data is generated from the biometric information for coarse collation and the biometric information for detailed collation (step S4), the packet data is encrypted by the packet data encryption unit 141 (step S21), and the encrypted packet data is transmitted to the server. (Step S5). Further, on the server side, packet data is received from the client (step S6), the packet data decoding unit 142 decodes the data (step S22), separates the data, and together with the user information, for rough collation biometric information and detailed collation. The biometric information is stored in a database (biological information storage unit for rough collation, biometric information storage unit for detailed collation) (step S8).

【0026】

FIG. 21 is a diagram showing an example of a device configuration in which encrypted packet data is transmitted via a network and biometric information collation processing is performed when multi-step collation is performed. In the one shown in FIG. 17, in the one shown in FIG. 17, the packet data encryption unit 141 for encrypting the packet data and the packet decryption unit 142 for decrypting the received encrypted packet are shown on the client side. The packet data decryption unit 142 for decrypting the encrypted packet data received on the server side and the packet data encryption unit 141 for encrypting the packet data are provided, and other configurations are provided. Is similar to that shown in FIG.

【0027】

FIG. 22 is a diagram illustrating a biometric information collation process in the case of transmitting encrypted packet data via a network and performing multi-step collation, and as a client-side process, a process of encrypting the packet data (step). S21), add the process of decrypting the received packet data (step S22), and as the process on the server side, the process of decrypting the received packet data (step S23) and the process of encrypting the packet data to be transmitted (step S23). Step S24) is added, and the other processes are the same as those shown in FIG. In FIG. 22, on the client side, biometric information for authentication is input (step S1), and biometric information for rough collation and biometric information for detailed collation are extracted from the input biometric information (step S2, S3). Then, the packet data is generated (step S4), the packet data is encrypted (step S21), the packet data is transmitted to the server side, and a verification request is made (step S5). Then, it waits for the notification from the server side (step S6), and if there is a notification, goes from step S7 to step S22, decodes the packet data (step S22), notifies the user of the collation result (step S8), and processes. To finish. On the server side, the packet data is received from the client and a verification request is received (step S9), the received packet data is decoded (step S23), the packet data is separated (step S10), and the biometric information for rough verification is stored. Refer to the section. Then, rough collation is performed with the received biometric information for authentication, and similar data is extracted (step S11). Similar data is stored in the similar data list. If there is no similar data, the process proceeds from step S12 to step 17, and the process ends with no matching data. If there is matching data, detailed collation is performed based on the narrowing result by rough collation as described above (step S13). Then, if there is no matching data that satisfies the threshold, the process proceeds from step S15 to step S17, the process is terminated with no matching data, and if there is matching data, the data with the highest degree of matching is determined as the person (step S15). The packet data is encrypted (step S24) and the matching result is notified to the client (step S16).

【0028】

(Appendix 1) This is a method of collating biometric information entered without personally identifiable information with multiple pre-registered biometric information to identify the individual, and narrows down the collated data for personal authentication in advance. The biometric information for rough collation and the biometric information for detailed collation are registered, and the biometric information for rough collation and the biometric information for detailed collation are created from the biometric information input by the user for collation at the time of collation. Then, the biometric information for rough collation registered in advance is collated with the biometric information for rough collation created for authentication, and only the biometric information for rough collation that has similarities is narrowed down. A multi-step verification method for personal authentication, which is characterized in that the biometric information for detailed collation that is linked to information is collated with the biometric information for detailed collation created for authentication, and the one with the highest degree of matching is determined as the principal. (Appendix 2) The fingerprint is used as the biometric information, and the ridge width of the fingerprint existing in a specific range is used as the biometric information for rough collation. The ridge width for rough collation created for authentication and the registered ridge for rough collation are used. The line width is collated, and those within the range close to the ridge width of the fingerprint for authentication are narrowed down as the population for detailed collation, and the registered detailed collation biometric information linked with the narrowed biometric information for coarse collation and authentication. A multi-step verification method for personal authentication in Appendix 1, which is characterized in that the biometric information for detailed collation created for is collated and the one with the highest degree of matching is determined as the principal. (Appendix 3) The fingerprint is used as the above biometric information, the number of fingerprint feature points is used as the rough collation biometric information, the number of rough collation feature points created for authentication, and the registered rough collation fingerprint. The registered detailed collation biometric information linked with the narrowed rough collation biometric information and the detailed collation biometrics created for authentication are narrowed down to the population in the close range as the population for detailed collation. A multi-step verification method for personal authentication in Appendix 1, which is characterized by collating information and determining the one with the highest degree of matching as the principal. (Appendix 4) A specific area of the fingerprint is used as the rough collation information for authentication, and only the specific area of the fingerprint is referred to during the rough collation, and the pattern of the specific area and the pattern of the specific area registered in the biometric information registration unit for rough verification are used. Pattern matching is performed, and those in a close range are narrowed down as a population for detailed matching, and the detailed matching biometric information linked to the narrowed down rough matching biometric information is collated with the detailed collation biometric information created for authentication. A multi-step verification method for personal authentication in Appendix 1, which is characterized by determining the person with the highest degree of matching as the person. (Appendix 5) Fingerprints are used as biometric information, and in rough collation, only a part of the detailed collation algorithm is used to perform a certain collation and interrupt, and the information is in a close range based on the collation results up to that point. Is narrowed down as a population for detailed collation, and the detailed collation biometric information linked to the narrowed down biometric information for rough collation is collated with the biometric information for detailed collation created for authentication by the above collation algorithm, and the one with the highest degree of matching is selected. A multi-step verification method for personal authentication in Appendix 1, which is characterized by determining the person as the person. (Appendix 6) This is a method of identifying a person by collating the biometric information entered without identifying an individual with a plurality of types of biometric information registered in advance, and is one of the multiple types of biometric information. The above-mentioned rough collation information obtained from the above is registered in the rough collation biometric information registration unit, and the above detailed collation information obtained from other biometric information among a plurality of types of biometric information is registered in the detailed collation biometric information registration unit. Register, create coarse collation information for authentication and detailed collation information from multiple types of biometric information, and collate the coarse collation biometric information with fast collation speed with the collation rough collation information at the time of collation. , The person who narrows down the biometric information that may be the person, collates the detailed collation biometric information with a slow collation speed with the detailed collation information for collation, and determines the person with the highest degree of matching as the person. Multi-step verification method for authentication. (Appendix 7) A client equipped with a biometric information input acquisition unit, a means for transmitting the acquired biometric information for rough collation for authentication, a biometric information for detailed collation to a server connected to the network, and a means for acquiring the collation result on the server. , The rough collation biometric information registration unit that holds the rough collation biometric information, the detailed collation biometric information registration unit that holds the detailed collation biometric information, and the rough collation for authentication sent from the above client. In order to collate the registered biometric information for rough collation with the biometric information rough collation unit that narrows down the biometric information that may be the person, and to determine the biometric information with the highest degree of agreement as the person. A biometric information detailed collation unit that collates the registered detailed collation biometric information linked with the narrowed-down rough collation biometric information and the detailed collation biometric information for authentication sent from the above client, and sends the collation result to the client. A multi-step verification system for personal authentication, which is characterized by having a server to perform. (Appendix 8) The first and second biometric information creation units that create rough collation biometric information for authentication and detailed collation biometric information from the biometric information input acquisition unit and the biometric information input from the biometric information input acquisition unit. A client for personal authentication, which comprises a means for transferring the above-mentioned biometric information for rough collation and biometric information for detailed collation to a server connected to a network, and a means for acquiring a collation result on the server. (Appendix 9) Coarse collation biometric information registration unit that registers rough collation biometric information, detailed collation biometric information registration unit that registers detailed collation biometric information, and authentication rough collation information sent from the client. And, in order to collate the registered biometric information for rough collation and narrow down the biometric information that may be the person, and to determine the biometric information with the highest degree of matching as the person, the above narrowing down is performed. The biometric information detailed collation unit that collates the registered detailed collation biometric information associated with the coarse collation biometric information with the detailed collation biometric information for authentication transmitted from the client, and transmits the collation result to the client. A server that performs multi-step verification for personal authentication, which is characterized by having means. (Appendix 10) The first and second biometric information creation units that create rough collation information and detailed collation information from the biometric information input acquisition unit and the biometric information input from the biometric information input acquisition unit, and the first biological body. The rough collation biometric information registration unit that holds the rough collation information created by the information creation unit and the detailed collation biometric information registration that holds the detailed collation information created by the second biometric information creation unit above. The rough collation information registered in the above-mentioned rough collation biometric information registration unit is collated with the rough collation biometric information created for authentication, and the biometric information that may be the person is narrowed down. In order to determine the biometric information that has the highest degree of agreement with the department as the person, the biometric information for detailed collation that is linked to the biometric information for coarse collation that has been narrowed down is acquired from the biometric information registration unit for detailed collation and created for authentication. A multi-step collation device for personal authentication, which is provided with a biometric information detailed collation unit for collating with biometric information for detailed collation. (Appendix 11) A program that identifies a person by collating the biometric information entered without identifying an individual with a plurality of pre-registered biometric information. The above program is input by a user for authentication. The process of creating rough collation biometric information and detailed collation biometric information from biometric information, and collating the pre-registered rough collation biometric information with the rough collation biometric information for authentication, and having similarities. The process of narrowing down only, and the process of collating the pre-registered detailed collation biometric information linked with the narrowed down biometric information for rough collation with the biometric information for detailed collation for authentication, and determining the one with the highest degree of matching as the person. A multi-step verification program for personal authentication characterized by having a computer execute it.

【0029】

[Effect of the invention]

As described above, in the present invention, the feature information for coarse collation and detailed collation is extracted from the biological information, and the population to be collated is narrowed down by the rough collation capable of high-speed processing, and the narrowed population is narrowed down. Since detailed collation that takes a long time to process is performed on the data, it is faster and more accurate IDless collation that does not impair reliability compared to performing detailed collation on all data. Is possible.

[Simple explanation of drawings]

FIG. 1 is a diagram showing an example of an apparatus configuration for performing a registration process of an embodiment of the present invention.

FIG. 2 is a diagram illustrating a flow of registration processing according to an embodiment of the present invention.

FIG. 3 is a diagram illustrating a pattern matching area according to an embodiment of the present invention.

FIG. 4 is a diagram showing an example of an apparatus configuration for performing collation processing according to an embodiment of the present invention.

FIG. 5 is a diagram illustrating a flow of collation processing according to an embodiment of the present invention. FIG. 6 is a diagram showing an example in which a feature score of a gentle portion with no change in the upper part of a fingerprint is used. [Fig. 7] Only It is a figure which shows the device configuration example for performing the registration process of the Example which uses the biological information of.

FIG. 8 is a diagram illustrating a flow of registration processing of an example using only biological information.

FIG. 9 is a diagram showing an example of a device configuration for performing collation processing of an embodiment using only biological information.

FIG. 10 is a diagram illustrating a flow of collation processing of an embodiment using only biological information.

FIG. 11 is a diagram showing an example of a device configuration for performing a biometric information registration process of an embodiment using two types of biometric information.

FIG. 12 is a diagram illustrating a flow of a biometric information registration process of an embodiment using two types of biometric information.

FIG. 13 is a diagram illustrating an example of a device configuration for performing a biometric information collation process of an embodiment using two types of biometric information.

FIG. 14 is a diagram illustrating a flow of a biometric information collation process of an embodiment using two types of biometric information.

FIG. 15 is a diagram illustrating an example of a device configuration of an embodiment in which biometric information is registered via a network.

FIG. 16 is a diagram illustrating a flow of a registration process of an embodiment in which a biometric information registration process is performed via a network.

FIG. 17 is a diagram illustrating an example of a device configuration of an embodiment in which biometric information is collated via a network.

FIG. 18 is a diagram illustrating a flow of a biometric information collation process of an embodiment in which multi-step collation is performed via a network.

FIG. 19 is a diagram showing an example of a device configuration of an embodiment in which encrypted packet data is transmitted via a network and biometric information is registered.

FIG. 20 is a diagram illustrating a flow of a registration process of an embodiment in which encrypted packet data is transmitted via a network and biometric information is registered.

FIG. 21 is a diagram showing an example of a device configuration of an embodiment in which encrypted packet data is transmitted via a network and biometric information is collated.

FIG. 22 is a diagram illustrating a flow of collation processing of an embodiment in which encrypted packet data is transmitted via a network and bioinformation collation processing is performed.

FIG. 23 is a diagram showing an example of a fingerprint determined to be a right-flow type, an example of a fingerprint determined to be a left-flow type, and an example of a fingerprint determined to be a wavy type.

FIG. 24 is a diagram showing an example of a fingerprint determined in a vortex shape.

[Explanation of symbols]

101 Biometric information input unit 121 Coarse collation biometric information storage unit 122 Coarse collation biometric information extraction unit 123 Detailed collation biometric information extraction unit 124 Detailed collation biometric information storage unit 125 Coarse collation unit 126 Detailed collation unit 127 Personal determination unit 131 Matching Result notification unit 132 Verification result user notification unit 133 Rough verification unit 134 Detailed verification unit 135 Identity determination unit 136 Verification result reception unit 141 Packet data encryption unit 142 Packet data decryption unit 143 Packet data generation unit 144 Packet data transmission unit 145 Packets Data receiving unit 146 Packet data separation unit 201 1st biometric information input unit 202 2nd biometric information input unit 300 Biological information extraction unit 301 Biological information storage unit