JP6798798B2 - Method and device for updating data for user authentication - Google Patents

Description

The present invention relates to a method and an apparatus for updating data for user authentication.

With the development of various mobile devices such as smartphones and wearable devices, the importance of security authentication is increasing. Biometric authentication uses fingerprints, irises, voices, faces, blood vessels, and the like to authenticate users. The biological characteristics used for authentication vary from person to person, and not only are there no inconveniences to possess, but there is less risk of plagiarism or forgery, and they will not change for life.

In particular, the fingerprint recognition method is currently the most commercially available for various reasons such as convenience, security, and economy. For example, the fingerprint recognition method can enhance security for user devices and easily provide various application services such as mobile payment.

According to one embodiment of the present invention, the results already calculated in the authentication process are fully utilized, the similarity is calculated with the best effort within the time available for updating, and if the time is insufficient, in advance. By setting the similarity to the determined minimum value, the template is updated adaptively regardless of the length of time allowed for the update.

The method of updating the biometric authentication data of the registration database according to the embodiment of the present invention is a step of authenticating an input image using the registration database during the time of the first length (the authentication is the input). Includes the generation of information for authentication of the image) and the step of updating the registration database based on the time of the first length and the information for authentication of the input image.

The steps of adaptively updating the registration database include a step of determining the similarity based on the size of the overlapping region between the registration image and the input image in the registration database, and the step based on the similarity. It may include a step of deciding whether or not to register the input image in the registration database.

The step of determining the similarity is a step of determining at least a part of the similarity based on the information for authenticating the input image, and the similarity is determined based on the time of the first length. At least one of the steps of determining at least a part of the minimum values may be included.

The steps for determining whether or not to register in the registration database are the step for determining whether or not the maximum similarity among the similarity is equal to or lower than the upper threshold, and whether or not the maximum similarity is equal to or higher than the lower threshold. It may include at least one of the steps to determine.

The upper threshold may correspond to the maximum similarity based on the overlapping area between the registered images.

The information for authenticating the input image includes the degree of correlation within the overlapping area between the input image and the registered image in the registered database, and the size of the overlapping area between the input image and the registered image. At least one of them may be included.

The step of authenticating the input image is a correlation between the step of determining the size of the overlapping area between the input image and the registered image in the registered image and the correlation within the overlapping area between the input image and the registered image. A step of determining the degree, a step of determining whether the degree of correlation is greater than or equal to the threshold, a step of outputting an acceptance signal when the degree of correlation is equal to or greater than the threshold, and the size of the overlapping region. And the step of storing the correlation degree in the memory (the size of the stored overlapping area and the correlation degree correspond to the registered image) may be included.

The step of authenticating the input image may further include a step of storing the time of the first length in the memory.

The step of authenticating the input image may include a step of authenticating the input image by using a multi-stage authentication method (multi-stage authentication).

The adaptively updating step is based on the step of estimating the second length time for updating the registration database based on the time of the first length and the step of estimating the time of the second length based on the time of the second length. , The step of adaptively updating the registration database may be included.

The step of estimating the second time is based on the difference between the time of the third length, which is the time of the length allocated for the biometric authentication, and the time of the first length. It may include a step of calculating the time of two lengths.

The step of adaptively updating the registration database is at least one of a step of adding the input image to the registration database and a step of replacing any one of the registration images in the registration database with the input image. May include.

The replacement step may include selecting a replacement image based on the similarity based on the size of the overlapping region between the registered images.

According to one embodiment of the present invention, the method of updating the biometric authentication data in the registration database is a duplication between the step of receiving an input image including biometric information and the input image and the registered image in the registration database. A step of determining a first maximum similarity based on the size of an overlapped area and an upper threshold based on the size of the overlapping region where the first maximum similarity is between the registered images. Includes a step of determining whether or not the first maximum similarity reaches the higher threshold, and a step of registering the input image in the registration database based on the determination result of whether or not the first maximum similarity reaches the upper threshold.

The step of determining the first maximum similarity is a step of determining the similarity between the input image and the registered image based on the size of the overlapping region between the input image and the registered image. , The step of determining the first maximum similarity having the largest similarity value among the similarity may be included.

The steps of determining the similarity are the step of determining the correlation within the overlapping region between the registered image of any one of the registered images and the input image, and the registered image of any one of the registered images. The degree of similarity between the registered image and the input image using the step of determining the size of the overlapping region between the image and the input image, and the degree of correlation and the size of the overlapping region. Of these, the step of determining the first similarity may be included.

The step of determining the similarity is performed by using a movement vector and a rotation angle based on image relocation between the registered image of any one of the registered images and the input image. A step of determining a normalized cross-correlation (NCC) value, a step of determining a shared area ratio (sharing area ratio) between any one of the registered images and the input image, and the normalization. It may include a step of determining the first similarity among the similarity between the registered image and the input image by using the cross-correlation value and the shared area ratio.

The method of updating the registered database may further include a step of selecting a replacement image from the registered images based on a first similarity based on the size of the overlapping region between the registered images.

The step of selecting the replacement image is a step of selecting a candidate image to be replaced from the registered images based on the first similarity based on the size of the overlapping region between the registered images (each of the candidate images). Is related to the portion of the first similarity (which includes the first similarity between the related candidate image and the remaining registered image) and the step of accumulating each of the portions. And the step of selecting a candidate image having a high cumulative portion from the candidate images as the replacement image may be included.

The step of registering the input image in the registration database is a step of updating the registration database by replacing the replacement image with the input image if the first maximum similarity is smaller than the upper threshold value, and the first step. If the maximum similarity is greater than the higher threshold, it may include a step of retaining the replacement image in the registration database.

The upper threshold may include a second maximum similarity based on the size of the overlapping region between the registered images.

The steps of registering the input image in the registration database include a step of determining whether or not the first maximum similarity exceeds the lower threshold value and whether or not the first maximum similarity exceeds the lower threshold value. It may include a step of registering the input image in the registration database based on the determination of.

In the step of registering the input image in the registration database based on the determination as to whether or not the first maximum similarity exceeds the lower threshold, the first maximum similarity is larger than the lower threshold and the upper threshold is used. If it is smaller, the step of registering the input image in the database may be included.

The registration database update method includes a step of determining an authentication score between the input image and the registered image, and a step of performing user authentication for the input image based on the authentication score and the authentication threshold (the user authentication). The step to be performed further includes (including the step of generating the similarity between the registered image and the input image), and the step of determining the first maximum similarity is based on the generated similarity. The step of determining the first maximum similarity may be included.

The biological information may include at least one of fingerprint information, blood vessel information, and iris information.

According to one embodiment of the present invention, the database management method includes a step of receiving an input image including biometric information and a step of updating the database when the input image is authenticated based on the registered image in the database. The updating step includes determining whether or not to add the input image to the database, and the authentication step and the updating step are performed at the same time as or less than a set time. ..

The biological information may be fingerprint information.

The update step is a step of determining the similarity based on the size of the overlapping region between the registered image included in the database and the input image, and the input image is transferred to the database based on the similarity. It may include a step of deciding whether or not to register.

The step of determining the similarity is based on the authentication of the input image, the step of determining at least a part of the similarity, and the length of the set time and the time associated with the authentication. It may include at least one of the steps of determining at least a portion of the similarity to a set value.

The steps for determining whether or not to register in the registration database are the step for determining whether or not the maximum similarity among the similarity is equal to or lower than the upper threshold, and whether or not the maximum similarity is equal to or higher than the lower threshold. It may include at least one of the steps to determine.

The updating step may include selecting replacement images based on similarity based on the size of overlapping regions between the registered images.

According to one embodiment of the present invention, the method of updating the biometric authentication data in the database includes a step of authenticating an input image including biometric information based on the registered image in the database, and the input image and the registered image. Includes a step of determining the similarity associated with, and a step of updating the database based on the similarity, the lower threshold, and a higher threshold higher than the lower threshold.

The authentication step, the determination step, and the update step may be performed at a set length of time.

The authenticating step authenticates the input image at the time of the first length, the updating step updates the database at the time of the second length, and the updating step is the second length. May include a step of applying such that the time of is the difference between the time of the set length and the time of the first length.

The updating step may include replacing one of the registered images with the input image when the similarity is between the upper threshold and the lower threshold.

The upper threshold may be associated with the maximum similarity and the lower threshold may be associated with the minimum similarity.

By making the best use of the results already calculated in the authentication process, the template can be updated adaptively regardless of the length of time allowed for the update.

It is a figure explaining the operation of acquiring the input image including the biological information which concerns on one Embodiment of this invention. It is a figure explaining the recognition rate which is improved by updating the registration database which concerns on one Embodiment of this invention. It is a flowchart which showed the update method of the registration database which concerns on one Embodiment of this invention. It is a flowchart which showed the method of calculating the degree of similarity which concerns on one Embodiment of this invention. It is a figure for demonstrating the method of determining the maximum degree of similarity which concerns on one Embodiment of this invention. It is a flowchart which showed the update method of the registration database which concerns on other embodiment of this invention. It is a flowchart which showed the method of selecting the replacement image which concerns on one Embodiment of this invention. It is a figure for demonstrating the method of selecting the replacement image which concerns on one Embodiment of this invention. It is a figure for demonstrating the method of selecting the replacement image which concerns on other embodiment of this invention. It is a flowchart which showed the update method of the registration database which concerns on one Embodiment of this invention. It is a figure explaining the authentication process of biometrics which concerns on one Embodiment of this invention. It is a figure explaining the adaptive update of the registration database which concerns on one Embodiment of this invention. It is an operation flowchart which shows the method of adaptively updating the registration database which concerns on one Embodiment of this invention. It is a block diagram which shows the multi-stage authenticator which concerns on one Embodiment of this invention. It is a figure explaining the rough search which concerns on one Embodiment of this invention. It is a figure explaining the precise search which concerns on one Embodiment of this invention. It is a figure explaining the update apparatus which adaptively updates the registration database which concerns on one Embodiment of this invention. It is a figure explaining the update apparatus which adaptively updates the registration database which concerns on other embodiment of this invention. It is a figure explaining the update apparatus which adaptively updates the registration database which concerns on other embodiment of this invention. It is a figure explaining the update apparatus which adaptively updates the registration database which concerns on other embodiment of this invention. It is a figure explaining the update apparatus which adaptively updates the registration database which concerns on other embodiment of this invention. It is a figure explaining the update apparatus which adaptively updates the registration database which concerns on other embodiment of this invention. It is a figure explaining the update apparatus which adaptively updates the registration database which concerns on other embodiment of this invention. It is an operation flowchart which showed the method of adaptively updating the registration database which concerns on one Embodiment of this invention. It is a block diagram of the update device of the registration database which concerns on one Embodiment of this invention. It is a figure explaining the method of recognizing the vein pattern which concerns on one Embodiment of this invention.

The specific structural or functional description disclosed in the present invention is merely exemplified for the purpose of explaining the embodiment. The present invention may be modified into various different embodiments and the scope of the invention is not limited to the disclosed embodiments.

Terms such as first or second are used to describe the various components, but these terms should be interpreted only for the purpose of distinguishing one component from the other. For example, the first component may be named the second component, and similarly the second component may be named the first component.

When it is mentioned that one component is "connected" or "connected" to another component, it may be directly linked to or connected to another component. It should be understood that there may be other components in between. On the other hand, when it is mentioned that one component is "directly linked" or "directly connected" to another component, it should be understood that there is no other component in between. Expressions that describe the relationships between the components, such as "between" and "directly between" or "adjacent to" and "directly adjacent to", should be interpreted similarly. is there.

A singular expression includes multiple expressions unless they have distinctly different meanings in the context. As used herein, terms such as "including" or "having" attempt to specify the existence of features, numbers, steps, actions, components, parts, or combinations thereof described herein. It must be understood as not prescribing the possibility of existence or addition of one or more other features, numbers, steps, actions, components, components, or combinations thereof.

Unless defined as different, all terms used herein, including technical or scientific terms, shall be generally understood by those with ordinary knowledge in the technical field to which the embodiment belongs. Has the same meaning. Commonly used, pre-defined terms should be construed as having a meaning consistent with the meaning they have in the context of the relevant technology and are ideal unless expressly defined in this application. It cannot be interpreted as having or overly formal meaning.

The embodiment of the present invention described below can be used to perform an operation of recognizing an input image including biological information, and can be used to update a registration database. For example, a fingerprint recognition device, a vein recognition device, an iris recognition device, or the like can perform an operation to authenticate or identify a user, and the authenticated or identified biometric information can be used for updating a conventional registration database.

Embodiments of the present invention can be realized in various forms of products such as personal computers, laptop computers, tablet PCs, smartphones, televisions, smart home appliances, intelligent automobiles, kiosks, and wearable devices. For example, embodiments of the present invention may be applied to authenticate users in smartphones, mobile devices, smart home systems and the like. In the same manner, the embodiment of the present invention may be applied to a payment service using user authentication, an intelligent automobile system, and the like. Embodiments of the present invention can provide robust user authentication results even when the user's finger is small or the fingerprint input is inaccurate in the scenarios described above. Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. The same reference numerals presented in each drawing indicate the same members.

Method for updating the registration database using the similarity based on the size of the overlapping region FIG. 1 is a diagram illustrating an operation of acquiring an input image including biometric information according to an embodiment of the present invention. Hereinafter, for convenience of explanation, it is assumed that the biological information is a fingerprint. However, the embodiments of the present invention are similarly applied to various biological information that can be recognized in image forms such as veins and irises.

Referring to FIG. 1, the sensor 110 according to an embodiment of the present invention detects a user's fingerprint. For example, the sensor 110 may include a plurality of detection elements. The plurality of detection elements may be arranged in an array or matrix structure. The sensor 110 detects the fingerprint input in the form of an analog signal using a plurality of detection elements. The sensor 110 converts the analog signal detected by using the analog-to-digital converter into a digital image. Hereinafter, the input image may be referred to as a converted digital image.

Here, the size of the sensor 110 may be different from the size of the user's finger 120. For example, when the sensor 110 is mounted on a mobile device, the size of the sensor 110 can be smaller than the size of the user's finger 120 in order to satisfy the structural constraints of the mobile device. In this case, the sensor 110 can detect only a part of the fingerprint of the user's finger 120. Hereinafter, the input image detected by the sensor 110 is a partial image of the user fingerprint.

The sensor standard differs depending on the manufacturer of the sensor 110, and the sensor standard includes the size of the sensor array, the resolution of the sensor, and the like. The size of the sensor array indicates the size of the detection area formed by the plurality of detection elements. For example, the size of the sensor array may be 1 cm (length) x 3 cm (width). The resolution of the sensor indicates the number of detection elements per unit length or unit area.

An embodiment of the present invention provides a method and an apparatus for updating a registration database (enrollment database: enrollment DB). The registration database includes images that have already been registered, and the registered image is a partial image of the fingerprint of the registered user (enrollment user). To update the registration database, for example, add a new registration image to the registration database, replace a part of the registration image included in the registration database with a new registration image, or part of the registration image included in the registration database. Including removing etc. The registration database may be referred to as a template.

With reference to FIG. 2, the recognition rate increases in proportion to the registered area. The registered area means the size of the user's fingerprint area covered by the registered images contained in the registered database. As an example, when the registered image included in the registration database covers the first registered area 130, the recognition rate is 93%, and when the registered image included in the registered database covers the second registered area 140, it is recognized. The rate is 99%.

In the embodiment of the present invention, the registration database is updated in the direction of increasing the registration area. For example, the registered image included in the registration database at the first time corresponds to the first registered area 130. The first time point may be the time when the user first registers the fingerprint in order to use the electronic device to which the embodiment of the present invention is applied.

The user inputs a fingerprint for user authentication while using the electronic device. In the embodiment of the present invention, the registration area is increased by updating the registration database using the input image in which the user authentication is successful. For example, the registered image included in the registration database at the second time point corresponds to the second registered area 140. The second time point may be a time point when the registration database is updated using the fingerprint image input for user authentication.

According to the embodiment of the present invention, the convenience of the user for fingerprint registration can be improved. For example, in the embodiment of the present invention, only a predetermined number and / or a selected number of registered images are registered in the initial registration stage, and the user updates the registration database in an unrecognizable state in the actual use stage. Can provide the technology to do.

As described in detail below, an embodiment of the present invention determines the maximum similarity based on the size of the overlapping region between the input image and the registered image. By deciding whether to register the input image in the registration database using the determined maximum similarity, it is possible to register a new fingerprint area as compared with the already registered registered image.

In the embodiment of the present invention, the size of the registered database is changed by replacing the replacement image and the input image selected from the registered images based on the similarity based on the size of the overlapping area between the registered images. Even without it, more various registered images can be secured. Further, in the embodiment of the present invention, the outlier reflects the registered fingerprint by comparing the maximum similarity with the preset and / or selected lower threshold value and registering the input image in the registration database. Can be prevented.

Further, in the embodiment of the present invention, it is possible to prevent the fingerprint recognition rate from being lowered by excluding the input image containing less fingerprint information from the authentication target. In the embodiment of the present invention, an erroneously input fingerprint (incorporate fingerprint), a damaged fingerprint, or the like is registered as a registered image by considering registration of an input image for which user authentication has been accepted in a registration database. However, the fingerprint recognition rate of the fingerprint recognition device can also be improved.

Not only that, embodiments of the present invention can provide techniques for adaptively updating a registration database based on the time required to authenticate an input image. An embodiment of the present invention can also provide a technique for utilizing an already calculated result for updating a registration database in order to authenticate an input image.

FIG. 3 is a flowchart showing a method of updating the registration database according to the embodiment of the present invention. The method of updating the registration database is performed by the registration database update device (hereinafter, “update device”) according to the embodiment of the present invention. The renewal device may be included in the fingerprint authenticator, for example, or may be configured by a separate device.

The update device may be embodied in hardware or may be embodied in a combination of hardware that runs the software. When a processor issues instructions related to a function stored on a computer recording medium, the software-embodied function is performed by the processor. The processor and hardware may exchange signals via an input / output bus or the like.

Referring to FIG. 3, the updater receives an input image including biometric information (S210). The updater receives an input image from a sensor that detects biological information. The input image includes biological information such as fingerprint information, vein information, and iris information.

The updater determines the maximum similarity based on the size of the overlapping region between the input image and the registered image (S220). The updater calculates the degree of similarity between the input image and the registered image based on the size of the overlapping area between the input image and the registered image. For example, assuming that the number of registered images is 15, the updater has a similarity between the input image and the first registered image, a similarity between the input image and the second registered image, ... Fifteen similarities, such as the similarity between and the fifteenth registered image, can be calculated. The updater can determine the similarity having the largest value out of the calculated 15 similarity as the maximum similarity.

Each calculated similarity according to an embodiment of the present invention not only reflects how similar the two images are within the overlapping region, but also reflects the size of the overlapping region. For example, the first case where two images are the same in the overlapping area but the size of the overlapping area is very small, and the second case where the two images are not the same in the overlapping area but the size of the overlapping area is very large. A case can be assumed. The similarity of the first case can be calculated to be higher than the similarity of the second case when only considering how similar the two images are within the overlapping region. On the other hand, according to the embodiment of the present invention, the similarity of the second case can be calculated higher than the similarity of the first case.

A method of calculating the similarity between the registered image and the input image by the updater will be described with reference to FIG. 4, and a method of determining the maximum similarity will be described with reference to FIG. ..

With reference to FIG. 3, the updater compares the maximum similarity with the upper threshold based on the size of the overlapping region between the registered images (S230). The upper threshold value may include a second maximum similarity having the largest value among the similarity based on the size of the overlapping region between the registered images. For example, the higher threshold may be a similarity based on the size of the overlapping region between the two registered images that overlap most of the registered images.

The update device registers the input image in the registration database according to the comparison result in step S230 (S240). The update device determines whether or not the maximum similarity determined between the input image and the registered image is smaller than the upper threshold value. When the maximum similarity is smaller than the upper threshold value, the update device updates the registration database in the direction of increasing the registration area by registering the input image in the registration database.

Here, "the maximum similarity is smaller than the upper threshold value" means that the similarity based on the overlapping area between the input image and the registered image is lower than the similarity based on the overlapping area between the registered images. To do. "The maximum similarity is smaller than the upper threshold" may be understood in the sense that the input image contains a certain level or more of a new fingerprint area that does not overlap with the registered image registered in the registration database.

When the maximum similarity is smaller than the upper threshold value, the update device adds the input image to the registered image registered in the registration database. Alternatively, the updating device selects a replacement image to be replaced with the input image (replace) from the registered images, replaces the input image with the replacement image, and registers the input image.

The updater cannot update the registration database if the maximum similarity is greater than or equal to the higher threshold. For example, the updater can keep the replacement image as a registration image in the registration database as it is if the maximum similarity is greater than or equal to the higher threshold.

According to an embodiment of the invention, the renewal device compares the maximum similarity with a preset subthreshold. The update device registers the input image in the registration database based on the comparison result with the lower threshold value. When the maximum similarity is larger than the lower threshold value, the update device registers the input image in the registration database. The "lower threshold" may be understood as the minimum similarity required in order for the input image to be recognized as the fingerprint of the user corresponding to the recognition target. Here, "the maximum similarity is larger than the lower threshold value" may be understood in the sense that the input image includes a region overlapping with the registered image to be shown at a certain level or higher for the same user. The updater prevents the outliers from being reflected as the registered fingerprint by registering the input image in the registration database when the maximum similarity is greater than the lower threshold.

In an embodiment of the present invention, the registered database is updated only when the maximum similarity falls between the upper threshold and the lower threshold, thereby preventing the registered user from being contaminated by the fingerprints of other users. It is possible to provide a technique for improving the fingerprint recognition rate.

FIG. 4 is a flowchart showing a method of calculating the similarity according to the embodiment of the present invention.

Referring to FIG. 4, the updating device according to the embodiment of the present invention calculates the degree of correlation within the overlapping region between the registered image and the input image of any one of the registered images (S310). The degree of correlation within the overlapping region may be, for example, a normalized cross-correlation (NCC) value.

The updater calculates the size of the overlapping area between any one of the registered images and the input image (S320). The size of the overlapping area may also be expressed as "sharing image area". The updater calculates the shared area ratio between any one of the registered images and the input image using the shared image area.

The updater calculates the degree of similarity between any one of the registered images and the input image by using the degree of correlation and the size of the overlapping region (S330). For example, the similarity is calculated by the product of the normalized cross-correlation value and the shared image area.

FIG. 5 is a diagram for explaining a method for determining the maximum similarity according to an embodiment of the present invention. With reference to FIG. 5, a registered image 410 and an input image 420 are shown.

The update device according to the embodiment of the present invention calculates the similarity between the input image and the registered image, and determines the similarity having the largest value among the similarity as the maximum similarity. For example, when the number of registered images is 15 and the first registered image out of 15 is the registered image 410, the method by which the updater calculates the similarity between the registered image 410 and the input image 420 is as follows. It is a street.

The updater calculates the movement vector and rotation angle between the registered image 410 and the input image 420 by image registration between the registered image 410 and the input image 420. Image alignment is a processing method that transforms images that are different from each other and shows them in one coordinate system. By aligning the images, it is possible to understand how the images obtained by different measurement methods correspond to each other. Image alignment can be divided into a spatial domain method and a frequency domain method.

The spatial area method is a method of matching the pattern and characteristics of the pixel intensity of an image in the space of the image. When the number of feature point pairs contained in the two images is larger than the minimum number required for the image deformation, for example, the image deformation between the two images is detected by using the RANSAC (random sample consensus) method. May be good. Further, the frequency domain method is a method of directly detecting the parameter required for deformation between two images in the frequency domain. The parameters required for the deformation that can be detected here are, for example, the movement of the image, the rotation angle, and the change in size.

The updater calculates the normalized cross-correlation value using the calculated movement vector and rotation angle. For example, the normalized cross-correlation value between the two images I ₁ and I ₂ is calculated using the following equation (1).

ここで、Ｗは、画像Ｉ_１と画像Ｉ_２との間の重複領域を示す。ｉは画像のＸ軸座標を、ｊは画像のＹ軸座標を示し、ｘはＸ軸移動値、ｙはＹ軸移動値を示す。Ｉ_１（ｉ、ｊ）は画像Ｉ_１の（ｉ、ｊ）座標におけるピクセル値を示し、画像Ｉ_２（ｘ＋ｉ、ｙ＋ｊ）は画像Ｉ_２の（ｘ＋ｉ、ｙ＋ｊ）座標におけるピクセル値を示す。画像Ｉ_１は登録画像４１０として、画像Ｉ_２は入力画像４２０として理解されてもよい。

Here, W indicates an overlapping region between the image I ₁ and the image I ₂ . i indicates the X-axis coordinate of the image, j indicates the Y-axis coordinate of the image, x indicates the X-axis movement value, and y indicates the Y-axis movement value. I ₁ (i, j) shows the pixel value at the (i, j) coordinate of the image I ₁ , and the image I ₂ (x + i, y + j) shows the pixel value at the (x + i, y + j) coordinate of the image I ₂ . Image I ₁ may be understood as the registered image 410 and image I _{2 as} the input image 420.

The updater calculates the size of the overlapping area between the registered image 410 and the input image 420. The size of the overlapping area may be the shared image area. The updater calculates the shared area ratio between the registered image 410 and the input image 420 using the shared image area. The shared area ratio can be obtained, for example, by the following formula (2).

The update device calculates the shared image area with respect to the total image area of the registered image 410 and the input image 420 by the shared area ratio. Each of the registered image 410 and the input image 420 may be understood to have the same overall image area. According to the embodiment of the present invention, the updating device can also calculate the shared area ratio by weighting the shared image area.

The updater calculates the degree of similarity between the registered image 410 and the input image 420 using the normalized cross-correlation value and the shared area ratio. The degree of similarity between the registered image 410 and the input image 420 can be determined by the following equation (3).

_Soverlap (I ₁ , I ₂ ) = ncc (I ₁ , I ₂ ) x r _overlap (I ₁ , I ₂ ) (3)

The updater calculates the degree of similarity between the registered image 410 and the input image 420 by multiplying the normalized cross-correlation value and the shared area ratio.

The update device calculates the similarity between the 14 registered images and the input image 410 excluding the registered image 410 for which the similarity has been calculated by the same method described above. The updating device determines the similarity having the largest value among the calculated 15 similarity as the maximum similarity.

FIG. 6 is a flowchart showing a method of updating the registration database according to another embodiment of the present invention. Referring to FIG. 6, the updater receives an input image including biometric information (S510). The updater calculates the similarity between the input image and the registered image based on the size of the overlapping area between the input image and the registered image (S520), and the largest value among the calculated similarities. The similarity having the above is determined to be the maximum similarity (S530). The updating device selects a replacement image from the registered images based on the similarity based on the size of the overlapping region between the registered images (S540). A method for the updater to select a replacement image will be described with reference to FIGS. 7 to 9.

The updater determines whether the maximum similarity is less than the upper threshold (S550). The upper threshold may be, for example, the second maximum similarity having the largest value of the similarity based on the size of the overlapping region between the registered images. As a result of determination in step S550, if the maximum similarity is smaller than the upper threshold value, the updater updates the registration database by replacing the replacement image with the input image (S560). As a result of determination in step S550, if the maximum similarity is greater than or equal to the upper threshold value, the updater keeps the replacement image as it is in the registration database (S570).

FIG. 7 is a flowchart showing a method of updating the registration database according to another embodiment of the present invention. According to one embodiment of the present invention, the updating device temporarily selects a candidate image that duplicates and stores the most information, and selects a candidate image that has more information that overlaps with the remaining registered images among the candidate images. Final selection as a replacement image.

Referring to FIG. 7, the updating device according to the embodiment of the present invention selects a candidate image to be replaced from the registered images based on the similarity based on the size of the overlapping region between the registered images (S610). .. The reason for selecting a registered image having a high degree of similarity as a candidate image to be replaced in one embodiment of the present invention is that it is not necessary to have duplicate registered images that are substantially the same or substantially similar to the registered database, and they are duplicated. This is to remove the registered image.

The updating device detects, for example, the registered image corresponding to the maximum similarity from the similarity based on the size of the overlapping region between the registered images. The update device is selected as a candidate image to be replaced with the registered image corresponding to the maximum similarity. Since the similarity based on the size of the overlapping region is calculated between two images different from each other, the candidate image corresponding to the maximum similarity may be two registered images.

The updating device accumulates the similarity between each of the candidate images selected in step S610 and the remaining registered images (S620). Here, the "remaining registered image" may be understood as the remaining registered image excluding the candidate image to be replaced from the registered image. For example, the updater can calculate the cumulative similarity between the candidate image and the remaining registered images using equation (4).

ここで、ｓ_{ｏｖｅｒｌａｐ}（ｉ、ｊ）は、ｉ番目の登録画像とｊ番目の登録画像との間の類似度を示す。ｊ番目の登録画像は、２個の候補画像のうちの１つである。Ｍは登録画像の数である。本発明の他の実施形態によれば、ｉ＝ｊの場合、ｓ_{ｏｖｅｒｌａｐ}（ｉ、ｊ）は、ｊと関係なく、同一の値（例えば、最大値）として算出されるため、ｓ（ｊ）のΣを算出するとき、ｉ≠ｊの条件が省略されてもよい。

Here, _soverlap (i, j) indicates the degree of similarity between the i-th registered image and the j-th registered image. The j-th registered image is one of the two candidate images. M is the number of registered images. According to another embodiment of the present invention, for the case of _{i = j, s overlap (i} , j) is independent of the j, calculated as the same value (e.g., maximum value), s (j) When calculating Σ of, the condition i ≠ j may be omitted.

For example, it is assumed that the number of registered images is 18, and the candidate images to be replaced are the 13th registered image and the 18th registered image. The updater calculates the cumulative similarity (s13) between the 13th registered image and the remaining registered images, and the cumulative similarity (s18) between the 18th registered image and the remaining registered images. Is calculated.

The updating device selects a candidate image having a high cumulative similarity from the candidate images as a replacement image (S630). The updater selects a candidate image having a higher value of the accumulated similarity (s13) and the accumulated similarity (s18) as the replacement image.

8 and 9 are diagrams for explaining a method of selecting a replacement image according to an embodiment of the present invention. With reference to FIGS. 8 and 9, graphs 730 and 810 are shown in which the degree of similarity between the registered image 710 and the registered image 710 is projected onto a three-dimensional matrix. In the graphs 730 and 810, the X-axis and the Y-axis show the index of the registered image, and the Z-axis shows the similarity between the registered images.

For example, it is assumed that the similarity between the registered image 713 and the registered image 716 of the registered images 710 is the maximum similarity. The updating device detects the registered image having the maximum similarity by detecting the registered image corresponding to the point 815 displayed in the graph 810 of FIG.

Graph 830 shows the cumulative similarity of the candidate images. Graph 830 is the result of one-dimensional projection of graph 810 of FIG. 9 projected on the three-dimensional matrix. In graph 830, the X-axis shows the index of the registered image and the Y-axis shows the cumulative similarity.

As an example, when the index of the candidate image is 13, 18, the updater has a cumulative similarity (s18) corresponding to the 18th registered image and a cumulative similarity (s13) corresponding to the 13th registered image. To compare. Since "the cumulative similarity is higher" means that the registered image overlaps more with other registered images, the registered image with more overlap is selected as the replacement image. The updater selects the 18th registered image, which has a higher cumulative similarity, as the replacement image. The thirteenth registered image is retained in the registration database because it is more likely to cover a new area than the eighteenth registered image.

FIG. 10 is a flowchart showing a method of updating the registration database according to the embodiment of the present invention.

Referring to FIG. 10, the updater according to an embodiment of the present invention receives an input image (S910) and extracts a valid image from the input image (S920). The updating device determines whether or not the size of the effective image is larger than the preset first threshold value (S930). As a result of the determination in step S930, if the size of the effective image is larger than the preset first threshold value, the updater calculates the authentication score (S940). If the size of the valid image is less than or equal to the preset first threshold, the updater rejects the authentication and provides the user with a phrase that prompts the user to enter a new input image. To receive (S910).

In step S940, the renewal device calculates the authentication score by various methods. For example, the renewal device may calculate the number of authentication points considering the size of the overlapping area, or may calculate the number of authentication points not considering the size of the overlapping area. The update device divides the input image into small blocks, then calculates the block-specific score, and integrates the block-specific scores to calculate the authentication score.

In step S950, the update device determines whether or not the authentication score is greater than the preset authentication threshold. The preset authentication threshold is, for example, the number of authentication points between registered images. The update device authenticates the user for the input image by comparing the authentication score with the authentication threshold. As a result of the determination in step S950, if the authentication score is smaller than or the same as the preset authentication threshold value, the updater may determine that the user authentication has failed based on the input image. As a result of the determination in step S950, if the authentication score is larger than the preset authentication threshold value, the update device accepts the user authentication. It is considered to update the input image for which authentication is accepted to the registered image.

The updater determines the maximum similarity (Smax) based on the similarity calculated between the input image for which user authentication has been accepted and the registered image (S970). In step S970, the calculated similarity is the similarity based on the size of the overlapping region. In the following, the input image may be understood as an input image for which user authentication has been accepted.

The updater determines whether the maximum similarity determined in step S970 is greater than the lower threshold and smaller than the upper threshold (S980). The upper threshold is the maximum similarity based on the size of the overlapping region between the registered images contained in the registration database 905. For example, the updating device detects the registered image pair having the highest similarity among the registered images, and defines the similarity of the detected registered image pair as the second maximum similarity. The second maximum similarity is used as the upper threshold. In one embodiment of the present invention, when the input image and the registered registered image overlap in a certain area and / or a selected area or more, in other words, when the maximum similarity is larger than the lower threshold value, the input image is registered in the registration database. By registering with, the outline is prevented from being reflected as a registered fingerprint.

If the maximum similarity is greater than the lower threshold and less than the upper threshold, the updater updates the registration database 905 in step S960. The updater may add the input image to the registration database 905. Alternatively, it may be replaced with any one of the registered images in the registration database 905. In this case, the updating device selects a replacement image from the registered images already registered in the registration database 905 based on the similarity based on the size of the overlapping area between the registered images (S960). If the maximum similarity is less than or greater than the upper threshold, the updater cannot update the registration database 905 in step S960.

Method for adaptively updating the registration database according to the time required for authentication FIG. 11 is a diagram illustrating an authentication process of biometrics according to an embodiment of the present invention. Referring to FIG. 11, the biometrics certification process according to an embodiment of the present invention includes a certification step and an adaptive renewal step.

The input image is authenticated at the authentication stage. For example, in the authentication stage, the registered image included in the registration database is compared with the input image, and the score is calculated. If the calculated score is greater than the pre-determined authentication threshold, the authentication may succeed, otherwise the authentication may fail.

In the adaptive update stage, the operation for updating the registration database is performed. For example, in the adaptive update stage, a method is performed in which the registration database is updated using the similarity based on the size of the overlapping area described with reference to FIGS. 1 to 10.

Although not shown, the biometrics certification process may further include a pretreatment step prior to the certification step. In the pre-processing stage, a pre-processing operation for authenticating the input image is performed. The preprocessing operation is an operation of improving the input image, but may include an operation of detecting an effective area in the input image.

FIG. 12 is a diagram illustrating adaptive update of the registration database according to an embodiment of the present invention. Referring to FIG. 12, the biometrics authentication process has a maximum permissible time limit (hereinafter, maximum permissible time). If the maximum permissible time is determined, the certification and adaptive renewal stages must be performed within the maximum permissible time.

According to one embodiment of the present invention, at the authentication stage, a method of reducing the search candidate group via a multi-stage authentication method and performing authentication is applied. The multi-stage authentication method is composed of two stages, a coarse search and a fine search, or is composed of three or more stages. In this case, in each stage, the search candidate group is reduced by the minimum calculation, and precise matching is performed in the final stage.

In the multi-stage authentication method, if the calculation result is a certain score or more even in the stage before the final stage, the authentication or search operation is stopped and it is decided to accept the operation at an early stage. Therefore, the authentication accuracy can be guaranteed and the authentication speed can be improved.

When the multi-stage authentication method is used, the time required in the authentication stage differs from case to case. For example, the time required for authentication in Case A is longer than the time required for authentication in Case D.

If the maximum allowed time is determined, the longer the time required for authentication, the shorter the time allowed for adaptive renewal. For example, the time allowed for adaptive renewal in Case A is shorter than the time allowed for adaptive renewal in Case D.

An embodiment of the present invention can provide a technique for selecting the best registered database update method by predicting the time required in the authentication stage and predicting the time that can be used in the future. The embodiment of the present invention can guarantee the authentication performance while minimizing the authentication time.

For example, when the registration database is updated only when the time for updating the registration database is sufficiently guaranteed as in case D, the performance of the fingerprint recognition device is not improved. This is because the input image that is accepted early is likely to be very similar to the registered image. As described with reference to FIG. 2, the recognition rate of the fingerprint recognition device increases in proportion to the registered area. Since the input image, which is very similar to the registered image, contains almost no information other than the information contained in the registered image, the registered area does not increase even if the input image is added to the registered database.

In the embodiment of the present invention, not only when the time for updating the registered database is sufficiently guaranteed as in case D, but also when the time for updating the registered database is not sufficient as in case A. We will provide the technology to update the registration database with the best efforts.

FIG. 13 is an operation flowchart showing a method of adaptively updating the registration database according to the embodiment of the present invention. Referring to FIG. 13, the updater according to one embodiment of the present invention collects biometric data in step S1310. For example, the updater receives an input image containing fingerprint information from the fingerprint sensor. The renewal device authenticates in step S1320. In some cases, the renewal device may perform preprocessing before step S1320.

If it is determined in step S1330 that the authentication is accepted, the updater estimates in step S1340 the time required to authenticate the input image. For example, the updater uses a timer to measure the number of clocks required for authentication of the input image, and estimates the time required for authentication of the input image from the measured number of clocks. The renewal device also predicts the time available for adaptive renewal from the estimated time.

The update device adaptively updates the registration database based on the time estimated in step S1350. The renewal device controls the amount of computation for the adaptive renewal stage so that all authentication and adaptive renewal stages can be performed within the maximum permissible time.

FIG. 14 is a block diagram showing a multi-stage authenticator 1400 according to an embodiment of the present invention. Referring to FIG. 14, the multi-stage authenticator 1400 according to an embodiment of the present invention includes a plurality of authenticators corresponding to a plurality of stages. For example, the multi-stage authenticator 1400 includes a first authenticator 1410 and a second authenticator 1420. Although not shown in the figure, the multistage authenticator 1400 may include three or more stages.

The first authenticator 1410 authenticates by a coarse search method. The first authenticator 1410 determines whether or not an authentication score (authentication score) equal to or higher than the first authentication threshold value is derived while comparing the input image and the registered image by a coarse search method. The first authenticator 1410 accepts the authentication and ends the authentication operation in response to the determination that the authentication score equal to or higher than the first authentication threshold value has been derived. In this case, acceptance signal (the accept Signal) is output at time _{t 1.}

If no authentication score greater than or equal to a predetermined threshold is derived during the coarse search, the first authenticator 1410 transmits the reduced search space to the second authenticator 1420. For example, the first authenticator 1410 extracts a predetermined number of candidate groups in the order in which the authentication score is excellent from the registered images included in the registration database, and secondly extracts information related to the extracted candidate groups. Output to the authenticator 1420.

FIG. 15 is a diagram illustrating a rough search according to an embodiment of the present invention. With reference to FIG. 15, the first authenticator 1410 calculates the authentication score by comparing the input image and the registered image on a one-to-one basis. As an example, the first authenticator 1410 uses the normalized cross-correlation value of the equation (1) as the authentication score. As another example, the first authenticator 1410 uses the similarity based on the size of the overlapping region of the equation (3) as the authentication score.

Referring to FIG. 14 again, the second authenticator 1420 authenticates in the reduced search space by the fine search method. The second authenticator 1420 determines whether or not an authentication score equal to or higher than the second authentication threshold value is derived while comparing the input image and the registered image by a precise search method. The second authenticator 1420 accepts the authentication and ends the authentication operation in response to the determination that the authentication score equal to or higher than the second authentication threshold value has been derived. In this case, acceptance signal is output at time t _2. Time t ₂ is a time after time t ₁ .

If it is determined that the authentication score equal to or higher than the second authentication threshold cannot be derived in the reduced search space, the second authenticator 1420 ends the authentication operation without authenticating the input image.

FIG. 16 is a diagram illustrating a precise search according to an embodiment of the present invention. Referring to FIG. 16, the second authenticator 1420 divides the input image into a plurality of blocks. The second authenticator 1420 calculates the block authentication score via block pattern registration. The second authenticator 1420 determines whether or not the input image can be authenticated by using the total value or the average value of the top K pieces of the block authentication scores.

17 to 22 are diagrams illustrating an update device that adaptively updates the registration database according to the embodiment of the present invention. Referring to FIG. 17, the update device 1700 according to an embodiment of the present invention includes an authentication module 1710, an update module 1720, a time estimation module 1730, a registration database 1740, and a memory 1750.

The authentication module 1710 authenticates the input image and outputs an acceptance signal or a rejection signal. The time estimation module 1730 estimates the time required to authenticate the input image or the time available for adaptive update in response to the acceptance signal. The update module 1720 updates the registration database 1740. The update module 1720 controls updates to the registration database 1740 based on the time estimated by the time estimation module 1730.

The authentication module 1710 includes a candidate selector 1711, a first authenticator 1712, and a second authenticator 1713. With reference to FIG. 18, candidate selector 1711 receives the input image. The input image is the result of preprocessing. The candidate selector 1711 extracts the first candidate group from the registered images included in the registration database 1740. The candidate selector 1711 extracts the first candidate group by various methods.

The registration database 1740 contains 20 registered images corresponding to the first identifier (id1), 20 registered images corresponding to the second identifier (id2), and 20 registered images corresponding to the third identifier (id3). including. The first identifier, the second identifier, and the third identifier correspond to different fingers of the same user. The 60 registered images are partial images including a part of the fingerprint of the finger.

The candidate selector 1711 extracts the first candidate group by processing the input image and the registered image in the frequency domain. The candidate selector 1711 generates a third image showing the phase correlation between the input image and the registered image in the Fourier space, and calculates the variance of the phase relationship in the third image. When the calculated variance is larger than the predetermined threshold value, the candidate selector 1711 classifies the registered image as the first candidate group.

The candidate selector 1711 transmits information related to the first candidate group to the first authenticator 1712. As will be described in detail below, in order to minimize the amount of calculation in the adaptive update operation, the result already calculated in the authentication operation is utilized. The candidate selector 1711 will store information related to the first candidate group in the memory 1750 provided in advance for the adaptive update operation in the future.

With reference to FIG. 19A, the first authenticator 1712 receives information related to the first candidate group. The first authenticator 1712 authenticates the input image. In the first authenticator 1712, the operation described in FIG. 15 is applied. However, the first authenticator 1712 uses only the first candidate group instead of using all 60 registered images in the registration database 1740 in order to authenticate the input image.

The first authenticator 1712 outputs an acceptance signal if an authentication score equal to or higher than the first authentication threshold value is derived during authentication. The output acceptance signal is transmitted to the time estimation module 1730. The result already calculated by the first authenticator 1712 will be stored in the memory 1750 for the adaptive update operation in the future.

When the authentication score equal to or higher than the first authentication threshold value is not derived, the first authenticator 1712 extracts the second candidate group corresponding to the predetermined number of higher authentication scores. The first authenticator 1712 transmits information related to the second candidate group to the second authenticator 1713. The information related to the second candidate group and the result already calculated by the first authenticator 1712 are stored in the memory 1750 for the adaptive update operation in the future.

With reference to FIG. 19B, the memory 1750 stores the matching information of the registration unit and the number of calculations or the time until authentication. s is ncc according to the formula (1) or _soverlap according to the formula (3), and area is the shared area ratio or the size of the overlapping area according to the formula (2). As an example, the information stored in the first row and first column of the table in memory is the result of comparing the input image with the registered image having an index of 1 among the registered images whose identifier corresponds to the thumb. Is.

Referring to FIG. 20, the second authenticator 1713 receives information related to the second candidate group. The second authenticator 1713 authenticates the input image. In the second authenticator 1713, the operation described in FIG. 16 is applied. The second authenticator 1713 outputs an acceptance signal if an authentication score equal to or higher than the second authentication threshold value is derived during authentication. The output acceptance signal is transmitted to the time estimation module 1730. The result already calculated by the second authenticator 1713 is stored in the memory 1750 for the adaptive update operation in the future. If no authentication score equal to or greater than the first authentication threshold is derived, the first authenticator 1712 outputs a rejection signal.

Referring to FIG. 21, the time estimation module 1730 receives an acceptance signal. The time estimation module 1730 responds to the acceptance signal and estimates the time required for authentication of the input image or the time available for adaptive update. The update module 1720 updates the registration database 1740 based on the input image and the time estimated by the time estimation module 1730.

As an example, the update module 1720 calculates the maximum similarity based on the size of the overlapping area between the input image and the registered image. Here, the update module 1720 inputs the 20 registered images corresponding to the accepted identifiers, instead of determining all the similarities between the 60 registered images and the input images contained in the registration database 1740. Only the degree of similarity with the image is determined. When adding an input image to the registration database, score calculation is required 20 times, and when replacing the input image with any one of the 20 registered images, 19 times excluding the score calculation with the registered image to be replaced. Score calculation is required.

Here, the update module 1720 uses the already calculated results (described in FIGS. 18 to 20) stored in the memory 1750. With reference to FIG. 22, the update module 1720 has already been calculated first stored in memory 1750 to determine the similarity 2000 between the 20 registered images corresponding to the accepted identifiers and the input images. The results are used to determine some 2210 of similarity.

The number of score operations for determining whether or not to update in the update module 1720 is 20 or 19 times. If the score between the input image and the registered image is already calculated by the authentication module 1710 and the calculation result is stored in the memory 1750, the update module 1720 can reduce the calculation amount. For example, if the first similarity between the first registered image and the input image has already been calculated by the first authenticator, the update module 1720 acquires the first similarity from the memory 1750. The above-mentioned operation may be described as a score reuse method.

The update module 1720 calculates a similarity 2220 that has not yet been determined in the best effort based on the time estimated by the time estimation module 1730. For example, the first authenticator 1712 calculates the registration information. The registration information includes a movement vector and a rotation angle using image alignment. The first authenticator 1712 rotates and moves the image to calculate the overlap area ratio and the normalized cross-correlation.

If there is no similar overlapping area in the registration stage, the first authenticator 1712 skips the subsequent calculation process and skips to the next registered image. Although the average authentication time is reduced through such an intermediate end operation, there is no score for the registered image that has been intermediately completed. In this case, the update module 1720 must directly calculate the similarity. Here, the update module 1720 calculates the similarity within the maximum permissible time.

If there is not enough time to calculate all the similarity, the update module 1720 sets the remaining similarity 2230 to a predetermined and / or selected minimum. The predetermined and / or selected minimum value may be smaller than the lower threshold shown in FIGS. 9 and 10. The above-mentioned operation may be described as a selective score reuse method.

The update module 1720 processes the registered images in the order of the index of the registered images or at random in order to calculate the similarity 2000 for a given time. Alternatively, the update module 1720 preferentially processes the registered images selected as the first candidate group or the second candidate group.

The operation of the first authenticator 1712 is represented by the algorithm shown in Table 1 below.

第２認証器１７１３の動作は、下記の表２のアルゴリズムで表現される。

The operation of the second authenticator 1713 is represented by the algorithm shown in Table 2 below.

時間推定モジュール１７３０の動作は、表３のアルゴリズムで整理される。

The operation of the time estimation module 1730 is organized by the algorithms in Table 3.

入力画像を登録画像のうちの１つに置き換える場合、更新モジュール１７２０の動作は、表４のアルゴリズムに整理される。

When replacing the input image with one of the registered images, the operation of the update module 1720 is organized by the algorithms in Table 4.

図２３は、本発明の一実施形態に係る登録データベースを適応的に更新する方法を示した動作フローチャートである。図２３を参照すると、更新装置は、認証ステップＳ２３１０のサブステップＳ２３１１において認証点数（ｓ）が算出される。認証段階Ｓ２３１０のサブステップＳ２３１２において算出された認証点数（ｓ）が認証閾値（Ｔｈ１）より大きいと判断されれば、更新装置は、認証を受諾する。認証が受諾される場合、認証のために演算された結果がメモリ２３３０（ｎｏｎ−ｔｒａｎｓｉｔｏｒｙ ｃｏｍｐｕｔｅｒ−ｒｅａｄａｂｌｅ ｓｔｏｒａｇｅ ｍｅｄｉａ）に格納される。

FIG. 23 is an operation flowchart showing a method of adaptively updating the registration database according to the embodiment of the present invention. With reference to FIG. 23, the update device calculates the authentication score (s) in the sub-step S2311 of the authentication step S2310. If it is determined that the authentication score (s) calculated in the sub-step S2312 of the authentication step S2310 is larger than the authentication threshold value (Th1), the renewal device accepts the authentication. When the authentication is accepted, the result calculated for the authentication is stored in the memory 2330 (non-transition computer-readable storage media).

In the sub-step S2321 and the sub-step S2322 of the update stage S2320, the update device updates the score (S') of the area within the time limit. In substep S2321, the updater determines the score (S') of the individual region for calculating the score (S') of the region. If there is information already stored in the memory 2330, the score (S') of the individual area can be loaded from the memory 2330 by the updater. For example, the score (S') of the individual area of the first registered image, the second registered image, and the 19th registered image corresponding to the thumb can be loaded from the memory 2330. If there is no stored information for the registered image in memory 2330, the updater determines the authentication score s and the overlapping area and is the individual associated with the registered image based on the authentication score (s) and the overlapping area (area). The score (S') of the area can be calculated. For example, the score of the individual area of the third registered image to the eighteenth registered image corresponding to the thumb can be calculated by f (s, area). Here, f (s, area) = s × area may be used.

In sub-step S2322, the updater determines whether the current time t is within the T _{time limit} . When the current time t is within the T _{time limit} , step S2321 is performed by the update device to calculate the score (S') of the individual region for updating the score (S') of the continuous registered image and region. It can be carried out. When the current time t has passed the T _{time limit} , the score of the individual area that has not yet been calculated by the updater may be determined to a value predetermined and / or selected by the updater.

For example, the score of the individual region of the 20th registered image corresponding to the thumb may be set to a predetermined and / or selected minimum value (for example, 0.1).

In substep S2323, whether the _updater falls between a predetermined and / or selected first threshold (Th _{AREA_1} ) and a second threshold (Th _{AREA_2} ) for the region score (S'). Is judged. The score (S') of the region may be calculated by the updater using the equation (5).

ここで、Ｍは予め決定した及び／又は選択された自然数であり、ｓｏｒｔ_{ｄｅｃｅｎｄ}（Ｓ’）は個別領域の点数を降順に整列した結果を示す。式（５）によれば、領域の点数（Ｓ’）は、更新装置によって高い値を有するＭ個の個別領域の点数（Ｓ’）の平均として算出されてもよい。

Here, M is a predetermined and / or selected natural number, and sort _descend (S') indicates the result of arranging the scores of the individual regions in descending order. According to the equation (5), the region score (S') may be calculated as the average of the scores (S') of M individual regions having high values by the updater.

When M is 1, the score (S') of the region corresponds to the maximum similarity. In this case, the sub-step S2323 may directly correspond to the step S980 of FIG.

In step S2323, if it is determined that the score (S') of the area falls between the predetermined first threshold value (Th _{AREA_1} ) and the second threshold value (Th _{AREA_2} ), the registration database is updated by the update device. You may. For example, if the score (S') of the area satisfies the condition of step S2323, the registered image may be added to the registration database.

As an example, the number of registered images contained in the registration database may be changed to the following scenario.

1. 1. At the time of the first user registration, 18 registered images are registered in the registration database.

2. 2. In the process of using fingerprint authentication, add new fingerprint images that meet the update conditions of the registration database to the registration database and register up to 25 registered images.

3. 3. If up to 25 registered images are secured, the registered images will be retained at 25 by replacing them with new fingerprint images that meet the registration database update conditions and registered images.

In order to add the fingerprint image to the registration database, the operation of removing the image already registered in the registration database is inactivated, or to replace the fingerprint image with the registered image, it has already been registered in the registration database. The action of removing the image may be activated.

In sub-step S2323, the first threshold value (Th _{AREA_1} ) corresponds to the upper threshold value, and the second threshold value (Th _{AREA_1} ) corresponds to the lower threshold value. Here, the upper threshold value may be the score of the replacement candidate having the most overlapping information among the registered images. Although not shown in the figure, the upper threshold may not be considered when adding a fingerprint image to the registration database. In this case, in substep S2323, only whether or not the area score (S') is larger than the lower threshold value is determined, and if the area score (S') is larger than the lower threshold value, the corresponding fingerprint image is added to the registration database. May be done.

FIG. 24 is a block diagram of a registration database update device according to an embodiment of the present invention.

Referring to FIG. 24, the update device according to an embodiment of the present invention includes a sensor 2410, a memory 2420, a processor 2430, and a receiver 2440. The update device further includes a template management unit 2450. The sensor 2410, the memory 2420, the processor 2430, the receiving unit 2440, and the template management unit 2450 communicate via the bus 2460.

The sensor 2410 may be the fingerprint sensor 110 shown in FIG. The sensor 2410 captures a fingerprint image by a well-known method (for example, a method of converting an optical image into an electric signal). The image is output to the processor 2430 via the receiving unit 2440. The processor 2430 performs the authentication operation and the update operation described in FIGS. 1 to 23 by issuing an instruction stored in the memory 2420.

As another example, the sensor 2410 may include a sensor that recognizes the user's vein pattern. Sensor 2410 extracts vein patterns from the skin on the back of the user's hand. The sensor 2410 uses infrared illumination and a filter to generate a brightness contrast of the blood vessels to the skin and then acquires an image containing the vein pattern. The processor 2430 updates the registration database with the input image corresponding to the vein pattern and the image of the registered vein pattern.

Referring to FIG. 25, the method of authenticating the fingerprint pattern with reference to FIGS. 1 to 23 is directly applied to the method of authenticating the vein pattern. For example, the user's smartwatch 2510 detects the user's vein pattern 2520. The input image 2530 generated by detecting the user's vein pattern 2520 is authenticated through the coarse search 2550 and the fine search 2580. As a result of the coarse search 2550, the candidate group 2560 is extracted from the registered images included in the registration database 2540. The input image 2530 is blocked 2570 for the fine search 2580.

As another example, the sensor 2410 may include a sensor that recognizes the user's iris pattern. Sensor 2410 scans or captures the iris pattern between the user's pupil and the sclera (the area of the white of the eye). The processor 2430 updates the registration database with the input image corresponding to the iris pattern and the image of the registered iris pattern.

The memory 2420 stores a registration database including registered images. The memory 2420 stores an input image including biometric information captured by the sensor 2410, a newly registered input image, and the like. The memory 2420 may be a volatile memory or a non-volatile memory.

Processor 2430 determines the maximum similarity based on the size of the overlapping area between the input image and the registered image. The processor 2430 causes the input image to be registered in the registration database according to the comparison result of the upper threshold based on the maximum similarity and the size of the overlapping region between the registered images. The processor 2430 calculates the similarity between the input image and the registered image based on the size of the overlapping area between the input image and the registered image, and maximizes the similarity having the largest value among the similarities. Determined as similarity.

The processor 2430 calculates the degree of correlation in the overlapping area between the registered image and the input image of any one of the registered images, and determines the size of the overlapping area between the registered image and the input image of any one of them. calculate. Processor 2430 uses the degree of correlation and the size of the overlapping region to calculate the degree of similarity between any one of the registered images and the input image. In addition to this, the processor 2430 may perform at least one method described in FIGS. 1 to 23.

The processor 2430 executes the program by executing the program code and controls the update device. The program code executed by the processor 2430 is stored in the memory 2420. The updater connects to an external device (eg, a personal computer or network) via an input / output device (not shown) to replace the data.

The receiving unit 2440 receives an input image including biometric information detected by the sensor 2410.

The template management unit 2450 selects a replacement image from the registered images based on the similarity based on the size of the overlapping area between the registered images. The template management unit 2450 selects a candidate image to be replaced from the registered images based on the similarity based on the size of the overlapping area between the registered images. The template management unit 2450 accumulates the similarity between each of the candidate images and the remaining registered images, and selects the candidate image having a high accumulated similarity as the replacement image among the candidate images. According to the embodiment of the present invention, the template management unit 2450 is not composed of separate elements as shown in FIG. 24, and the operation of the template management unit 2450 is performed by the processor 2430 that executes the instruction word stored in the memory 2420.

Update devices include mobile devices such as mobile phones, smartphones, PDAs, tablet PCs and laptop computers, computer devices such as tablet PCs and netbooks, and electronic products such as TVs, smart TVs and security devices for gate control. , Various electronic systems may be included.

The above-described embodiment is embodied by a combination of hardware components, software components, and / or hardware components and software components. For example, the devices, methods and components described in this embodiment include, for example, a processor, a controller, an ALU (arithmetic logic unit), a digital signal processor, a microprocessor, an FPGA (field program matrix array), and a PLL. It is embodied using one or more general purpose computers or special purpose computers, such as different devices that execute and respond to (programmable logistic unit), microprocessors, or instructions. The processing device executes an operating system (OS) and one or more software applications running on said operating system. The processing device also accesses, stores, manipulates, processes, and generates data in response to software execution. For convenience of understanding, one processing device may be described as being used, but for those with ordinary knowledge in the art, the processing device may have multiple processing elements and / or It can be seen that it contains multiple types of processing elements. For example, the processing device includes a plurality of processors or one processor and one controller. Other processing configurations, such as a parallel processor, are also possible.

The software includes computer programs, codes, instructions, or a combination of one or more of these, configures the processor to operate as desired, and commands the processor independently or in combination. The software and / or data is interpreted by the processing device and is any type of machine, component, physical device, virtual device, computer recording medium or device, or signal transmitted to provide instructions or data to the processing device. It is embodied permanently or temporarily through the waves. The software is distributed on a networked computer system and is stored or executed in a distributed manner. The software and data are stored on one or more computer-readable recording media.

The method according to this embodiment is embodied in the form of program instructions implemented via various computer means and recorded on a computer-readable recording medium. The recording medium includes program instructions, data files, data structures, etc. alone or in combination. The recording medium and program instructions may be specially designed and configured for the purposes of the present invention, and may be known and usable by those skilled in the art of computer software. Good. Examples of computer-readable recording media include magnetic media such as hard disks, floppy (registered trademark) disks and magnetic tapes, optical recording media such as CD-ROMs and DVDs, and magnetic-optical such as floppy disks. It may include a medium and a hardware device specially configured to store and execute program instructions such as ROM, RAM, flash memory and the like. Examples of program instructions include not only machine language code as generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

As described above, although the embodiments have been described by the limited drawings, various technical modifications and modifications can be applied based on the above description by anyone having ordinary knowledge in the art. .. As an example, the techniques described may be performed in a different order than the methods described, and / or components such as the described systems, structures, devices, circuits, etc. may be combined or combined in a different manner than the methods described. Appropriate results can be achieved even if substituted or replaced by other components or equivalents. Therefore, the scope of the present invention is not limited to the disclosed embodiments, but is defined by the scope of claims and the equivalent of the scope of claims.

2410: Sensor 2420: Memory 2430: Processor 2440: Receiver 2450: Template management unit 2460: Bus

Claims (27)

It is a method to update the biometric data of the registration database.
During the first length of time, the step of authenticating the input image using the registration database, the authentication comprising generating information for authentication of the input image,
Including the step of updating the registration database based on the time of the first length and the information for authenticating the input image .
The step of adaptively updating the registration database is
A step of determining the similarity based on the size of the overlapping area between the registered image in the registration database and the input image, and
Including a step of determining whether or not to register the input image in the registration database based on the similarity.
The step of determining the similarity is
A step of determining at least a part of the similarity based on the information for authenticating the input image.
A step of determining at least a part of the similarity to the minimum value based on the time of the first length, and
Including at least one of
Method. It is a method to update the biometric data of the registration database.
During the first length of time, the step of authenticating the input image using the registration database, the authentication comprising generating information for authentication of the input image,
Including the step of updating the registration database based on the time of the first length and the information for authenticating the input image.
The step of authenticating the input image is
A step of determining the size of the overlapping area between the input image and the registered image in the registration database, and
A step of determining the degree of correlation within the overlapping region between the input image and the registered image, and
The step of determining whether the degree of correlation is greater than or equal to the threshold value,
When the degree of correlation is equal to or higher than the threshold value, the step of outputting an acceptance signal and
A step in which the size of the overlapping area and the degree of correlation are stored in a memory, and the size of the stored overlapping area and the degree of correlation correspond to the registered image.
including,
Method. The step of deciding whether or not to register in the registration database is
A step of determining whether or not the maximum similarity among the similarity is equal to or lower than the upper threshold value, and
The step of determining whether or not the maximum similarity is equal to or higher than the lower threshold value, and
Including at least one of
The method according to claim 1 . The upper threshold corresponds to the maximum similarity based on the overlapping area between the registered images.
The method according to claim 3 . The information for authenticating the input image includes the degree of correlation within the overlapping area between the input image and the registered image in the registered database, and the size of the overlapping area between the input image and the registered image. , Including at least one of
The method according to any one of claims 1 to 4 . The step of authenticating the input image further includes a step of storing the time of the first length in the memory.
The method according to claim 2 . The step of authenticating the input image includes a step of authenticating the input image using a multi-stage authentication method.
The method according to claim 2 or 6 . The adaptively updating step
A step of estimating the second length time for updating the registration database based on the first length time, and
A step of adaptively updating the registration database based on the time of the second length,
including,
The method according to claim 1. The step of estimating the second length time is based on the difference between the third length time, which is the length time allocated for biometrics, and the first length time. Including the step of calculating the time of the second length.
The method according to claim 8 . The step of adaptively updating the registration database is
The step of adding the input image to the registration database and
A step of replacing any one of the registered images in the registration database with the input image,
Including at least one of
The method according to claim 1. The replacement step includes selecting replacement images based on similarity based on the size of overlapping regions between the registered images.
The method according to claim 10 . It is a method to update the biometric data of the registration database.
Steps to receive input images containing biometric information,
A step of determining the first maximum similarity based on the size of the overlapping region between the input image and the registered image in the registration database.
A step of determining whether or not the first maximum similarity reaches an upper threshold value based on the size of the overlapping region between the registered images, and
Including a step of registering the input image in the registration database based on a determination result as to whether or not the first maximum similarity reaches the upper threshold value .
The upper threshold includes a second maximum similarity based on the size of the overlapping region between the registered images.
Method. The step of determining the first maximum similarity is
A step of determining the degree of similarity between the input image and the registered image based on the size of the overlapping region between the input image and the registered image.
The step of determining the first maximum similarity having the largest similarity value among the similarity, and
including,
The method according to claim 12 . The step of determining the similarity is
A step of determining the degree of correlation in the overlapping region between the registered image of any one of the registered images and the input image, and
A step of determining the size of the overlapping region between the registered image of any one of the above and the input image, and
A step of determining the first similarity among the similarity between the registered image and the input image using the correlation degree and the size of the overlapping region.
including,
13. The method of claim 13 . The step of determining the similarity is
A step of determining a normalized cross-correlation (NCC) value using a movement vector and rotation angle based on image alignment between the registered image of any one of the registered images and the input image.
A step of determining a shared area ratio between any one of the registered images and the input image, and
A step of determining the first similarity among the similarity between the registered image and the input image by using the normalized cross-correlation value and the shared area ratio.
including,
13. The method of claim 13 . Further comprising the step of selecting a replacement image from the registered images based on a first similarity based on the size of the overlapping region between the registered images.
The method according to claim 12 . The step of selecting the replacement image is
It is a step of selecting a candidate image to be replaced from the registered images based on the first similarity based on the size of the overlapping region between the registered images, and each of the candidate images has the first similarity. In relation to the portion of, the portion comprises a first similarity between the relevant candidate image and the remaining registered images.
Steps to accumulate each of the above parts,
A step of selecting a candidate image having a high cumulative portion from the candidate images as the replacement image, and
including,
16. The method of claim 16 . The step of registering the input image in the registration database is
If the first maximum similarity is smaller than the higher threshold, the step of updating the registration database by replacing the replacement image with the input image, and
If the first maximum similarity is greater than the higher threshold, the step of retaining the replacement image in the registration database and
including,
16. The method of claim 16 . The step of registering the input image in the registration database is
The step of determining whether or not the first maximum similarity exceeds the lower threshold value, and
A step of registering the input image in the registration database based on a determination as to whether or not the first maximum similarity exceeds the lower threshold value.
including,
The method according to claim 12 . In the step of registering the input image in the registration database based on the determination as to whether or not the first maximum similarity exceeds the lower threshold, the first maximum similarity is larger than the lower threshold and the upper threshold is used. If smaller, it comprises the step of registering the input image in the registration database.
19. The method of claim 19 . A step of determining the authentication score between the input image and the registered image, and
A step of performing user authentication on the input image based on the authentication score and the authentication threshold, and a step of generating a similarity between the registered image and the input image.
Including
The step of determining the first maximum similarity is
Including the step of determining the first maximum similarity based on the generated similarity.
The method according to claim 12 . The biological information includes at least one of fingerprint information, vein information, and iris information.
The method according to claim 12 . In combination with hardware, a computer program stored in the medium in order to perform the method of any one of claims 1 to 22. It is a database management method
Steps to receive input images containing biometric information,
Steps to authenticate the input image based on the registered image in the database,
Including the step of updating the database.
The updating step includes determining whether or not to add the input image to the database.
The authentication step and the update step are performed at the same time as or less than the set time .
The update step further
A step of determining the similarity based on the size of the overlapping area between the registered image included in the database and the input image, and
A step of determining whether or not to register the input image in the database based on the similarity is included.
The step of determining the similarity is
A step of determining at least a part of the similarity based on the authentication of the input image,
A step of determining at least a portion of the similarity to a set value based on the set time and the length of time associated with the authentication.
Including at least one of
Database management method. The biological information is fingerprint information.
The database management method according to claim 24 . The step of deciding whether or not to register in the database is
A step of determining whether or not the maximum similarity among the similarity is equal to or lower than the upper threshold value, and
The step of determining whether or not the maximum similarity is equal to or higher than the lower threshold value, and
Including at least one of
The database management method according to claim 24 . The updating step includes selecting replacement images based on similarity based on the size of overlapping regions between the registered images.
The database management method according to claim 24 .

Images