KR102664795B1 - Device and method for generating cancelable iris templates using column-based slicing - Google Patents

Abstract

An apparatus and method for generating a disposable iris template using heat-based slicing are disclosed. The disposable iris template generating device includes a binary code generating unit that generates an iris binary code from an iris image, an input vector generating unit that generates an input vector for an IoM hashing (Index-of-Max Hashing) technique from the iris binary code, and a template generator that performs IoM hashing to generate a hash code corresponding to the input vector.

Description

Apparatus and method for generating discardable iris templates using column-based slicing {DEVICE AND METHOD FOR GENERATING CANCELABLE IRIS TEMPLATES USING COLUMN-BASED SLICING}

The present invention relates to a method of generating a disposable iris template that can improve the accuracy and security of authentication in an iris recognition environment. In particular, a discardable iris template that can improve security and efficiency by utilizing a heat-based slicing technique is provided. It relates to a device and method for generating

Recently, as an untact-oriented culture was formed due to COVID-19, and the amendment to the Electronic Signature Act was passed in the plenary session of the National Assembly in May 2020, interest in user authentication technology increased. In particular, biometric authentication using iris is attracting more attention because it has high accuracy and uses a non-contact method. However, the immutability of the iris is an advantage during authentication, but if it is leaked, other users can also access the system with the iris information. Therefore, in order to prevent such threats in advance, templates created using iris information must be designed and stored so that they can be discarded. Disposable biometric templates regenerate multiple indistinguishable new templates from the same iris.

Kim et al.’s Index-of-Max (IoM) hashing by Jin et al. (Z. Jin, J.Y. Hwang, Y.L. Lai, S. Kim and A.B.J. Teoh, “locality sensitive hashing-enabled cancelable biometrics: Index-of- max hashing,”IEEE Transactions on Information Forensics and Security, vol. 13, no. 2, pp. 393-407, 2018.) was proven to be strong against leakage attacks by applying it to iris templates rather than fingerprints. , Jae-yeol Jeong, Ki-seong Kim, and Ik-rae Jeong, “Disposable iris template using hashing,” Journal of Information Security Society, vol. 29, no. 3, 565-577, 2019. However, in Kim et al.'s iris template creation process, additional operations occur as shift operations are performed for pre-sorting.

In the present invention, paying attention to this point, the iris binary code is divided on a column basis rather than a row basis in order to improve calculation processing speed. If the input vector for IoM hashing is constructed by slicing the iris binary code vertically rather than 10 bits horizontally, the efficiency of existing operations can be improved.

The technical problem to be achieved by the present invention is to provide a method and device for generating a discardable iris template with improved computational processing speed.

A disposable iris template generation device according to an embodiment of the present invention includes a binary code generator that generates an iris binary code from an iris image, and an input vector for an IoM hashing (Index-of-Max Hashing) technique from the iris binary code. It includes an input vector generator that generates an input vector, and a template generator that performs IoM hashing to generate a hash code corresponding to the input vector.

In addition, a method of generating a disposable iris template according to an embodiment of the present invention is performed by a computing device including at least a processor, and includes generating an iris binary code from an iris image, IoM hashing (Index- of-Max Hashing), generating an input vector for the technique, and performing IoM hashing to generate a hash code corresponding to the input vector.

The disposable iris template generation device and method according to an embodiment of the present invention has the effect of improving the matching performance time of IoM hashing through a column-based slicing technique.

In order to more fully understand the drawings cited in the detailed description of the present invention, a detailed description of each drawing is provided.
1 shows an example of an iris image.
Figure 2 shows an example of iris binary code.
Figure 3 is a diagram for explaining the input value conversion step of IoM hashing according to an embodiment of the present invention.
Figure 4 is a graph showing experimental results of the present invention and conventional techniques.
Figure 5 is a functional block diagram of an iris template generating device according to an embodiment of the present invention.
Figure 6 is a flowchart for explaining a method for generating an iris template according to an embodiment of the present invention.

Specific structural or functional descriptions of the embodiments according to the concept of the present invention disclosed in this specification are merely illustrative for the purpose of explaining the embodiments according to the concept of the present invention. They may be implemented in various forms and are not limited to the embodiments described herein.

Since the embodiments according to the concept of the present invention can make various changes and have various forms, the embodiments will be illustrated in the drawings and described in detail in this specification. However, this is not intended to limit the embodiments according to the concept of the present invention to specific disclosed forms, and includes all changes, equivalents, or substitutes included in the spirit and technical scope of the present invention.

Terms such as first or second may be used to describe various components, but the components should not be limited by the terms. The above terms are used only for the purpose of distinguishing one component from another component, for example, without departing from the scope of rights according to the concept of the present invention, a first component may be named a second component and similarly a second component The component may also be named a first component.

When a component is said to be "connected" or "connected" to another component, it is understood that it may be directly connected to or connected to that other component, but that other components may also exist in between. It should be. On the other hand, when it is mentioned that a component is “directly connected” or “directly connected” to another component, it should be understood that there are no other components in between. Other expressions that describe the relationship between components, such as "between" and "immediately between" or "neighboring" and "directly adjacent to" should be interpreted similarly.

The terms used in this specification are only used to describe specific embodiments and are not intended to limit the invention. Singular expressions include plural expressions unless the context clearly dictates otherwise. In this specification, terms such as “comprise” or “have” are intended to designate the presence of features, numbers, steps, operations, components, parts, or combinations thereof described in this specification, but are not intended to indicate the presence of one or more other features. It should be understood that it does not exclude in advance the existence or possibility of addition of elements, numbers, steps, operations, components, parts, or combinations thereof.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by a person of ordinary skill in the technical field to which the present invention pertains. Terms as defined in commonly used dictionaries should be interpreted as having meanings consistent with the meanings they have in the context of the related technology, and unless clearly defined in this specification, should not be interpreted in an idealized or overly formal sense. No.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings attached to this specification. However, the scope of the patent application is not limited or limited by these examples. The same reference numerals in each drawing indicate the same members.

The method for generating a discardable iris template proposed in the present invention is largely structured as follows.

1) Generate iris binary code.

2) The binary value extracted from the iris binary code is converted to a real number and converted into an input vector value for IoM hashing (Index-of-Max Hashing).

3) IoM hashing is performed using GRP (Gaussian Random Projection) or URP (Uniform Random Permutation) method.

4) IoM hashing matching is performed based on LSH (Locality Sensitive Hashing).

The column-based slicing proposed in the present invention is used in the second step, the input vector conversion step of IoM hashing. Therefore, in this specification, we will describe in detail how to improve efficiency in the relevant step and then briefly describe the remaining steps.

Binary code generation steps for iris image

The iris image (see Figure 1) goes through a process of converting it to binary code for hashing and matching. The Daugman algorithm used at this time removes noise from the iris image, performs normalization, and detects iris features using a Garbor Filter. The iris binary code generated in this way has a size of a bit horizontally (a is a random natural number, and an exemplary value is 480), and b bits vertically (b is a random natural number, and an exemplary value is 20) (see FIG. 2). ).

Input conversion step of IoM hashing

Since the input value of IoM hashing (which may also be named an input vector or feature vector) is a feature vector of real numbers, the iris binary code derived in the previous step requires a real number conversion process. In the present invention, real number conversion is performed through column-based slicing (see Figure 3).

The real value conversion process through column-based slicing has a clear advantage in that it does not require an alignment process (alignment-free). In general, biometric information must be accurately sorted in advance to enable comparison. In the case of the iris, perfect alignment is found by shifting the code. However, when dividing and sorting the code based on vertical (column), the process becomes simpler because a separate shift process is not required. Therefore, efficiency in terms of processing speed increases.

The real conversion sequence of the iris code is as follows.

First, the iris binary code is generated as a table of b × a bits (e.g., 20 × 480 bits). Next, data in row b (e.g., 20) and column 1 is extracted by slicing by column, and then the matrix is transposed into the matrix in row 1 and column b (e.g., 20). If a (e.g., 480) times are performed in this manner, a table with a total of 1 × (a × b) bits (e.g., 1 × 9600 bits) is formed. In order to have smooth real number conversion and security advantages, vector values must be generated as negative and positive values. Therefore, in order to proceed with the operation, the table of 1 ×b)/c (e.g., 960) vectors are generated. Then, each c (e.g., 10) bit binary number is converted to a decimal number in the range of 0 to 1023, divided by 1,024, and 0.5 is subtracted to obtain a value between -0.5 and 0.4990. Depending on the embodiment, it is also possible to subtract 512 and then divide by 512.

Steps to perform IoM hashing

IoM hashing has the following advantages by converting a real-valued feature vector as input into a rank-based code with an index with the maximum value.

1) Since the original feature vector cannot be restored using the hashed code, irreversibility is satisfied.

2) Because it is a method dependent on relative order, it is not affected by the size of the feature vector. Therefore, it is strong against noise and deformation.

3) The size independence of IoM hashing makes the hash code scale-invariant.

IoM hashing can be realized in two ways. Regarding IoM hashing and matching, which will be described below, the paper by Z. Jin et al. (Z. Jin, J.Y. Hwang, Y.L. Lai, S. Kim and A.B.J. Teoh, “locality See “sensitive hashing-enabled cancelable biometrics: Index-of-max hashing,” IEEE Transactions on Information Forensics and Security, vol. 13, no. 2, pp. 393-407, 2018.

First, GRP-based IoM hashing receives a d (d is a random natural number)-dimensional feature vector Outputs the index of the maximum value. Algorithm 1 for performing GRP-based IoM hashing is as follows.

[Algorithm 1]

Second, URP-based IoM hashing receives the feature vector Outputs the index of the maximum value. Algorithm 2 for performing URP-based IoM hashing is as follows.

[Algorithm 2]

Matching step between IoM hashing results

IoM hashing matching requires high similarity. and It follows the rank-based LSH (Locality Sensitive Hashing) algorithm, which has a high probability of collision. If the similarity between two iris vectors is high, the hash code collision probability is high, and if the similarity is low, the hash code (which may mean the result of IoM hashing) has a low probability of collision.

Registered hash code and query hash code The probability of collision between similarities is It is expressed as in other words, This happens.

1) GRP-based IoM matching

cast Let be the LSH function defined as . ego am. here am.

Similarity in GRP-based IoM hashing am.

The matching score in the implementation is based on the total entries (m) of the hash code. and This is the number of times it becomes '0' (collision) by subtraction of each element.

2) URP-based IoM matching

Similarity in URP-based IoM hashing is a relative order measurement within a certain range, and is expressed in Equation 1.

[Equation 1]

The matching score in the implementation is based on the total entries (m) of the hash code. and This is the number of times it becomes '0' (collision) by subtraction of each element.

Ultimately, in GRP-based IoM matching and URP-based IoM matching, if the matching score exceeds a predetermined threshold (or is equal to or exceeds the threshold), it can be determined that the two hash codes match. Otherwise, the two hashes can be determined to match. It may be determined that the code does not match. Matching two hash codes may mean that the two irises being compared are from the same person.

Below, the experiment and evaluation results are described.

The experimental environment for evaluating the efficiency of the proposed methodology is shown in Table 1 below.

[Table 1]

In the experiment, to evaluate the performance of this methodology, it was compared and analyzed based on GRP-based IoM hashing, which shows a relatively large difference in IoM hashing matching time. According to Kim et al., among the existing horizontal-based slicing techniques, GRP-based IoM hashing was measured at 0.065 seconds, and URP-based IoM hashing was measured at 0.058 seconds.

For the experimental data, CASIA v3 was used in the same way for equal comparison with Kim et al. Images of 249 subjects were extracted, but the left and right irises of one subject were treated as different subjects, and subjects with only 7 or more images were selected.

The measure of efficiency was set as the time required to match one iris query template with the registered hash binary code according to the values of the parameters, Q, the Guassian random projection vector, and Gaussian random matrix, m.

Experimental results for the existing horizontal-based slicing technique showed that as the values of q and m increased, the time required also increased proportionally. In particular, as m grew above 100, the amount of increase also increased. However, this methodology was measured to take less than 0.02 seconds even if the value of m increased.

Figure 4 shows the trend of time required when q is 5, which is the point where the difference is clearly revealed. Except for the moment when m is small, which is 10, all other values showed fast speeds.

Table 2 shows the time required when q is 5 and m is 300, which is the point where the result is clearly evident. This methodology showed a throughput rate about 21 times higher than the column-based technique that generates 17 templates by performing shift operations. Meanwhile, EER, a measure of accuracy, was confirmed to be approximately 10 times higher than that of Kim et al.

[Table 2]

Figure 5 is a functional block diagram of an iris template generating device according to an embodiment of the present invention.

Referring to FIG. 5 , the iris template generating device 100 may be implemented as a computing device including at least a processor and/or memory. Computing devices may include desktop PCs, tablet PCs, servers, smartphones, navigation devices, etc. The iris template generating device 100 includes an image acquisition unit 110, a binary code generating unit 120, an input vector generating unit 130, a template generating unit 140, a matching unit 150, and a storage unit 160. It may include at least one of the following. In describing the iris template generating device 100, detailed descriptions of content that overlaps with the previous description will be omitted.

The image acquisition unit 110 may acquire a target iris image from which to create a discardable iris template. Depending on the embodiment, the iris image may be stored in advance in the storage unit 160, and in this case, the iris template generating device 100 may not include the image acquisition unit 110.

Specifically, the image acquisition unit 110 may receive the user's iris image through a wired or wireless communication network from a scanner capable of scanning the iris. Depending on the embodiment, it is also possible to receive an iris image from a portable storage device such as a USB storage device through a predetermined input interface. As another example, the user's iris image may be obtained by capturing the user's iris image using an imaging device such as a camera provided in the iris template generating device 100.

The binary code generator 120 may generate a binary code (or iris binary code) corresponding to the iris image acquired by the image acquisition unit 110 or the iris image stored in the storage unit 160. To this end, the binary code generator 120 may perform at least one of iris area detection, noise removal, and normalization operations from the iris image.

The input vector generator 130 may generate an input vector for IoM hashing from the iris binary code. As described above, the input vector generator 130 may generate an input vector for IoM hashing by applying column-wise slicing.

The template generator 140 may generate a discardable iris template from the input vector generated by the input vector generator 130. Here, the iris template is the result of IoM hashing using an input vector as an input, and may be named a hashing code.

The iris template generating device 100 may further include a matching unit 150. In this case, the iris template generating device 100 may also be called an iris recognition device, an iris matching device, etc.

The matching unit 150 may compare the generated iris template with at least one of a plurality of pre-stored iris templates and calculate similarity to determine whether the two iris templates match.

The storage unit 160 may store programs, software, etc. necessary for operating the iris template generating device 100. Additionally, the storage unit 160 temporarily or non-temporarily stores data generated during the process of generating a discardable iris template, such as iris image, iris binary code, input vector, iris template, similarity, matching result, etc. It can be.

Figure 6 is a flowchart for explaining a method for generating an iris template according to an embodiment of the present invention.

The iris template generating method may be performed by the iris template generating device 100 shown in FIG. 5, that is, a computing device, and may be referred to as an iris recognition method, an iris matching method, etc.

First, an iris image is acquired by the image acquisition unit 110 of the iris template generating device 100 (S110). The acquired iris image may be stored in the storage unit 160.

Next, the binary code generator 120 of the iris template generating device 100 may generate an iris binary code corresponding to the acquired iris image (S120). Since the iris binary code can be generated using known techniques, detailed description thereof will be omitted.

The input vector generator 130 of the iris template generating device 100 may generate an input vector for input of IoM hashing from the iris binary code (S130). In order to generate a hash code, or iris template, through the IoM hashing technique, input of a predetermined standard is required. Accordingly, the input vector generator 130 can convert the iris binary code into an input vector to meet a predetermined standard.

The template generator 140 of the iris template generator 100 may generate a hash code corresponding to the input vector using an IoM hashing technique (S140). The hash code may refer to an iris template corresponding to the input iris image. Accordingly, an iris template that corresponds to the user's iris image and can be discarded can be created.

The matching unit 150 of the iris template generating device 100 is configured to provide at least one of a query hash code (a query iris template, which may mean a hash code generated by the template generating unit 140) and pre-stored hash codes. It is possible to decide whether to match (S150). Through this, authentication using iris can be performed. In other words, if there is a hash code matching the query hash code, authentication for the user may be completed, and if there is no hash code matching the query hash code, authentication for the user may fail.

The device described above may be implemented as a hardware component, a software component, and/or a set of hardware components and software components. For example, devices and components described in the embodiments may include, for example, a processor, a controller, an Arithmetic Logic Unit (ALU), a Digital Signal Processor, a microcomputer, a Field Programmable Array (FPA), It may be implemented using one or more general-purpose or special-purpose computers, such as a Programmable Logic Unit (PLU), microprocessor, or any other device capable of executing and responding to instructions. The processing device may execute an operating system (OS) and one or more software applications running on the operating system. Additionally, a processing device may access, store, manipulate, process, and generate data in response to the execution of software. For ease of understanding, a single processing device may be described as being used; however, those skilled in the art will understand that a processing device may include multiple processing elements and/or multiple types of processing elements. It can be seen that it may include. For example, a processing device may include a plurality of processors or one processor and one controller. Additionally, other processing configurations, such as parallel processors, are also possible.

Software may include a computer program, code, instructions, or a combination of one or more of these, and may configure a processing unit to operate as desired, or may be processed independently or collectively. You can command the device. Software and/or data may be used on any type of machine, component, physical device, virtual equipment, computer storage medium or device to be interpreted by or to provide instructions or data to a processing device. , or may be permanently or temporarily embodied in a transmitted signal wave. Software may be distributed over networked computer systems and stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., singly or in combination. Program instructions recorded on the medium may be specially designed and configured for the embodiment or may be known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. -Includes specially configured hardware devices to store and execute program instructions, such as magneto-optical media, ROM, RAM, flash memory, etc. Examples of program instructions include machine language code, such as that produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

The present invention has been described with reference to the embodiments shown in the drawings, but these are merely exemplary, and those skilled in the art will understand that various modifications and other equivalent embodiments are possible therefrom. For example, the described techniques are performed in a different order than the described method, and/or components of the described system, structure, device, circuit, etc. are combined or combined in a different form than the described method, or other components are used. Alternatively, appropriate results may be achieved even if substituted or substituted by an equivalent. Therefore, the true scope of technical protection of the present invention should be determined by the technical spirit of the attached registration claims.

100: Iris template creation device
110: Image acquisition unit
120: Binary code generation unit
130: input vector generation unit
140: Template creation unit
150: matching unit
160: storage unit

Claims (8)

A binary code generator that generates an iris binary code from an iris image;
an input vector generator that generates an input vector for an IoM hashing (Index-of-Max Hashing) technique from the iris binary code; and
A template generator that performs IoM hashing to generate a hash code corresponding to the input vector,
The input vector generator generates the input vector by applying column-based slicing to the iris binary code,
The input vector generator,
Separate each column of the iris binary code in row b (b is a random natural number) and column a (a is a random natural number) to extract a number of b rows and column 1 data,
Convert each of the extracted data in row b, column 1 to data in row 1, column b,
Combine the data in row 1 and column b to create a table with 1 row and column (a×b),
Split the table in row 1 (a
Generating the input vector by performing a real conversion operation on each of the values converted to decimal numbers,
The template generator performs IoM hashing based on Gaussian Random Projection (GRP), and the number (q) of Gaussian random projection vectors is 5,
Disposable iris template generation device. According to paragraph 1,
The disposable iris template generating device further includes a matching unit that determines whether the hash code matches at least one of a plurality of pre-stored hash codes,
Disposable iris template generation device. delete According to paragraph 1,
In the GRP-based IoM hashing, the number (m) of Gaussian random matrices is greater than 10 and less than or equal to 300,
Disposable iris template generation device. A method for generating a disposable iris template, the method performed by a computing device comprising at least a processor, the method comprising:
Generating an iris binary code from an iris image;
Generating an input vector for an Index-of-Max Hashing (IoM) hashing technique from the iris binary code; and
A step of performing IoM hashing to generate a hash code corresponding to the input vector,
In the step of generating the input vector, the input vector is generated by applying column-based slicing to the iris binary code,
The step of generating the input vector is,
Separate each column of the iris binary code in row b (b is a random natural number) and column a (a is a random natural number) to extract a number of b rows and column 1 data,
Convert each of the extracted data in row b, column 1 to data in row 1, column b,
Combine the data in row 1 and column b to create a table with 1 row and column (a×b),
Split the table in row 1 (a
Generating the input vector by performing a real conversion operation on each of the values converted to decimal numbers,
The step of generating the hash code involves performing IoM hashing based on GRP (Gaussian Random Projection), where the number (q) of Gaussian random projection vectors is 5,
How to create a disposable iris template. According to clause 5,
The disposable iris template generation method further includes determining whether the hash code matches at least one of a plurality of pre-stored hash codes,
How to create a disposable iris template. delete According to clause 5,
The step of generating the input vector is,
Using Gaussian random matrices greater than 10 and less than or equal to 300,
How to create a disposable iris template.

Images