KR102685541B1 - Handwritten access security systems and methods - Google Patents

Abstract

The purpose of the present invention to solve the above-mentioned conventional problems is to provide a handwriting-based access security system and method that can reduce manpower usage, can be provided at a low cost, and can significantly reduce the risk of personal information leakage. In order to achieve the above-mentioned purpose, the handwriting-based access security system according to the present invention includes a handwriting input unit that inputs handwriting process information and handwriting image information of an entrant; an access authorization determination unit that compares and analyzes the handwriting process information and handwriting image information collected from the handwriting input unit with the handwriting process information and handwriting image information of a pre-stored entrant based on artificial intelligence to determine whether to grant access; and an access control unit that generates access control information to control the access system according to the access authorization determined by the access authorization determination unit.

Description

{Handwritten access security systems and methods}

The present invention relates to an access security system and method, and more specifically, to an access security system and method through writing that can reduce the risk of personal information leakage at low cost.

The access security system has been one of the very important systems for organizations and companies that handle secrets for a long time. If secrets are leaked, it can cause property damage or, in the worst case, national damage.

Therefore, organizations and companies that manage secrets operate various access security systems to ensure that the facilities they operate and the documents they manage are disclosed only to trusted people.

A typical access security system is operated by having a separate security staff member who manages access and issuing passes after verifying identity, such as an ID card. However, the above method incurs relatively high fixed expenses because it requires hiring employees, and because it is managed by people, loopholes in access control may arise due to unintentional mistakes.

With the recent development of IT technology, access security systems based on biometric information such as fingerprints and iris are also being actively introduced.

This access security system is not suitable for visitors who allow temporary access because only fixed users can enter, and because it uses biometric information, there is a risk of personal and biometric information being leaked if the access security system is hacked.

Republic of Korea Patent Publication No. 10-1534930 (Registration date: July 1, 2015) Republic of Korea Patent Publication No. 10-2002-0087938 (publication date: November 23, 2002)

The purpose of the present invention to solve the conventional problems described above is to provide an access security system and method through writing that reduces the use of manpower, can be provided at a low cost, and can greatly reduce the risk of personal information leakage. am.

In order to achieve the above object, an access security system through handwriting according to the present invention includes a handwriting input unit for inputting handwriting process information and handwriting image information of an entrant; an access authorization determination unit that compares and analyzes the handwriting process information and handwriting image information collected from the handwriting input unit with pre-stored handwriting process information and handwriting image information of a person permitted access based on artificial intelligence to determine whether access is granted; and an access control unit that generates access control information to control the access system according to the access permission determined by the access permission determination unit.

In addition, in the access security system through writing according to the present invention, the handwriting input unit is characterized as an AIoT pen that can obtain writing process information and a handwriting image when the person entering and exiting writes on paper.

Additionally, in the access security system through handwriting according to the present invention, the handwriting input unit is characterized as a smart device that inputs the handwriting information by writing on a touch screen.

Additionally, in the access security system through handwriting according to the present invention, the access authorization determination unit includes a handwriting data collection unit that collects writing process information and handwriting image information received from the handwriting input unit; a handwriting process similarity module that calculates similarity with the handwriting process information using recurrent neural network (RNN)-based artificial intelligence technology based on the writing process information; a handwriting image similarity module that calculates similarity with the handwriting image information using convolutional neural network (CNN)-based artificial intelligence technology based on the handwriting image information; and an access authorization unit that converts the judgment results of the handwriting process similarity module and the handwriting image similarity module into a similarity distance scale, and compares the similarity distance scale with a preset similarity judgment standard distance to determine whether access is permitted. do.

In addition, in the access security system through handwriting according to the present invention, the access authorization determination unit further includes a preprocessor that processes the data collected by the handwriting data collection unit into digital data for similarity judgment processing based on artificial intelligence technology. It is characterized by

In addition, in the access security system through handwriting according to the present invention, the handwriting process similarity module calculates the similarity between the visitor's handwriting process information and the handwriting process information using DeepSignNet.

In addition, in the access security system through handwriting according to the present invention, the handwriting image similarity module calculates similarity between the visitor's handwriting image information and the handwriting image information using SigNet.

Additionally, in the access security system through writing according to the present invention, the writing process information includes at least one of handwriting position, handwriting angle, pen pressure, writing speed, writing acceleration, and handwriting stroke information.

And, in order to achieve the above object, the access security method through handwriting according to the present invention includes (a) the access authorization register registers and stores in the DB the handwriting process information and handwriting image information entered by the person with access through the handwriting input unit; steps; (b) a step where the handwriting input unit inputs the visitor's handwriting process information and handwriting image information; (c) The access authorization determination unit compares and analyzes the handwriting process information and handwriting image information of the visitor collected from the handwriting input unit with the registered handwriting process information and handwriting image information based on artificial intelligence to determine whether access is granted. step; and (d) an access control unit generating access control information and controlling the access system according to the access approval determined by the access approval determination unit.

Additionally, in the access security method through handwriting according to the present invention, step (c) includes: (c1) a handwriting data collection unit collecting writing process information and handwriting image information received from the handwriting input unit; (c2) a handwriting process similarity module calculating similarity with the handwriting process information using recurrent neural network (RNN)-based artificial intelligence technology based on the handwriting process information; (c3) a handwriting image similarity module calculating similarity with the handwriting image information using convolutional neural network (CNN)-based artificial intelligence technology based on the handwriting image information; and (c4) a step where the access authorization unit converts the judgment results of the handwriting process similarity module and the handwriting image similarity module into a similarity distance scale, and compares the similarity distance scale with a preset similarity judgment standard distance to determine whether or not access is authorized. It is characterized by including.

Specific details of other embodiments are included in “Specific Details for Carrying Out the Invention” and the attached “Drawings.”

The advantages and/or features of the present invention and methods for achieving them will become clear by referring to the various embodiments described in detail below along with the accompanying drawings.

However, the present invention is not limited to the configuration of each embodiment disclosed below, but may also be implemented in various different forms. However, each embodiment disclosed in this specification ensures that the disclosure of the present invention is complete, and the present invention It is provided to fully inform those skilled in the art of the present invention, and it should be noted that the present invention is only defined by the scope of each claim.

According to the present invention, by reducing the use of manpower and using relatively inexpensive equipment, not only can it be provided at a lower cost than existing systems, but it can also greatly reduce the risk of information leakage by not using personal information directly. Provides access security systems and methods.

In addition, according to the present invention, an access security system and method is provided that can fundamentally prevent identity forgery, as signature data containing complex and diverse data makes imitation nearly impossible.

In addition, according to the present invention, the handwriting of a person attempting illegal entry is left on the record, thereby providing an access security system and method that can help identify suspects without personal information when an illegal incident occurs.

Figure 1 is a diagram showing the block configuration of an access security system through writing according to an embodiment of the present invention.
Figure 2 is a diagram showing the detailed flow of an access security method through writing according to an embodiment of the present invention.
Figure 3 is a diagram showing a schematic diagram of the access authentication process of an artificial intelligence module applied to the access security method through handwriting according to an embodiment of the present invention.
Figure 4 is a diagram showing an example of a dot pattern applied to an access security system and method through writing according to an embodiment of the present invention.
Figure 5 is a schematic diagram of the access security process applied to the access security system and method through writing according to an embodiment of the present invention.

Before explaining the present invention in detail, the terms or words used in this specification should not be construed as unconditionally limited to their ordinary or dictionary meanings, and the inventor of the present invention should not use the terms or words in order to explain his invention in the best way. It should be noted that the concepts of various terms can be appropriately defined and used, and furthermore, that these terms and words should be interpreted with meanings and concepts consistent with the technical idea of the present invention.

That is, the terms used in this specification are only used to describe preferred embodiments of the present invention, and are not used with the intention of specifically limiting the content of the present invention, and these terms refer to various possibilities of the present invention. It is important to note that this is a term defined with consideration in mind.

In addition, it should be noted that in this specification, singular expressions may include plural expressions, unless the context clearly indicates a different meaning, and may include singular meanings even if similarly expressed in plural. .

Throughout this specification, when a component is described as “including” another component, it does not exclude any other component, but rather includes any other component, unless specifically stated to the contrary. It could mean that you can do it.

Furthermore, if a component is described as being "installed within or connected to" another component, it means that this component may be installed in direct connection or contact with the other component and may be installed in contact with the other component and may be installed in contact with the other component. It may be installed at a certain distance, and in the case where it is installed at a certain distance, there may be a third component or means for fixing or connecting the component to another component. It should be noted that the description of the components or means of 3 may be omitted.

On the other hand, when a component is described as being “directly connected” or “directly connected” to another component, it should be understood that no third component or means is present.

Likewise, other expressions that describe the relationship between each component, such as "between" and "immediately between", or "neighboring" and "directly neighboring", have the same meaning. It should be interpreted as

In addition, in this specification, terms such as "one side", "other side", "one side", "the other side", "first", "second", etc., if used, refer to one component. It is used to clearly distinguish it from other components, and it should be noted that the meaning of the component is not limited by this term.

In addition, in this specification, terms related to position such as "top", "bottom", "left", "right", etc., if used, should be understood as indicating the relative position of the corresponding component in the corresponding drawing. Unless the absolute location is specified, these location-related terms should not be understood as referring to the absolute location.

In addition, in this specification, when specifying the reference numeral for each component in each drawing, the same component has the same reference number even if the component is shown in different drawings, that is, the same reference is made throughout the specification. The symbols indicate the same component.

In the drawings attached to this specification, the size, position, connection relationship, etc. of each component constituting the present invention is exaggerated, reduced, or omitted in order to convey the idea of the present invention sufficiently clearly or for convenience of explanation. It may be described, and therefore its proportions or scale may not be exact.

In addition, hereinafter, in describing the present invention, detailed descriptions of configurations that are judged to unnecessarily obscure the gist of the present invention, for example, known technologies including prior art, may be omitted.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the drawings.

Figure 1 is a diagram showing the block configuration of an access security system through writing according to an embodiment of the present invention.

As shown in FIG. 1, the access security system through writing according to an embodiment of the present invention includes a handwriting input unit 100, an access authorization register 210, an access authorization determination unit, and an access control unit 250. It may be configured to include an authorization server 200.

Here, the handwriting input unit 100 may be a handwriting input device that inputs the visitor's writing process information and handwriting image information.

The access authorization determination unit compares and analyzes the handwriting process information and handwriting image information collected from the handwriting input unit 100 through a network such as wireless communication with the pre-stored handwriting process information and handwriting image information of the person permitted to enter, based on artificial intelligence. It may be a configuration that determines whether to authorize or not.

The access control unit 250 may be configured to generate access control information to control the access system of a specific building or space according to the access approval determined by the access approval determination unit.

As such, the access security system through writing according to an embodiment of the present invention uses time-series writing process data such as writing speed and acceleration using an AIoT pen, an electronic pen that writes on paper, or a smart device, by a person who wishes to enter through the access system. After completion of the handwriting, it provides a system that allows access to the handwriting image data written on actual paper or a smart device and compares and analyzes the registered data through an artificial intelligence model to determine whether or not access is permitted.

Hereinafter, each component of the access security system through writing according to an embodiment of the present invention will be described in detail with reference to the drawings.

As shown in FIG. 1, the handwriting input unit 100 is a component that inputs handwriting information for entry by a person who wishes to enter through the access system of a building or space, and inputs handwriting information such as writing speed, acceleration, and pen pressure in the writing process. It may be an IoT pen or AIoT pen equipped with a sensor that can measure or detect various generated data.

Here, the AIoT pen may be composed of a small camera to recognize dot patterns formed on writing paper, a pressure sensor, an acceleration sensor, a tilt sensor, a general ballpoint pen nib, a Bluetooth module, and a battery.

In other words, as an example of a handwriting entry section, when a person who wants to enter using an AIoT pen writes a specific handwriting, time-series location information of the handwriting can be obtained by the camera provided in the AIoT pen, and each stroke is monitored through a pressure sensor. The applied pressure (pen pressure) can be measured, and time-series writing process data that occurs during the writing process can be obtained by measuring writing speed, writing acceleration, and tilt of the pen through an acceleration sensor or tilt sensor.

Additionally, when the writing process is completed, the handwriting input unit 100 may acquire completed handwriting image or handwriting image data through a camera, sensor, etc.

Such an Internet of Things (IoT) pen or AIoT pen, also called a smart pen, is a device that collects handwriting data for authentication, and has various sensors that can track and record handwriting process data in real time. Can be built-in.

1) The camera of an AIoT pen can generally be used to capture the movement of the pen and the path of writing on paper, and can also track the position of the pen relative to the paper and record a digital image of the handwriting. Here, if paper with a dot pattern printed on it is used, the location information of the handwriting can be obtained more accurately and quickly.

2) The tilt sensor is a sensor that measures the tilt of the pen relative to the writing surface and can detect the angle at which the pen is held when writing, which is a unique feature of handwriting.

3) The acceleration sensor is also called an accelerometer, and this sensor measures the acceleration of the pen. It provides data on the speed and direction of pen movement and can help understand the flow and speed of a person's writing.

4) The pressure sensor can measure the pressure (pen pressure) applied to the pen while writing. The intensity of each stroke can be captured to provide another piece of personalized information.

Additionally, as shown in FIG. 1, the handwriting input unit 100 may be a smart device such as a smartphone or smart pad that inputs the handwriting information by writing on a touch screen.

Such a smart device is carried by the person wishing to enter and writes with a touch pen, etc. on the smart device through Bluetooth communication already provided in the access system, and the corresponding writing process data and handwriting image data are input. It may be transmitted to the access authorization determination unit.

Here, the smart device is equipped with a communication device such as Bluetooth, and a program for acquiring writing process data generated during the writing process can be installed, as well as a tilt sensor and a contact pressure of the pen tip in the touch pen provided in the smart device. A sensor capable of measuring can be provided.

In this way, in addition to AIoT pens, other smart devices and technologies that can collect writing process data and handwriting image data can be applied.

For example, devices such as digital tablets and other tablets that support a stylus (touch pen) can capture handwriting data, and pressure-sensitive touch devices that can detect and record the strength, speed, and pattern of stylus strokes. A screen can be provided, and it is also possible to provide the above-described sensor in the stylus.

Additionally, the smartphone is equipped with a touch-sensitive screen and an accelerometer that can record handwriting data when writing on the screen using a finger or stylus, and can be used as a handwriting input unit 100 applied to an embodiment of the present invention.

And, as shown in FIG. 1, the access authorization determination unit may be configured to include a data collection unit 231, a handwriting process similarity module 234, a handwriting image similarity module 235, and an access authorization unit 237. there is.

Here, the data collection unit 231 is a component that collects the writing process information and handwriting image information received from the above-described handwriting input unit 100, and may be configured to include a database (DB) and a data management unit that manages the data. .

The data collection unit 231 may be configured to process the data collected by the above-described handwriting data collection unit 231 into digital data for similarity determination processing based on artificial intelligence technology.

In other words, the pre-processing unit 232 organizes the raw data collected from the handwriting input unit 100 and converts it into a format that the AI model can understand.

The preprocessing unit 232 of the access authorization determination unit applied to the embodiment of the present invention preprocesses the writing process data and handwriting image data through the following process for application of the recurrent neural network module and the convolutional neural network module.

The preprocessing process for calculating the similarity of handwriting process data is as follows.

1) Sequence encoding: The sequence data of the writing process must be encoded into a format understandable by the RNN, which may involve representing each writing action as a vector of numbers, with each number representing a specific property such as speed, pressure, angle, or position. It may apply to

2) Normalization: Numerical data should be normalized to a standard scale, usually with mean 0 and standard deviation 1, which is important to ensure that all features have equal importance and to ensure that the model is not distorted by features with large absolute values.

3) Sequence Padding: RNNs generally require input sequences of the same length, and if the handwritten sequences have different lengths, they may need to be "padded" with an additional value such as 0 to make them all the same length.

The preprocessing process for calculating handwriting image data similarity is as follows.

1) Image resizing: CNN requires input images of the same size, so all handwriting images must be resized to a standard size.

2) Grayscale conversion: If color is not important in the handwriting recognition task, the image can be converted to grayscale. While color images have three color channels (red, green, and blue), grayscale images have only one color channel, which simplifies data and reduces computational costs.

3) Normalization: Like handwriting process data, image data must be normalized to a standard scale. This may involve scaling pixel values from their original range (typically 0 to 255) to a standard range (such as 0 to 1 or -1 to 1).

Once preprocessing by the preprocessor 232 is completed, the handwriting data is ready to be input into RNN and CNN models for learning or prediction. The exact preprocessing steps may vary depending on the characteristics of the handwriting data and AI model.

And, as shown in FIG. 1, the writing process similarity module 234 uses recurrent neural network (RNN)-based artificial intelligence technology based on the collected writing process data to register and store access permitted in the DB in advance. It may be a configuration that calculates the similarity with the person's handwriting process data.

Handwriting process data refers to the series of actions and movements that occur during the writing process, including the order and direction of strokes, speed, pressure, and other nuances that make each person's handwriting unique.

Therefore, RNNs are particularly suited to analyzing such sequential data because they can maintain a kind of 'memory' of past inputs in the sequence. This can be very important for understanding stroke order, speed, and other time-dependent elements of handwriting.

Here, recurrent neural networks (RNNs) are suitable for sequential data analysis, and several variants have been developed to handle various problems such as long-term dependence and vanishing gradients. The most common types may include long short term memory (LSTM), gated cyclic unit (GRU), and bidirectional RNN.

Long short-term memory (LSTM) is designed to avoid long-term dependency problems by using a system of gates, which determine what information should be kept or discarded at each step in the sequence, allowing LSTMs to retain important information over longer sequences.

Gate cyclic unit (GRU) is a simplified version of LSTM, which uses gates to control the flow of information, but has the advantage of being easier to calculate due to fewer gates and fewer parameters.

Bidirectional RNN has the advantage of processing handwriting process data in both forward and backward directions and being able to capture the dependencies of not only past data points but also future data points, making it useful when extracting such feature data and applying it to similarity calculation.

In particular, in an embodiment of the present invention, DeepSign technology can be used to calculate the similarity of handwriting process data.

DeepSign is a technology that uses deep learning technology to recognize handwriting and authenticate users. By learning the user's handwriting, it can extract the user's unique characteristics and authenticate the user based on this.

Additionally, DeepSign has the advantage of being a convenient and safe authentication method compared to existing biometric authentication technologies such as fingerprint recognition, iris recognition, and face recognition.

More specifically, DeepSign has the following advantages:

First, DeepSign has the advantage of being able to authenticate using only the user's handwriting, so there is no need to wear separate equipment or register biometric information.

Second, DeepSign has the advantage of being difficult for others to forge because it learns the user's handwriting and extracts the user's unique characteristics.

Third, DeepSign has the advantage of being able to recognize a variety of handwriting, so it can be used regardless of the user's age, gender, or nationality.

And, as shown in FIG. 1, the handwriting image similarity module 235 is configured to calculate the similarity with the handwriting image information using convolutional neural network (CNN)-based artificial intelligence technology based on handwriting image information. You can.

Here, convolutional neural networks (CNNs) have the advantage of being particularly suitable for image analysis because they can effectively learn features of images with unique spatial hierarchies such as lines, curves, and loops in the context of handwriting. CNNs can work by using filters and pooling layers to extract and learn the most important features from an image.

Additionally, in an embodiment of the present invention, SigNet technology can be used to calculate the similarity of handwriting image data.

Here, SigNet is a deep learning-based approach to offline signature verification often used in biometric authentication. Offline signature verification is based on static images, unlike online methods that consider the process of creating a signature, such as stroke order, speed, and pressure. This is a process that verifies handwritten signatures.

SigNet uses a convolutional neural network (CNN) architecture to learn discriminant features from signature images. It is especially effective for tasks that require distinguishing between authentic signatures and skilled forged signatures.

Then, the access authorization unit 237 converts the judgment results of the handwriting process similarity module 234 and the handwriting image similarity module 235 into a similarity distance scale, and compares the converted similarity distance scale with a preset similarity judgment standard distance. It may be a configuration that determines whether or not to grant access.

When the above-mentioned two modules analyze the input handwriting data, the access authorization unit 237 calculates the similarity distance between the analyzed data and the data of the pre-registered approved person. This may include mathematical distance metrics such as cosine similarity or Euclidean distance.

Then, the access authorization unit 237 compares the calculated distance with the preset similarity judgment standard distance (d), and if the calculated similarity distance is greater than the preset similarity judgment standard distance (the input handwriting is very similar to the certified handwriting), Indicates similarity) You can decide to allow access, otherwise, you can restrict or deny access.

Figure 2 is a diagram showing the detailed flow of an access security method through writing according to an embodiment of the present invention.

As shown in FIG. 2, the access security method through handwriting according to an embodiment of the present invention is (a) the access authorization register records handwriting process information and handwriting image information entered by a person with access through the handwriting input unit 100. A step of registering and storing in the DB (S100), (b) a step of the handwriting input unit 100 inputting the handwriting process information and handwriting image information of the visitor (S200), and (c) an access authorization determination unit of the handwriting input unit ( A step (S300) of comparing and analyzing the handwriting process information and handwriting image information of the visitor collected from 100) with the registered handwriting process information and handwriting image information based on artificial intelligence to determine whether to grant access (S300); ) It may be configured to include a step (S400) in which the access control unit 250 controls the access system according to the access approval determined by the access approval determination unit.

Here, step (c) (S300) includes (c1) a step (S310) in which the handwriting data collection unit 231 collects the writing process information and handwriting image information received from the handwriting input unit 100, and (c2) A step (S320) in which the handwriting process similarity module 234 calculates the similarity with the handwriting process information using recurrent neural network (RNN)-based artificial intelligence technology based on the handwriting process information, and (c3) handwriting image similarity. A step (S330) in which the module 235 calculates the similarity with the handwritten image information using convolutional neural network (CNN)-based artificial intelligence technology based on the handwritten image information, (c4) the access authorization unit 237 ) converting the judgment results of the handwriting process similarity module 234 and the handwriting image similarity module 235 into a similarity distance scale, and comparing the similarity distance scale with a preset similarity judgment standard distance to determine whether entry is permitted. (S340) may be included.

More specifically, as shown in Figure 2, the process of the access security method through handwriting according to an embodiment of the present invention includes a data collection step, a data preprocessing step, a feature extraction and similarity calculation step, and an access authorization judgment and decision step. It may be configured to include.

The data collection step is to collect writing process data (e.g. stroke order, speed, pressure) and handwriting image data by writing a signature or specific phrase using the IoT pen or smart device you wish to access.

The data preprocessing step is a process of preprocessing the collected data to prepare it for use in the AI module, and may include sequence encoding and normalization for handwriting process data, and image resizing and normalization for handwriting image data.

In the feature extraction and similarity calculation stage, the preprocessed data is fed to the AI module, and in the case of handwriting process data, an RNN (e.g. LSTM or GRU) generates a series of predictions based on the input sequence, and in the case of handwriting image data, , CNN (e.g., SigNet) generates feature vectors that represent the essential features of handwriting.

Then, the similarity calculation compares the RNN's prediction and the CNN's feature vector with the data of the pre-registered access authorizer, respectively, and the similarity score can be calculated based on a distance metric such as Euclidean distance or cosine similarity. The similarity score indicates how well the entered handwriting matches the approved handwriting.

The access authorization judgment and decision step converts the similarity scores of the two (writing process data and handwriting image data) into similarity distances, compares the converted similarity distance with a preset similarity judgment standard distance (threshold), and calculates the converted similarity score. If the distance exceeds the standard distance for determining similarity, the newly entered handwriting is considered to match the approved handwriting, and this is the step to determine whether to grant access.

Then, when access authorization is determined, the access control unit 250, which controls the access system, receives the authorization information and controls the access system to open the door. Otherwise, opening the door will be restricted or denied.

The access security system and method through handwriting according to an embodiment of the present invention can provide a high level of security because it is difficult to forge by using both dynamic handwriting process data and static handwriting image data.

In general, artificial intelligence technology that compares handwriting has been developed for the purpose of judging and preventing copying and imitation of handwriting. Since handwriting data is generally created based on images, an artificial intelligence model is learned by comparing two written images.

SigNet uses Siamese Network to score the difference between two images as a distance and sets a critical distance to distinguish between copy and handwriting. However, because this method is based on images, there are limitations in distinguishing between copies made by people such as copyists.

Therefore, data during the writing process can be very important to overcome limitations regarding images.

AIoT pens, such as digitizer pens, have the advantage of being able to collect data such as pen tilt, acceleration, and pen pressure because they display strokes on the display based on data generated from the pen.

Since such writing process data is structured through temporal order, it is desirable to use a recurrent neural network model specialized for data with temporal order, such as RNN, as described above.

In addition, DeepSign compares the similarity of two handwriting data using a layer with a Twin RNN structure based on 23 types of time series data acquired from an electronic pen.

The above method can be powerful against imitation because it looks at the process data, but it has the disadvantage that the similarity to the original person may be low because the data of the handwriting process is not a handwriting image for the model to understand. In the embodiment of the present invention, We propose a system and method for calculating similarity between handwriting process data and handwriting image data using each artificial intelligence module, and for access authentication based on this.

Figure 3 is a diagram showing a schematic diagram of the access authentication process of an artificial intelligence module applied to the access security method through handwriting according to an embodiment of the present invention.

As shown in Figure 1, the structure of the artificial intelligence model largely consists of a module that determines the similarity of handwriting process data and a module that determines the similarity of handwriting images.

The similarity judgment module of the handwriting process data preprocesses the raw data to more accurately judge the handwriting similarity, generates 24 characteristic information such as pen pressure, speed, and acceleration as shown in [Table 1] below, and compares the similarity using DeepSign. there is.

The handwriting image similarity judgment module can compare similarity using SigNet, which has a Siamese Network structure. The similarity from the two modules is converted to a similarity scale distance, and the sum of the two distances is used as the final result.

Figure 4 is a diagram showing an example of a dot pattern applied to an access security system and method through writing according to an embodiment of the present invention.

The AIoT pen used in the access security system according to an embodiment of the present invention can use dot pattern recognition technology that recognizes fine dot patterns on paper.

As shown in Figure 4, a dot pattern refers to a fine dot pattern printed on paper that is not easily visible to the eye. Dot pattern recognition technology has a different dot pattern at each location, so it recognizes the dot pattern at the current location through a small camera attached to the AIoT pen and determines the location of the pen on the paper. It's technology.

The dot pattern can be printed with a regular printer, so it has the advantage of being easy to use by printing the dot pattern on a general guest book. AIoT pens support pen pressure detection and can connect multiple pens to one server computer at the same time via Bluetooth.

Figure 5 is a schematic diagram of the access security process applied to the access security system and method through writing according to an embodiment of the present invention.

As shown in FIG. 5, when attempting to enter after registering one's signature in the handwriting database for verification provided in the data collection unit 231, the access authorization server 200 uses the data in the handwriting database for verification. Artificial intelligence performs an identity verification process to determine whether or not to enter.

Here, the database (DB) of the data collection unit 231 may include an entry/exit record database and an entry/exclusion database for the results of entry/exit.

Entry follows the following procedures.

1) First-time visitor registers signature or handwriting with AIoT pen or smart device

2) Create visit records

3) Acquire handwriting image data and writing process data

4) Data transmission to the artificial intelligence-based access authorization server (200) corresponding to the access authorization determination unit.

5) Handwriting verification using artificial intelligence

6) Notification of handwriting verification results

As such, the present invention provides an access security system through signature or handwriting authentication and an access security method using this system.

The entry security system and method through writing according to the present invention is not only more cost-effective than existing systems by reducing the use of manpower and using relatively inexpensive equipment, but can also significantly reduce the risk of information leakage by not directly using personal information. there is.

In addition, because signature data contains complex and diverse data, imitation is nearly impossible, preventing identity forgery at its source.

In addition, the access security system through handwriting authentication according to the present invention leaves the handwriting of the person who attempted illegal entry in the record, so it can help identify suspects without personal information when an illegal incident occurs.

Above, various preferred embodiments of the present invention have been described by giving some examples, but the description of the various embodiments described in the "Detailed Contents for Carrying out the Invention" section is merely illustrative and the present invention Those skilled in the art will understand from the above description that the present invention can be implemented with various modifications or equivalent implementations of the present invention.

In addition, since the present invention can be implemented in various other forms, the present invention is not limited by the above description, and the above description is intended to make the disclosure of the present invention complete and is commonly used in the technical field to which the present invention pertains. It is provided only to fully inform those with knowledge of the scope of the present invention, and it should be noted that the present invention is only defined by each claim in the claims.

100: handwriting input section
200: Access authorization server
210: Access authorization register
230: Access authorization judgment unit
231: Data collection department
232: preprocessing unit
234: Handwriting process similarity module
235: Handwriting image similarity module
237: Entry authorization book
250: Access control unit

Claims (10)

a handwriting input unit that inputs the visitor's writing process information and handwriting image information;
an access authorization determination unit that compares and analyzes the handwriting process information and handwriting image information collected from the handwriting input unit with pre-stored handwriting process information and handwriting image information of a person permitted access based on artificial intelligence to determine whether access is granted; and
It includes an access control unit that generates access control information to control the access system according to the access permission determined by the access permission determination unit,
The access authorization determination unit,
a handwriting data collection unit that collects writing process information and handwriting image information received from the handwriting input unit;
a handwriting process similarity module that calculates similarity with the handwriting process information using recurrent neural network (RNN)-based artificial intelligence technology based on the writing process information;
a handwriting image similarity module that calculates similarity with the handwriting image information using convolutional neural network (CNN)-based artificial intelligence technology based on the handwriting image information; and
Characterized by an access authorization unit that converts the judgment results of the handwriting process similarity module and the handwriting image similarity module into a similarity distance scale, and compares the similarity distance scale with a preset similarity judgment standard distance to determine whether access is permitted. ,
Entry security system through writing.
According to paragraph 1,
The handwriting input unit,
Characterized in that it is an AIoT pen that can obtain writing process information and handwriting images when the person entering writes on paper.
Entry security system through writing.
According to paragraph 1,
The handwriting input unit,
Characterized as a smart device that inputs the handwriting information by writing on a touch screen,
Entry security system through writing.
delete According to paragraph 1,
The access authorization determination unit,
Characterized in that it further comprises a pre-processing unit that processes the data collected by the handwriting data collection unit into digital data for similarity judgment processing based on artificial intelligence technology,
Entry security system through writing.
According to paragraph 1,
The writing process similarity module is,
Characterized in calculating similarity between the visitor's handwriting process information and the handwriting process information using DeepSignNet,
Entry security system through writing.
According to paragraph 1,
The handwriting image similarity module is,
Characterized in calculating the similarity between the visitor's handwriting image information and the handwriting image information using SigNet,
Entry security system through writing.
According to paragraph 1,
The above writing process information is:
Characterized in that it includes at least one of handwriting position, handwriting angle, pen pressure, writing speed, writing acceleration, and handwriting stroke information,
Entry security system through writing.
(a) registering and storing the handwriting process information and handwriting image information entered by the person permitted access through the handwriting input unit in the database;
(b) a step where the handwriting input unit inputs the visitor's handwriting process information and handwriting image information;
(c) The access authorization determination unit compares and analyzes the handwriting process information and handwriting image information of the visitor collected from the handwriting input unit with the registered handwriting process information and handwriting image information based on artificial intelligence to determine whether access is granted. step; and
(d) an access control unit generating access control information and controlling the access system according to the access approval determined by the access approval determination unit;
In step (c),
(c1) a handwriting data collection unit collecting writing process information and handwriting image information received from the handwriting input unit;
(c2) a handwriting process similarity module calculating similarity with the handwriting process information using recurrent neural network (RNN)-based artificial intelligence technology based on the handwriting process information;
(c3) a handwriting image similarity module calculating similarity with the handwriting image information using convolutional neural network (CNN)-based artificial intelligence technology based on the handwriting image information; and
(c4) The access authorization unit converts the judgment results of the handwriting process similarity module and the handwriting image similarity module into a similarity distance scale, and compares the similarity distance scale with a preset similarity judgment standard distance to determine whether or not access is authorized. Characterized in that,
Access security method through writing.


delete