KR20210018749A - Method for recognizing object based on image and apparatus for performing the method - Google Patents

Abstract

The present invention relates to an object recognition method based on an image and to a device performing the method. The object recognition method based on an image comprises the steps of: receiving, by an authentication server, an object image (reference authentication information) from a user device; and extracting, by the authentication server, the reference authentication information based on the object image (reference authentication information). By increasing the accuracy of object recognition based on an image, accurate object identification may be possible when providing a blockchain-based service.

Description

An image-based object recognition method and an apparatus for performing the method.

The present invention relates to an image-based object recognition method and an apparatus for performing the method. In more detail, the present invention relates to a method and apparatus for providing an object target service by capturing an image of an object and recognizing an object through analysis of the captured image.

Blockchain may be a digital ledger that is shared between members of the network by encrypting transaction information occurring in a public network or a private network.

A copy of the transaction ledger is distributed to each network member so that each time a new transaction occurs, the authentication process for the transaction can proceed with the consent of the members.

Since transaction information based on the blockchain cannot be arbitrarily changed, the reliability of transactions is increased and information tracking is easy. In particular, in the field of financial services, efforts are being made to increase work efficiency and establish a new service base by applying blockchain technology.

It is possible to reduce security, supervision, and regulatory costs through transparent transactions through a distributed ledger system. Blockchain technology can be applied not only to the financial sector, but also to logistics, distribution, and even government public administration services.

In order to provide blockchain-based services to individual objects, it may be necessary to prioritize accurate identification of objects. Object recognition may be performed through a separately allocated key such as an existing private key, but image-based object recognition may also be performed. Therefore, there is a need for a more accurate and less error-prone object recognition method.

An object of the present invention is to solve all of the above-described problems.

In addition, an object of the present invention is to increase the accuracy of object recognition based on an image to enable accurate object identification when providing a blockchain-based service.

Another object of the present invention is to extract features of different animals from images through specialized animal image recognition to enable accurate identification of animals.

A typical configuration of the present invention for achieving the above object is as follows.

According to an aspect of the present invention, in an image-based object recognition method, an authentication server receives an object image (reference authentication information) from a user device, and the authentication server performs standard authentication based on the object image (reference authentication information). It may include the step of extracting the information.

Meanwhile, the authentication process may be performed through authentication request information determined based on the object image (authentication request) transmitted from the user device on the blockchain and standard authentication information based on the object image (standard authentication information).

Further, the reference request information may be determined based on a first adjustment to the object image (reference authentication information), and the authentication request information may be determined based on a second adjustment to the object image (authentication request).

Further, the first adjustment is performed based on at least one adjustment center object included in the object image (reference authentication information), and the second adjustment is at least one adjustment center included in the object image (authentication request). It can be performed on an object basis.

According to another aspect of the present invention, an authentication server performing image-based object recognition includes a communication unit implemented to communicate with an external node and a processor operatively connected to the communication unit, wherein the processor is from a user device. It may be implemented to receive an object image (reference authentication information) and extract reference authentication information based on the object image (reference authentication information).

Meanwhile, the authentication process may be performed through authentication request information determined based on the object image (authentication request) transmitted from the user device on the blockchain and standard authentication information based on the object image (standard authentication information).

Further, the reference request information may be determined based on a first adjustment to the object image (reference authentication information), and the authentication request information may be determined based on a second adjustment to the object image (authentication request).

Further, the first adjustment is performed based on at least one adjustment center object included in the object image (reference authentication information), and the second adjustment is at least one adjustment center included in the object image (authentication request). It can be performed on an object basis.

According to the present invention, it is possible to accurately identify an object when providing a blockchain-based service by increasing the accuracy of object recognition based on an image.

In addition, according to the present invention, it is possible to accurately identify an animal by extracting characteristics of different animals from an image through specialized animal image recognition.

1 is a conceptual diagram showing an object authentication system according to an embodiment of the present invention.
2 is a conceptual diagram illustrating an image adjustment operation of an object image adjustment unit according to an embodiment of the present invention.
3 is a conceptual diagram showing a method for extracting authentication information from an object image according to an embodiment of the present invention.
4 is a conceptual diagram showing a method of generating authentication information according to an embodiment of the present invention.
5 is a conceptual diagram showing a method for extracting authentication information from an object image according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE INVENTION The detailed description of the present invention to be described later refers to the accompanying drawings, which illustrate specific embodiments in which the present invention may be practiced. These embodiments are described in detail sufficient to enable those skilled in the art to practice the present invention. It is to be understood that the various embodiments of the present invention are different from each other, but need not be mutually exclusive. For example, specific shapes, structures, and characteristics described herein may be changed from one embodiment to another and implemented without departing from the spirit and scope of the present invention. In addition, it should be understood that the positions or arrangements of individual elements in each embodiment may be changed without departing from the spirit and scope of the present invention. Therefore, the detailed description to be described later is not intended to be limited, and the scope of the present invention should be taken as encompassing the scope claimed by the claims of the claims and all scopes equivalent thereto. Like reference numerals in the drawings indicate the same or similar elements over several aspects.

Hereinafter, various preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings in order to enable those skilled in the art to easily implement the present invention.

Hereinafter, in the embodiment of the present invention, an animal is disclosed as an example of an object. However, the object recognition method according to an embodiment of the present invention may be applied to various objects other than animals, and such an embodiment may also be included in the scope of the present invention.

1 is a conceptual diagram showing an object authentication system according to an embodiment of the present invention.

In FIG. 1, an object authentication system for authenticating an object based on an image of an object is disclosed.

Referring to FIG. 1, the object authentication system may include a user device 100, an authentication server (or authentication node) 120, and a block chain 140.

The user device 100 may be implemented to generate an object image for generating reference authentication information and transmit the generated object image to the authentication server 120. Hereinafter, an object image for generating the reference authentication information may be expressed in terms of an object image (reference authentication information).

For example, the user device 100 may be a user's smartphone and performs imaging of an object (eg, dog, cat), and the captured object image (reference authentication information) is transmitted to the authentication server 120. I can.

In order to use the service, the user device 100 may preferentially generate an object image (reference authentication information) and transmit it to the authentication server 120. After the authentication process is completed based on the object image (standard authentication information), the user device 100 may access the service by generating an object image (authentication request) for accessing the service.

The authentication server 120 may be implemented for an object authentication procedure. The authentication server 120 may generate reference authentication information for an object authentication procedure through pre-processing, feature extraction, and analysis of an object image (reference authentication information). The generated standard authentication information is stored on the blockchain and can be used for authentication of the object later.

For example, the authentication server 120 may receive object image information (reference authentication information) through an application program interface (API), and the object image (reference authentication information) may be transmitted to the reference authentication information generation unit 150. have.

The reference authentication information generation unit 150 may be implemented to generate reference authentication information for an object. The reference authentication information generation unit 150 may include an object image adjustment unit 160 and a reference authentication information extraction unit 180. The object image adjustment unit 160 may adjust (or edit) an object image (reference authentication information) in order to accurately extract an object feature. For example, the object image adjustment unit 160 may adjust the direction of the object image (reference authentication information) to extract object feature information. The reference authentication information extractor 180 may be implemented to extract reference authentication information from an object image (reference authentication information) (or an adjusted object image (reference authentication information)). The standard authentication information is feature vector information including information on sub-objects such as eyes and nose and mouth located on the face of the object, pattern information, curvature information, and object information (sub-object information, pattern information, curvature information, etc.) And the like.

The reference authentication information generated by the reference authentication information generator 150 may be matched with object information input by the user device 100. The object information is information related to an object corresponding to the reference authentication information, and may include the name of the object, the age of the object, and information on the manager of the animal when the object is an animal. At least one of object information and standard authentication information may be stored on the block chain 140. For security, at least one of object information and standard authentication information may be hashed and stored on the authentication server 120 and/or the blockchain 140, and if there is a request for authentication later, authentication Information recorded on the server 120 and/or the blockchain 140 may be utilized.

For example, when the procedure for generating standard authentication information for an object is completed, the user can access the service through the authentication procedure for the object later. The user may request authentication by transmitting an image of an object on the blockchain 140. The object image for requesting authentication may be expressed in terms of an object image (authentication request).

Authentication request information is extracted from the object image (authentication request), and comparison between the authentication request information and object authentication information recorded on the blockchain 140 may be performed. The authentication request information can be extracted on the user device 100, and the authentication request information is directly transmitted to the block chain 140 through the user device 100, and object authentication information and authentication request recorded on the block chain 140 It may be determined whether information is matched. When the object authentication information and the authentication request information are matched, access to the service may be possible.

The object image (authentication request) that has successfully authenticated is converted to an object image (standard authentication information), and the extracted authentication request information may be used as standard authentication information. Through this method, the recognition rate for an object can be increased.

2 is a conceptual diagram illustrating an image adjustment operation of an object image adjustment unit according to an embodiment of the present invention.

In FIG. 2, a method for adjusting an object image to extract reference recognition information by an object image adjusting unit is disclosed. The same method can be used to extract authentication request information from an object image (authentication request).

Referring to FIG. 2, the object image adjustment unit may extract the adjustment center object 200 from the object image and rotate the object image based on the position of the adjustment center object 200 to adjust the object image. The adjustment center object 200 of the object image may be a sub-object that necessarily exists in the object. For example, when there are individual objects, the eyes, nose, and ears of a dog may be the coordination center object 200.

The object image adjusting unit may adjust the object image by determining the position of the adjustment center object 200 and aligning the position of the adjustment center object 200 by rotating a preset vertical line and a horizontal line.

For example, a first imaging line connecting the positions of the two eyes 210, which is the adjustment center object 200, is parallel to the first horizontal line 215, and two noses located at the nose, which is the adjustment center object 200. The direction of the object image may be adjusted so that the second imaging line connecting the position of the hole 220 is parallel to the second horizontal line 225.

In addition, according to an embodiment of the present invention, the direction of the object image may be adjusted based on the direction of the edge located on the adjustment center object 200. For example, the edges of the ears and the mouth are extracted, and the direction of the object image may be adjusted in consideration of the directions of the ears and the edges of the mouth.

In addition, with respect to both ears, a first angle between the first edge 240 and the third horizontal line 230 extracted from the first ear, and the second edge 245 and the third horizontal line 230 extracted from the second ear. If the second angle with) is different from each other, the direction of the object image may be adjusted to match the first angle and the second angle equally. Alternatively, the direction of the object image may be adjusted so that the third edge 250 and the fourth edge 255 extracted from the mouth have the same angle with respect to a horizontal line or a vertical line.

In addition, when the first edge group-based shape 260 extracted from the first ear and the second edge group-based shape 265 extracted from the second ear are different for both ears, the first edge group-based shape 260 The direction of the object image may be adjusted to make the and the second edge group-based shape 265 the same.

The object image adjustment method as described above may be performed by setting a priority according to recognition accuracy, or may be performed based on a degree of consistency of the object image adjustment set based on a higher priority. For example, the most accurate method of recognizing an adjustment center object, edge, or edge group-based shape on the object image may be set as a relatively higher priority object image adjustment method.

An object image adjustment method capable of relatively high accuracy and relatively fast recognition based on an object image adjustment result based on a plurality of object images may be set to have a higher priority image adjustment result.

For example, the priority may be set as follows.

First priority object image adjustment method: Of the object image adjustment methods using the adjustment center object, the object image adjustment method using the position of the first adjustment center object (eg, eye) and the first horizontal line

Second priority object image adjustment method: Among the object image adjustment methods using the adjustment center object, the position of the second adjustment center object (for example, nostril) that has a lower priority than the first adjustment center object and a second horizontal line are determined. Object image adjustment method

Third priority object image adjustment method: An object image adjustment method based on the angle between an edge and a horizontal line extracted from a third adjustment center object (eg, ear) among the object image adjustment methods using the adjustment center object

Fourth priority object image adjustment method: An object image adjustment method based on the shape of the edge group extracted from the third adjustment center object (eg, ear) among the object image adjustment methods using the adjustment center object

In consideration of the priority of such an object image adjustment method, the object image may be preferentially adjusted based on the higher priority. For example, when the first priority object image adjustment method is available, the object image may be adjusted based on the first priority object image adjustment method. When the first priority object image adjustment method is not available, the object image may be adjusted based on the second priority object image adjustment method.

Alternatively, the first adjustment object image adjusted based on the higher priority and the second adjusted object image adjusted based on the next higher priority by performing object image adjustment of the higher priority and the next higher priority in consideration of the priority of the object image adjustment method Verification of whether or not the object image has been accurately adjusted may be performed through a comparison between the two.

Further, a plurality of object image adjustment methods may be performed, and a final adjustment object image may be determined and used based on each of the first adjustment object image to the n-th adjustment object image adjusted based on the plurality of object image adjustment methods.

3 is a conceptual diagram showing a method for extracting authentication information from an object image according to an embodiment of the present invention.

In FIG. 3, a method of extracting standard authentication information from an object image (standard authentication information) and authentication request information from an object image (authentication request) is disclosed.

Referring to FIG. 3, the object may be an animal such as a dog or a cat, and a method for extracting authentication information (standard authentication information, authentication request information) based on an image of an animal is disclosed.

According to an embodiment of the present invention, an object image (object image (reference authentication information), object image (authentication request)) can be converted into a black and white image (step S300).

An image reversal is performed on the object image converted to a black-and-white image (step S310), and then authentication information may be extracted from the image using a posterization filter (step S320) (step S330). ).

The authentication information extraction procedure may be performed after performing the image adjustment operation described above in FIG. 2, or may be performed separately.

Through the above procedure, the animal pattern can be more accurately extracted from the object image, and the information on this pattern can be used as one authentication information.

For example, information on an area having a value less than or equal to a threshold brightness may be generated as authentication information. An area having a value less than or equal to the threshold brightness based on the surrounding brightness value among the black and white information located within the face of the animal may be extracted as the authentication information extraction area. The location of the authentication information extraction region may be determined based on the position of the above-described adjustment center object, and information on the shape of the authentication information extraction region may be determined based on a specific point located in the authentication information extraction region.

4 is a conceptual diagram showing a method of generating authentication information according to an embodiment of the present invention.

In FIG. 4, a method for generating information on an area having a value less than or equal to a threshold brightness as authentication information is disclosed.

Referring to FIG. 4, a procedure for generating authentication information may be performed on an image converted based on the procedure disclosed in FIG. 3.

First, the authentication information extraction region 420 may be extracted from black and white information located within the face of the animal. For convenience of explanation, the authentication information extraction area 420 is determined as an area having a value less than or equal to the threshold brightness compared to the brightness value of the surrounding area, but the range of the critical brightness for determining the authentication information extraction area 420 is variously set. Can be, and this method may also be included in the scope of the present invention.

The location of the authentication information extraction area 420 may be determined based on the location of the adjustment center object 400. On the authentication information extraction area 420, a specific position determination point (eg, a center of gravity point, a point nearest or farthest from the control center object 400) 450 may be determined, and the location of the control center object 400 And the relative position on the authentication information extraction area 420.

The shape of the authentication information extraction region 420 may be determined based on a distance of the authentication information extraction region 420 based on the location determination point 450. For example, a plurality of distances 460 on the boundary point of the authentication information extraction area 420 are extracted based on the location determination point 450, and the authentication information extraction area 420 is Information about the shape can be extracted. Alternatively, the shape of the authentication information extraction area may be stored as a separate image.

According to an embodiment of the present invention, the object authentication information extraction area 420 may be selectively used. As described above, the object image adjustment unit may adjust the object image by extracting the adjustment center object 400 from the object image and rotating the object image based on the position of the adjustment center object 400. When such an adjustment angle increases, distortion of the authentication information extraction region of the object may increase.

Accordingly, among the plurality of authentication information extraction regions 420 located on the object according to the adjustment angle, the authentication information extraction region 420 having relatively small distortion may be selected and used for the authentication procedure. As the adjustment angle is relatively large, the authentication information extraction region 420 located in a region relatively close to the center of the object image is selected to generate authentication information, and as the adjustment angle is relatively small, the region is relatively far from the center of the object image. The located authentication information extraction area 420 may also be selected to generate authentication information.

Alternatively, according to an embodiment of the present invention, an adjustment center object having the least change in an adjustment angle or image when adjusting a plurality of adjustment center objects located on the object may be selected as the adjustment center object (minimum change). Among the plurality of authentication information extraction regions 420, an object closest to the adjustment center object (minimum change) may be selected as the authentication information extraction region 420 and used to generate authentication information.

When generating standard authentication information from an object image (standard authentication information), standard authentication information can be generated for each of a plurality of authentication information extraction areas, and authentication request information is extracted from the object image (authentication request) when performing the authentication procedure. In the step, authentication request information extracted from at least one authentication information extraction region selected from among the plurality of authentication information extraction regions may be used for an authentication procedure.

5 is a conceptual diagram showing a method for extracting authentication information from an object image according to an embodiment of the present invention.

In FIG. 5, a method for extracting standard authentication information from an object image (standard authentication information) and authentication request information from an object image (authentication request) is disclosed.

The object may be an animal such as a dog or a cat, and a method for extracting authentication information (standard authentication information, authentication request information) based on an image of an animal is disclosed.

Referring to FIG. 5, the authentication information may include curvature information and color information centered on an adjustment center object.

The curvature information may include information on curvature of the face based on the adjustment center object. For example, by determining the curve of the face based on the eyes, ears, and nose, a plurality of bending vectors 500 may be generated, and the plurality of bending vectors 500 are used for the curve of the face through length and direction. May contain information.

Also, the color information may be determined based on the color vector 520 determined based on the adjustment center object. For example, the color vector 520 determined based on the adjustment center object may include color information according to changes in direction and length. As shown in the illustration in the example, a color vector 520 including color information according to a change in a region within a specific region may be determined, and sub color information for each of a plurality of sub regions determined based on the color vector 520 is extracted. Can be.

The color vector may be located in an area in which a color change range is clearly greater than or equal to a threshold value in the object image, but an area (or length) in which the color change range exists is greater than or equal to the threshold width (or length).

For example, if a dog's hair is mixed with black, white, and brown, the color change clearly exists above the threshold value so that the color characteristics of black, white, and brown can be recognized, and the color change range exists. A color vector may be located on an area in which the area (or length) is equal to or larger than the critical area (or length). A plurality of such color vectors may be generated for one object.

The color vector according to the exemplary embodiment of the present invention may be a vector that further includes not only a length characteristic, a direction characteristic, but also an area characteristic. That is, it may be a color vector that can be extracted including not only the color change in the first direction (eg, in the vertical direction) but also the color change in the second direction (eg, in the horizontal direction). Based on this color vector, one color vector may be defined to include characteristics of color change in a plurality of directions.

Such curvature information and color information may be used as authentication request information based on an object image (authentication request) and extract standard authentication information based on an object image (standard authentication information).

According to an embodiment of the present invention, among the authentication information for each object, object-specific authentication information that only an object has may be set. For example, authentication information a, authentication information b, and authentication information c may exist as individual A authentication information, and authentication information d, authentication information e, and authentication information f may exist as individual B authentication information. In this case, when the similarity between authentication information of different objects is compared and the similarity between authentication information is less than or equal to a threshold value, it is classified as unique authentication information, and the lower the similarity is, the higher the comparison priority can be used. When performing a comparison between the reference authentication information and the authentication request information, the unique authentication information is compared with priority, and after that, the authentication information having a higher comparison priority may be compared first. According to the strength of the authentication, the set of authentication information to be compared and matched may also be changed.

The embodiments according to the present invention described above may be implemented in the form of program instructions that can be executed through various computer components and recorded in a computer-readable recording medium. The computer-readable recording medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded in the computer-readable recording medium may be specially designed and configured for the present invention or may be known and usable to those skilled in the computer software field. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical recording media such as CD-ROMs and DVDs, and magnetic-optical media such as floptical disks. medium), and a hardware device specially configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of the program instructions include not only machine language codes such as those produced by a compiler but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware device can be changed to one or more software modules to perform the processing according to the present invention, and vice versa.

In the above, the present invention has been described by specific matters such as specific elements and limited embodiments and drawings, but this is provided only to help a more general understanding of the present invention, and the present invention is not limited to the above embodiments. Anyone with ordinary knowledge in the technical field to which the invention belongs can make various modifications and changes from these descriptions.

Accordingly, the spirit of the present invention is limited to the above-described embodiments and should not be defined, and all ranges equivalent to or equivalently changed from the claims to be described later as well as the claims to be described later are the scope of the spirit of the present invention. It will be said to belong to.

Claims (1)

The image-based object recognition method,
Receiving, by the authentication server, an object image (reference authentication information) from a user device; And
And extracting, by the authentication server, reference authentication information based on the object image (reference authentication information).

Images