Method for registrating and identifying information of pet and pet identification system

Abstract

The present invention provides a method for registering information of a pet animal that can link and manage information of a pet animal and a pet owner, a method for identifying information of a pet animal, and a pet animal recognition system that can register and identify information of a pet animal. The method for registering identity information of a pet animal according to the present invention comprises the steps of: acquiring a registration image containing a pet animal and a pet owner; extracting first feature information related to the pet animal and second feature information related to the pet owner from the registration image; generating association information between the first feature information and the second feature information; and registering the association information in a database.

Description

Technology Field

[0001] The present invention relates to a method for registering information of a companion animal, a method for identifying information of a companion animal, and a companion animal recognition system capable of registering and identifying information of a companion animal. Background Technology

[0002] In modern society, there is a growing demand for companion animals that can live alongside humans and provide emotional support. To provide various supplementary services, such as life cycle management and health care, there is an increasing need to manage information about companion animals by creating databases. This requires individual identification information, and specific objects that can be used can be defined for each type of animal. For example, similar to human fingerprints, a dog's nose print (the shape of the nose wrinkles) is known to differ from individual to individual, and it is generally known that once a dog reaches adulthood (at six months of age), these characteristics do not change for the remainder of its life. Therefore, in the case of dogs, nose prints can be utilized as a means of identification.

[0003] However, unlike humans who can assume specific postures or use various sensors, it is not easy for companion animals to do so. Since verbal communication is impossible, methods such as using physical force or tools like restraint frames to forcibly fix the animal in a specific posture or make contact with sensors could be considered; however, this is considered undesirable in terms of protecting animal rights and avoiding resistance from pet owners. Furthermore, since both a person to restrain the animal and a person to operate the biometric data acquisition device are required, there is a practical problem in that it is very difficult to collect biometric data alone.

[0004] As society becomes one where humans and animals coexist, situations requiring the linking of information between companion animals and their owners are occurring frequently. For example, when entering a pet facility or when a pet boarding facility (such as a dog hotel) is looking for a pet, the facility has no way to verify if the person is indeed the owner. Furthermore, regarding the issue of abandonment, cases where individuals deny ownership of the pet are occurring frequently. The problem to be solved

[0005] The present invention provides a method for registering information of a pet that can link and manage information of a pet and a pet owner, a method for identifying information of a pet, and a pet recognition system that can register and identify information of a pet.

[0006] The problems solved by the present invention are not limited to those mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the description below. means of solving the problem

[0007] A method for registering identity information of a companion animal according to the present invention comprises: a step of acquiring a registration image including a companion animal and a companion animal; a step of extracting first feature information related to the companion animal and second feature information related to the companion animal from the registration image; a step of generating association information between the first feature information and the second feature information; and a step of registering the association information in a database.

[0008] According to the present invention, the step of extracting the first feature information and the second feature information comprises: a step of normalizing the registered image to within a preset range of size and pixel values; and a step of extracting the first feature information and the second feature information from the normalized registered image.

[0009] According to the present invention, the step of extracting the first feature information and the second feature information comprises: inputting the registered image into a convolutional neural network having a plurality of stages; extracting a plurality of first feature region candidates and second feature region candidates of different scales using the convolutional neural network; detecting the location and type of a first feature object for biometric recognition of the pet in the first feature region candidates and detecting the location of a second feature object for biometric recognition of the pet owner in the second feature region candidates; and detecting a first feature region containing the first feature object and a second feature region containing the second feature object.

[0010] According to the present invention, the step of generating the association information comprises: a step of calculating a first hash value of the first feature information; a step of calculating a second hash value of the second feature information; a step of embedding the first hash value into the second feature information; and a step of embedding the second hash value into the first feature information.

[0011] According to the present invention, the step of embedding the second hash value into the first feature information comprises: merging the first feature information and the second hash value to generate first merged feature information; calculating a first merged hash value of the first merged feature information; and merging the first feature information and the first merged hash value to generate first dual feature information.

[0012] According to the present invention, the step of embedding the first hash value into the second feature information comprises: merging the second feature information and the first hash value to generate second merged feature information; calculating a second merged hash value of the second merged feature information; and merging the second feature information and the second merged hash value to generate second dual feature information.

[0013] A pet recognition system is provided that includes a mobile device and a server configured to perform a method for registering biometric information of a pet according to the present invention.

[0014] A method for identifying the identity information of a pet animal according to the present invention comprises: a step of acquiring an identification image including a pet animal and a pet owner; a step of extracting first feature information related to the pet animal and second feature information related to the pet owner from the identification image; a step of searching a database for pet animal identity information and pet owner identity information that match the first feature information and the second feature information; and a step of outputting the identity information of the pet animal and the pet owner when pet animal identity information and pet owner identity information that match the first feature information and the second feature information are found.

[0015] According to the present invention, the step of searching for the pet identity information and the pet owner identity information comprises: a step of searching for the pet identity information that matches the first feature information in the database; and a step of searching for the pet owner identity information that matches the second feature information in the database when the pet identity information that matches the first feature information is found.

[0016] According to the present invention, the step of searching for the pet identity information and the pet owner identity information comprises: a step of searching for the pet owner identity information that matches the second feature information in the database; and a step of searching for the pet identity information that matches the first feature information in the database when the pet owner identity information that matches the second feature information is found.

[0017] According to the present invention, the step of searching for the pet identity information and the pet owner identity information comprises: extracting a first feature vector from the first feature information; sorting animal feature vectors registered in the database in order of high similarity to the first feature vector; and outputting pet identifier information of the animal feature vectors in order of high similarity to the first feature vector.

[0018] According to the present invention, the step of searching for the pet identity information and the pet owner identity information comprises: a step of extracting a second feature vector from the second feature information; a step of sorting the pet owner feature vectors registered in the database in order of high similarity to the second feature vector; and a step of outputting the pet owner identifier information of the pet owner feature vectors in order of high similarity to the second feature vector.

[0019] A pet recognition system comprising a mobile device and a server configured to perform a method for identifying the biometric information of a pet according to the present invention is provided. Effects of the invention

[0020] According to the present invention, information about the pet and the owner can be managed by associating them with each other by extracting characteristic information of the pet and the owner, respectively, and then registering association information in a database.

[0021] The effects of the present invention are not limited to those mentioned above, and other unmentioned effects will be clearly understood by those skilled in the art from the description below. Brief explanation of the drawing

[0022] FIGS. 1 and FIGS. 2 illustrate a pet recognition system according to the present invention. FIG. 3 is a flowchart illustrating a method for registering information of a companion animal according to the present invention. Figures 4 and 5 are flowcharts illustrating examples of the process of detecting feature information. Figure 6 is a flowchart illustrating the process of normalizing an image. Figure 7 is an exemplary diagram illustrating the process of normalizing an image. Figures 8 and 9 illustrate an example of the process of detecting feature regions through a convolutional neural network. FIGS. 10 to 12 illustrate examples of the process of generating association information between a pet and its owner. FIGS. 13 and 14 illustrate the procedure of a method for registering information of a pet animal performed by a pet animal recognition system according to the present invention. FIG. 15 is a flowchart illustrating a method for identifying information of a companion animal according to the present invention. FIGS. 16 to 19 are flowcharts illustrating the process of searching for information on companion animals and their owners. FIGS. 19 and 20 illustrate the procedure of a method for identifying information of a pet animal performed by a pet animal recognition system according to the present invention. Specific details for implementing the invention

[0023] Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings so that those skilled in the art can easily implement the present invention. The present invention may be embodied in various different forms and is not limited to the embodiments described herein.

[0024] To clearly explain the present invention, parts unrelated to the explanation have been omitted, and the same reference numerals are used for identical or similar components throughout the specification.

[0025] In addition, in various embodiments, components having the same configuration are described using the same reference numerals only in the representative embodiment, and in other embodiments, only configurations different from the representative embodiment are described.

[0026] Throughout the specification, when a part is described as being "connected (or combined)" with another part, this includes not only cases where they are "directly connected (or combined)" but also cases where they are "indirectly connected (or combined)" with other members in between. Furthermore, when a part is described as "comprising" a certain component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components.

[0027] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the present invention pertains. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this application.

[0028] The present invention provides a method for registering information of a pet animal, a method for identifying information of a pet animal, and a pet animal recognition system (10) capable of registering and identifying information of a pet animal. Biometric information of a pet animal is acquired for the identification of the pet animal. Biometric information of a pet animal may include an iris, facial features, body features, and nose wrinkle features (nose print). Although this document focuses on the use of a dog's nose print as the biometric information of a pet animal, the scope of pet animals in the present invention is not limited to dogs, and various physical features of pet animals may be used as features used as identification information, not limited to nose prints.

[0029] Referring to FIG. 1, a pet recognition system (10) includes a mobile device (1) that captures images to acquire biometric information of a pet and performs processing on the images, and a server (2) that stores and manages biometric information of a pet. The mobile device (1) may be a device such as a smartphone, tablet, or PC. The server (2) may collect data from multiple mobile devices (1) and provide data requested by the mobile devices (1).

[0030] Referring to FIG. 2, a mobile device (1) according to the present invention includes a display (20), a camera (30), a processor (40), and a memory (50). Optionally, the mobile device (1) may further include a gyroscope sensor (71), a touch sensor (72), or an ambient light sensor (73). Optionally, the mobile device (1) may further include a communication module (60), a speaker (81), or a microphone (82).

[0031] The display (20) is a device that outputs visual information. The display (20) may include a panel and a control circuit. The panel may be equipped with a touch sensor (72). In particular, the display (20) may output an image of the pet being filmed (preview image, captured frame). Additionally, the display (20) may output an interface that provides information to the user during the process of acquiring the pet's biometric information.

[0032] The camera (30) is a device for capturing images (still images or video images). The camera may include one or more image sensors, lenses, camera controllers, image processing processors, and flashes. The camera may capture images for acquiring biological information of a pet. The characteristics of the camera (e.g., focus, exposure) may be controlled so that feature objects for acquiring biological information of the pet can be properly identified. The camera (30) may include not only a camera built into the mobile device (1) but also an external camera coupled to the mobile device (1) via a wired or wireless connection.

[0033] The processor (40) can perform computations and data processing to perform operations of the mobile device (1), and control each module of the mobile device (1). The processor (40) can execute one or more software programs stored in memory (50). The processor (40) may include a general-purpose processor (e.g., CPU) that runs an operating system managing the overall operation of the mobile device (1). Additionally, the processor (40) may include a dedicated processor (e.g., GPU, NPU) for specific computation processing. The processor (40) may include one or more processing circuits. The processor (40) can perform data processing and control operations to acquire biological information of a pet.

[0034] The memory (50) can store programs such as an operating system and applications for the operation of the mobile device (1), and data used during the execution of the programs. The memory (50) may include volatile memory (e.g., DRAM) and non-volatile memory (e.g., flash memory). The memory (50) can store images for acquiring biological information of a pet.

[0035] The communication module (60) provides an environment for the mobile device (1) to communicate with an external entity (e.g., server (2), base station, IoT device). The communication module (60) may include an antenna, an RF circuit, and a communication processor. The communication module (60) may provide data received from an external entity to the processor (40). The communication module (60) may convert the data provided by the processor (40) and transmit it to the external entity. The communication module (60) may receive video data for biometric authentication of a pet provided by the processor (40) and transmit the video data to the server (2). The mobile device (1) may communicate with the server (2). The mobile device (1) may acquire biometric information of the pet and transmit the acquired biometric information to the server (2). The server may store the biometric information in a database. The server (2) may perform data processing for learning or identification of the pet from the biometric information. The server (2) can generate identification information for the pet from the biometric information and transmit the identification information for the pet to the mobile device (1).

[0036] The server (2) may include a communication module (160) capable of exchanging data with a mobile device (1), a processor (140) that performs processing on the data, and a memory (150) that stores the data. The server (2) may store the identity information of the pet using the memory (150) as a database, or may store the identity information of the pet in an external database.

[0037] In the registration and identification of identity information using the biometric data of companion animals, the characteristics of the person (owner) and the animal differ, and since the selection of the animal to be adopted occurs randomly by the owner, there are no elements that can be correlated between the characteristic information obtained from the person and the animal. For example, there can be no correlation between a specific owner's fingerprint and the pattern of that owner's dog's nose print. To link these two completely unrelated characteristic pieces of information, a correlation can be generated through a simultaneous acquisition process. In other words, by ensuring that the two pieces of characteristic information existed together at the same time and place, it can be said that the two pieces of characteristic information have a close correlation.

[0038] In order to perform the above process, it is necessary to ensure that the two feature information existed at the same time and place, that is, that temporal and spatial simultaneity is guaranteed. To this end, the present invention proposes a method that utilizes object detection technology to acquire the two feature information while ensuring temporal and spatial simultaneity. The method for registering information about a pet and the method for identifying information about a pet described below may be performed by a processor (40) of a mobile device (1) and may also be performed by a processor (140) of a server (2). Furthermore, the method for registering information about a pet and the method for identifying information about a pet described below may be performed together by the mobile device (1) and the server (2).

[0039] FIG. 3 is a flowchart illustrating a method for registering information of a pet animal according to the present invention. The method for registering information of a pet animal according to the present invention includes the step of obtaining a registration image including a pet animal and a pet owner (S310); the step of extracting first feature information related to the pet animal and second feature information related to the pet owner from the registration image (S320); the step of generating association information between the first feature information and the second feature information (S330); and the step of registering the association information in a database (S340).

[0040] In step S310, a registered image containing a pet and a pet owner is captured by the camera (30) of the mobile device (1), and in step S320, feature information (object images) corresponding to the pet and the pet owner are respectively extracted from the registered image. For example, the first feature information related to the pet may be an image of the pet's nose, and the second feature information related to the pet owner may be a facial image. If both pieces of information (objects corresponding to the pet and the pet owner) are detected in a single input image (registered image), temporal and spatial simultaneity is guaranteed, so subsequent procedures can be performed; otherwise, simultaneity is not guaranteed, so the acquisition of identification information can be stopped.

[0041] In step S330, hash information (hash values) is calculated for each of the pet's characteristic information and the owner's characteristic information. Here, in order to link pairs of mutually corresponding characteristic information regions, the hash information of the counterpart can be embedded in the characteristic information. That is, the association information consists of the first characteristic information (pet's characteristic information) in which the hash value of the second characteristic information (owner's characteristic information) is embedded, and the second characteristic information (owner's characteristic information) in which the hash value of the first characteristic information (pet's characteristic information) is embedded. In step S340, the association information, namely the first characteristic information (pet's characteristic information) in which the hash value of the second characteristic information (owner's characteristic information) is embedded, and the second characteristic information (owner's characteristic information) in which the hash value of the first characteristic information (pet's characteristic information) is embedded, is stored in the database.

[0042] Referring to FIG. 4, the step of extracting first feature information and second feature information (S320) includes the step of normalizing a registered image within a preset range of size and pixel values ​​(S410) and the step of extracting feature information (first feature information and second feature information) from the normalized registered image (S420). Here, the feature information includes first feature information related to a pet and second feature information related to a pet owner. That is, the step of extracting first feature information and second feature information (S320) includes the step of normalizing a registered image within a preset range of size and pixel values ​​(S410) and the step of extracting first feature information from the normalized registered image (S420). Additionally, the step of extracting first feature information and second feature information (S320) includes the step of normalizing a registered image within a preset range of size and pixel values ​​(S410) and the step of extracting second feature information from the normalized registered image (S420). The extraction of the first feature information and the second feature information may be performed in parallel or sequentially.

[0043] FIG. 5 illustrates a process for extracting a nose print image of a dog as feature information of a companion animal. For example, the step of extracting a nose print image of a dog as a step (S320) of extracting first feature information and second feature information includes a step (S510) of normalizing an image (registered image or identification image) containing a dog's face, a step (S520) of detecting a face region within the image, a step (S530) of extracting an image of the face region by cutting out the face region detected in the image, a step (S540) of detecting a nose region within the extracted face region, and a step (S550) of extracting an image of the nose region by cutting out the nose region within the face region. In addition to such two steps of feature region detection, feature regions may be detected through three or more steps.

[0044] Prior to the extraction of feature information, image processing may be performed to detect feature information (e.g., images of a dog's nose print, images of a human face). Normalization for image processing may be executed, and the normalization process includes color space correction, cropping for image size adjustment, and resizing.

[0045] FIG. 6 is a flowchart illustrating the process of normalizing an image. According to the present invention, the step of normalizing an image (S510) includes the step of cropping the image at a set ratio (S610), the step of adjusting the size of the cropped image (S620), and the step of normalizing the pixel value range for the resized image. Referring to FIG. 7, an image having a size of 3,000 x 40,000 pixels is input as shown in FIG. 7 (a). A processor (40, 140) can crop the image according to a set ratio (1:1). The procedure for cropping the image can be executed through the following process.

[0046] L min = MIN(WIDTH, HEIGHT)

[0047] OFFSET x= (WIDTH - L min ) / 2

[0048] OFFSET y = (HEIGHT - L min ) / 2

[0049] CROP_BOX = BOX(LEFT=OFFSET x , TOP=OFFSET y ,

[0050] RIGHT=OFFSET x +L min , BOTTOM=OFFSET y +L min

[0051] First, the smaller value between the image width and height is the minimum length (L min It is entered as ). Minimum length (L) in width (WIDTH). min The value obtained by subtracting ) and dividing by 2 is the horizontal offset (OFFSET x As ), minimum length (L) at height (HEIGHT) min The value obtained by subtracting ) and dividing by 2 is the vertical offset (OFFSET y It is input as ). Subsequently, cropping (CROP_BOX) is executed on the image. The left edge (LEFT) of the image is the horizontal offset (OFFSET x The position of ), the top (TOP) is the vertical offset (OFFSET y The position of ), the right edge (RIGHT) is the horizontal offset (OFFSET x Minimum length (L) at ) min The position where ) is added, the bottom (BOTTOM) is the vertical offset (OFFSET y Minimum length (L) at ) min Cropping is performed based on the location where ) is added. In this way, as shown in Fig. 7(b), cropping can be performed at a 1:1 size while maintaining the center point of the image.

[0052] Afterwards, the size of the preview image (100) cropped as in Fig. 7 (b) is transformed as in Fig. 7 (c). Various methods can be used to adjust the size of the image, for example, bilinear interpolation can be applied.

[0053] Pixel value range normalization is performed on the scaled image. For example, the pixel values ​​of the image can be scaled to a range of [0, 255] with UINT8 format. The input format (normalization format) of the object detector can vary. The input format of the object detector is determined during the design and training phases. For example, if the object detector is set to a range of [-1.0, 1.0] in FLOAT32 format, pixel values ​​in the range of [0, 255] with UINT8 format can have their pixel value range scaled to a range of [-1.0, 1.0] in FLOAT32 format through the following process.

[0054] P`(x,y) = (P(x,y) / 127.5 - 1.0)

[0055] Through this process, the normalized image can finally be input into an object detector that detects feature regions for the registration and identification of companion animals and their owners.

[0056] The process of extracting feature information for identity registration and identification from a normalized image, that is, the process of detecting a feature region containing a feature object, can be executed as shown in FIGS. 8 and 9.

[0057] As illustrated in FIG. 8, the step of extracting first feature information and second feature information (S320) includes the step of inputting an image (registered image or identification image) into a convolutional neural network having multiple stages (S810), the step of extracting multiple feature region candidates of different scales (first feature region candidates and second feature region candidates) using the convolutional neural network (S820), the step of detecting the location and type of a feature object for biometric recognition in the feature region candidates (the step of detecting the location and type of a first feature object for biometric recognition of the pet in the first feature region candidates and detecting the location of a second feature object for biometric recognition of the owner in the second feature region candidates) (S830), and the step of detecting a feature region containing a feature object (a first feature region containing a first feature object and a second feature region containing a second feature object) (S840). Here, the feature objects for biometric recognition may each include a feature object of the pet (e.g., nose) and a feature object of the owner (e.g., face).

[0058] That is, the step of extracting first feature information and second feature information (S320), which is the step of extracting first feature information corresponding to the feature information of a pet, comprises the step of inputting a registration image (or identification image) into a convolutional neural network having multiple stages (S810), the step of extracting multiple first feature region candidates of different scales using the convolutional neural network (S820), the step of detecting the location and type of a first feature object for biometric recognition of a pet from the first feature region candidates (S830), and the step of detecting a feature region containing the first feature object (S840).

[0059] The step of extracting first feature information and second feature information (S320), wherein the second feature information corresponding to the feature information of the companion includes the step of inputting a registered image (or identification image) into a convolutional neural network having multiple stages (S810), the step of extracting multiple second feature region candidates of different scales using the convolutional neural network (S820), the step of detecting the location and type of a second feature object for biometric recognition of the companion from the second feature region candidates (S830), and the step of detecting a feature region containing the second feature object (S840).

[0060] Referring to FIG. 9, a normalized preview image is input into a convolutional neural network during the preprocessing stage (S910). The convolutional neural network consists of several stages, and in each stage, feature object candidates of the input preview image (100) are detected through operations of different scales (S920, S930). Here, the location and type of the feature object candidate (e.g., eyes, nose, ears) are detected by calculating the overlap of the regions of the feature object candidates extracted through the operations of the stages, so that the feature region with the highest probability of containing the feature object is detected (S940). The weight parameters used in the convolutional neural network can be mechanically generated through the learning process.

[0061] When a feature region is detected in an image through the process described above, the processor (40, 140) can perform a validation check to determine whether the image of the feature region is suitable for biometric recognition of the pet and the owner. An inter-linking procedure for identity registration can be performed for an image suitable for biometric recognition of the pet and the owner.

[0062] To interconnect pairs of corresponding feature information regions for pets and humans, methods such as assigning the same ID may be used.

[0063] As a more advanced method, a hash-based approach can be used to verify that the consistency between two sets of feature information is negated if one set is modified. By generating a corresponding hash value for a detected feature region and cross-embedding that hash value into the other region, consistency can be ensured and it can be confirmed that the photos match. By doing so, if one photo is modified, a hash mismatch occurs, thereby negating temporal and spatial consistency. In this document, internalization (embedding) refers to storing different information by associating it with one another. In other words, internalization (embedding) means merging or linking different information to store it.

[0064] A hash value is a value calculated by inputting specific data into a hash function, which is a function that takes data as input and converts it into an arbitrary number. When different data is input into a hash function, they are converted into different arbitrary numbers. In other words, a hash function is a function that maps an input value to another value within a certain range. As an example of a hash function, a modular function can be used, which outputs the remainder of dividing an input value by the size of a table as the hash value; in addition, various other widely known hash functions can be used.

[0065] Referring to FIG. 10, the step of generating association information (S330) includes the step of calculating a hash value of feature information (S1010) and the step of embedding the hash value into the counterpart feature information (S1020). That is, the step of generating association information (S330) includes the step of calculating a first hash value of the first feature information, the step of calculating a second hash value of the second feature information, the step of embedding the first hash value into the second feature information, and the step of embedding the second hash value into the first feature information.

[0066] As illustrated in FIG. 11, the nose image (100) of a pet and the face image (200) of a pet owner are input as feature information extracted from an image, respectively, as first feature information and second feature information, and the hash value of each other is internalized and stored in the feature information. Referring to FIG. 11 (a), the hash value of the second feature information is calculated, and the hash value of the second feature information and the first feature information are associated (internalized) and stored. Referring to FIG. 11 (b), the hash value of the first feature information is calculated, and the hash value of the first feature information and the second feature information are associated (internalized) and stored.

[0067] In addition, a method using double hashing as shown below may also be used.

[0068] Save = Image + HASH(Image + HASH(Relative Image))

[0069] Once feature information is extracted in this way, cross-verification becomes possible by utilizing independent identification recognition technologies. For example, one can prove ownership of a dog by authenticating a person's face and conditionally verifying the dog's paw print associated with that face, then confirming the consistency between the two authentication results. Conversely, one can verify ownership by first authenticating the dog's paw print and then conditionally verifying that the person's face is associated with that print. Furthermore, by simultaneously authenticating both the pet and the owner, information on both the dog and its owner can be acquired at the same time, thereby achieving higher security performance.

[0070] According to the present invention as illustrated in FIG. 12, the step (S1020) of embedding a hash value into the relative feature information comprises: a step (S1210) of merging the feature information (image) and the hash value (HASH (relative image)) of the relative feature information to generate merged feature information (image + HASH (relative image)); a step (S1220) of calculating the merged hash value (HASH (image + HASH (relative image))) of the merged feature information; and a step (S1230) of merging the feature information (image) and the merged hash value (HASH (image + HASH (relative image))) to generate double feature information (image + HASH (image + HASH (relative image))).

[0071] That is, the step of internalizing a second hash value (hash value of owner characteristic information) into the first characteristic information (pet characteristic information) includes the step of generating first merged characteristic information by merging the first characteristic information and the second hash value, the step of calculating a first merged hash value of the first merged characteristic information, and the step of generating first dual characteristic information by merging the first characteristic information and the first merged hash value.

[0072] The step of internalizing a first hash value (hash value of pet characteristic information) into second characteristic information (pet owner characteristic information) includes the step of generating second merged characteristic information by merging the second characteristic information and the first hash value, the step of calculating a second merged hash value of the second merged characteristic information, and the step of generating second dual characteristic information by merging the second characteristic information and the second merged hash value.

[0073] The process of associating the characteristic information of the pet and the owner described above with each other and registering it in a database can be performed by a pet recognition system (10) including a mobile device (1) and a server (2). As illustrated in FIG. 13, the mobile device (1) captures and acquires a registration image including the pet and the owner (S1310), and the registration image can be transmitted from the mobile device (1) to the server (2). The server (2) extracts characteristic information of the pet and the owner from the registration image, respectively (S1330), generates association information by associating the extracted characteristic information with each other (S1340), and registers the generated association information in a database (S1350).

[0074] Alternatively, as illustrated in FIG. 14, the mobile device (1) may capture and acquire a registered image including a pet and a pet owner (S1410), extract feature information of the pet and the pet owner from the registered image (S1420), generate association information by associating the extracted feature information with each other (S1430), store the generated association information (S1440), and transmit the association information to the server (2) (S1450). The server (2) that receives the association information may register the association information in a database (S1460).

[0075] The characteristic information of the registered pet and its owner (first characteristic information and second characteristic information) obtained through the aforementioned process is stored in a database and can be used for the identification of the pet in the future. In particular, since the owner's information (pet owner information) is stored in association with the pet, the association between the pet and its owner can be verified in the future, enabling more reliable management of the pet's identity information.

[0076] FIG. 15 is a flowchart illustrating a method for identifying information of a pet animal according to the present invention. The method for identifying identity information of a pet animal according to the present invention includes the step of acquiring an identification image including a pet animal and a pet owner (S1510), the step of extracting first feature information related to the pet animal and second feature information related to the pet owner from the identification image (S1520), the step of searching for pet animal identity information and pet owner identity information that match the first feature information and second feature information in a database (S1530), and the step of outputting pet animal and pet owner identity information that match the first feature information and second feature information when pet animal identity information and pet owner identity information that match the first feature information and second feature information is found (S1540).

[0077] In steps S1510 and S1520, the image capture, feature information extraction, and related image processing (normalization) that were performed when registering the identity information of the pet and its owner as previously described are performed similarly.

[0078] In step S1530, identity information of the pet and its owner that matches the feature information of the image is retrieved. While the retrieval of the two feature information may be performed in parallel, a method may be used to retrieve another feature information sequentially when one feature information is retrieved in order to improve search efficiency. The order in which feature information is retrieved first may be changed depending on the system design.

[0079] Referring to FIG. 16, the step of searching for pet information and owner identity information (S1530) includes the step of searching for pet information that matches the first feature information in a database (S1610), and when pet identity information that matches the first feature information is found (S1620), the step of searching for owner identity information that matches the second feature information in a database (S1630). That is, pet identity information is searched first, and when matching pet identity information is found, the search for owner information can be executed.

[0080] Conversely, the owner identity information may be searched first, and if matching owner identity information is found, the search for pet identity information may be performed. Referring to FIG. 17, the step of searching for pet information and owner identity information (S1530) includes the step of searching for owner identity information that matches the second feature information in the database (S1710), and the step of searching for pet identity information that matches the first feature information in the database when owner identity information that matches the second feature information is found (S1720) (S1730).

[0081] The statement that information matching the feature information is found means that identity information is found in the entire information stored in the database, where the similarity with the feature information is greater than or equal to a threshold (e.g., 95%). To output identity information matching the feature information, the server (2) sorts the data in order of similarity to the input feature information and outputs N (e.g., 5) identity information with the highest similarity, or can output identity information with a similarity greater than the threshold.

[0082] As illustrated in FIG. 18, the step (S1530) of searching for pet identity information and pet owner identity information includes the step (S1810) of extracting feature vectors from feature information, the step (S1820) of sorting feature vectors registered in a database in order of high similarity to the feature vectors, and the step (S1830) of outputting identifier information (ID) of the feature vectors in order of high similarity to the feature vectors. This search process can be applied identically to the search for pets and pet owners, respectively. As shown in FIG. 18, search results may be output in order of high similarity, or identity information with a similarity above a threshold may be output as search results. The method of outputting search results may differ for pet owners and pets. For example, search results for pets may be output in order of high similarity, and search results for pet owners may be output as search results of identity information with a similarity above a threshold (or vice versa).

[0083] The step of searching for pet information and owner identity information includes the step of extracting a first feature vector from first feature information, the step of sorting animal feature vectors registered in a database in order of high similarity to the first feature vector, and the step of outputting pet identifier information (animal ID) of the animal feature vectors in order of high similarity to the first feature vector.

[0084] The step of searching for pet information and owner identity information includes the step of extracting a second feature vector from second feature information, the step of sorting owner feature vectors registered in a database in order of high similarity to the second feature vector, and the step of outputting owner identifier information (owner ID) of the owner feature vectors in order of high similarity to the second feature vector.

[0085] The above-described method for identifying the identity information of a pet animal can be performed by a pet animal recognition system comprising a mobile device (1) and a server (2). For example, referring to FIG. 19, the mobile device (1) captures and acquires an identification image including a pet animal and a pet owner (S1910) and transmits the identification image to the server (2) (S1920). The server (2) extracts feature information of the pet animal and the pet owner (first feature information and second feature information) through processing the identification image (S1930) and searches for pet animal identity information and pet owner identity information (S1940). When pet animal identity information and pet owner identity information matching the first feature information and second feature information are found, the pet animal identity information and pet owner identity information are output (S1950), and the pet animal information and pet owner information are transmitted to the mobile device (1) (S1960).

[0086] Referring to FIG. 20 as another example, a mobile device (1) captures and acquires an identification image including a pet and a pet owner (S2010), extracts feature information (first feature information and second feature information) of the pet and pet owner through processing of the identification image (S2020), and transmits the feature information (first feature information and second feature information) to a server (2) (S2030). The server (2) searches for pet identity information and pet owner identity information that match the feature information (first feature information and second feature information) (S2040). When pet identity information and pet owner identity information that match the first feature information and second feature information are found, the server (2) outputs the pet identity information and pet owner identity information (S2050) and transmits the pet identity information and pet owner identity information to the mobile device (1) (S2060).

[0087] The embodiments and drawings attached to this specification merely clearly illustrate a part of the technical concept included in the present invention, and it is obvious that variations and specific embodiments that can be easily deduced by a person skilled in the art within the scope of the technical concept included in the specification and drawings of the present invention are all included within the scope of the rights of the present invention.

[0088] Accordingly, the scope of the present invention should not be limited to the described embodiments, and all things equivalent to or having equivalent variations to the claims set forth below, as well as the claims set forth below, shall be considered to fall within the scope of the concept of the present invention.

Claims

Claim 1 A method for registering identity information of a companion animal comprises: a step of acquiring a registration image in which a first feature object for biometric recognition of the companion animal and a second feature object for biometric recognition of the owner are included in a single frame; a step of extracting first feature information related to the companion animal and second feature information related to the owner from the single frame; a step of generating association information between the first feature information and the second feature information; and a step of registering the association information in a database, wherein the step of generating the association information includes: a step of calculating a first hash value of the first feature information; a step of calculating a second hash value of the second feature information; a step of internalizing the first hash value into the second feature information; and a step of internalizing the second hash value into the first feature information, wherein the step of internalizing the second hash value into the first feature information includes: a step of merging the first feature information and the second hash value to generate first merged feature information; and a step of calculating a first merged hash value of the first merged feature information. A method for registering information of a pet, comprising the step of generating first dual feature information by merging the first feature information and the first merged hash value, and the step of internalizing the first hash value into the second feature information includes: generating second merged feature information by merging the second feature information and the first hash value; calculating a second merged hash value of the second merged feature information; and generating second dual feature information by merging the second feature information and the second merged hash value. Claim 2 A method for registering information of a pet animal according to claim 1, wherein the step of extracting the first feature information and the second feature information comprises: a step of normalizing the registered image to within a preset range of size and pixel values; and a step of extracting the first feature information and the second feature information from the normalized registered image. Claim 3 A method for registering information of a pet animal according to claim 1, wherein the step of extracting the first feature information and the second feature information comprises: inputting the registered image into a convolutional neural network having a plurality of stages; extracting a plurality of first feature region candidates and second feature region candidates of different scales using the convolutional neural network; detecting the location and type of the first feature object in the first feature region candidates and detecting the location of the second feature object in the second feature region candidates; and detecting a first feature region containing the first feature object and a second feature region containing the second feature object. Claim 4 delete Claim 5 delete Claim 6 delete Claim 7 A pet recognition system comprising a mobile device and a server configured to perform the method for registering biometric information of a pet according to paragraph 1. Claim 8 A method for identifying the identity information of a companion animal comprises: a step of acquiring an identification image in which a first feature object for biometric recognition of the companion animal and a second feature object for biometric recognition of the companion animal are included in a single frame; a step of extracting first feature information related to the companion animal and second feature information related to the companion animal from the single frame; a step of searching a database for companion animal identity information and companion animal identity information that match the first feature information and the second feature information; and a step of outputting the companion animal and companion animal identity information when the companion animal identity information and companion animal identity information that match the first feature information and the second feature information are found, wherein the step of searching for companion animal identity information and companion animal identity information comprises: a step of extracting a second registration hash value that was embedded in the first animal feature information of the companion animal stored in the database and a first registration hash value that was embedded in the companion animal feature information of the companion animal; and a step of calculating a first identification hash value from the first feature information extracted from the identification image and a second identification hash value from the second feature information. A method for identifying information of a companion animal, comprising the step of verifying consistency by determining whether the first registered hash value and the first identification hash value match, and whether the second registered hash value and the second identification hash value match. Claim 9 A method for identifying information of a pet, wherein the step of searching for the pet identity information and the pet owner identity information comprises: a step of searching for the pet identity information that matches the first feature information in the database; and a step of searching for the pet owner identity information that matches the second feature information in the database when the pet identity information that matches the first feature information is found. Claim 10 A method for identifying information of a pet, wherein the step of searching for the pet identity information and the pet owner identity information comprises: a step of searching for the pet owner identity information that matches the second feature information in the database; and a step of searching for the pet identity information that matches the first feature information in the database when the pet owner identity information that matches the second feature information is found. Claim 11 A method for identifying information of a pet, wherein the step of searching for the pet identity information and the pet owner identity information comprises: a step of extracting a first feature vector from the first feature information; a step of sorting animal feature vectors registered in the database in order of high similarity to the first feature vector; and a step of outputting pet identifier information of the animal feature vectors in order of high similarity to the first feature vector. Claim 12 A method for identifying information of a pet, wherein the step of searching for the pet identity information and the owner identity information comprises: a step of extracting a second feature vector from the second feature information; a step of sorting the owner feature vectors registered in the database in order of high similarity to the second feature vector; and a step of outputting owner identifier information of the owner feature vectors according to the order of high similarity to the second feature vector. Claim 13 A pet recognition system comprising a mobile device and a server configured to perform the method for identifying the biometric information of a pet according to paragraph 8.

Images