Terminal for recognizing serial number of tire, system, method and program

Abstract

To read a serial number of a tire from an image through character recognition without taking off the tire from a vehicle or removing the tire from a storage location.SOLUTION: A terminal for recognizing a serial number of a tire is configured to: obtain an image including the whole image of the serial number indicated on the tire; detect the whole image of the serial number in the acquired image; detect two screw marks in the whole image; determine the inclination from the coordinates of the upper screw mark and the lower screw mark of the two screw marks; correct the inclination; detect respective characters in the whole image of the serial number corrected; determine whether the whole image of the serial number is inverted or not on the basis of the detected characters; vertically invert the whole image when a determination is made that the whole image of the serial number is inverted; and detect a character region located between the two screw marks to recognize the characters.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a terminal, a system, a method and a program for recognizing a serial number of a tire. [Background technology]

[0002] Generally, tires for vehicles manufactured in general have a unique serial number on the side, called a tire serial number. The tire serial number is a combination of letters and numbers, and contains information such as the tire manufacturer and manufacturing date, which is useful for checking the manufacturing quality and lifespan of the tire.

[0003] For this reason, it is important to accurately read and record the letters and numbers of the tire serial number when tracking tire manufacturing information, identifying the tire, controlling its quality, and tracking its safety, durability, and lifespan. In fact, tires fitted to vehicles are managed using a system that acquires and stores information such as tire serial numbers (Patent Document 1).

[0004] Incidentally, in the system of Patent Document 1, when acquiring and storing a tire serial number, the user must either look at the tire serial number directly and manually input it into a terminal or copy it by hand and input it all together into a terminal, which is a time-consuming process. For this reason, attempts have been made to recognize characters from an image of the tire serial number using a widely used character recognition system, but the accuracy of character recognition is reduced by the marks of the formwork and screws around the serial number.

[0005] In recent years, technology has been developed that recognizes characters from 3D images of tire serial numbers captured by a 3D camera, and by using this technology, manual input by the user is not required when registering tire serial numbers in a system such as that in Patent Document 1 (Non-Patent Document 1). [Prior art documents] [Patent documents]

[0006] [Patent Document 1] JP2019-104473A [Non-patent literature]

[0007] [Non-Patent Document 1] Technical Systems Co., Ltd., “Tire Marking Character Recognition Inspection System”, [online], November 1, 2019, iPROS Co., Ltd., [Retrieved October 2, 2023], Internet<URL:https: / / www.ipros.jp / product / detail / 2000261167 / > Summary of the Invention [Problem to be solved by the invention]

[0008] However, in the technology of Non-Patent Document 1, when recognizing the serial number of a tire, the tire needs to be removed from the vehicle, and recognition cannot be performed by a simple procedure. In addition, since the serial number of the tire needs to be captured in 3D images, it is difficult to handle compared to 2D images because it requires special equipment and the data size is large.

[0009] Therefore, the inventors realized that by performing character recognition on an image of a tire's serial number captured using a terminal or camera, it is possible to recognize the tire's serial number without having to remove the tire from the vehicle or carry it from its storage location, and furthermore, accurate character recognition is possible even with a 2D image.

[0010] In view of these problems, the present invention aims to provide a terminal, system, method, and program for recognizing tire serial numbers that can read the tire serial number from an image by character recognition, without removing the tire from the vehicle or carrying it from the storage location. [Means for solving the problem]

[0011] The present invention provides the following solutions.

[0012] According to a first aspect of the present invention, there is provided a terminal for recognizing a serial number of a tire, comprising: A first acquisition unit that acquires an image including an entire image of the serial number displayed on the tire; a first detection unit that detects an entire image of the serial number in the acquired image and detects two screw marks in the entire image; a correction unit that determines a tilt from coordinates of an upper screw mark and a lower screw mark of the two screw marks and corrects the tilt; a second detector that detects each character within an overall image of the corrected serial number; a first determination unit that determines whether an entire image of the serial number is inverted based on each of the detected characters; an inversion unit that inverts the entire image of the serial number upside down when it is determined that the entire image of the serial number is inverted; a recognition unit that detects a character area between the two screw marks and recognizes the characters; A terminal comprising:

[0013] According to a first feature of the present invention, character recognition can be performed from an image of a tire's serial number captured using an unfixed terminal or camera, eliminating the need to remove the tire from the vehicle or carry it from the storage location when recognizing the tire's serial number, and furthermore, accurate character recognition is possible even with a 2D image.

[0014] The second feature of the present invention is the invention according to the first feature, The first determination unit provides a terminal that determines that the entire image of the serial number is an inverted image based on the detected images of each of the characters and the line connecting the two screw marks.

[0015] According to the second aspect of the present invention, even if the serial number is composed only of characters such as 0, 8, and S, whose inversion cannot be determined, it is possible to determine that the entire serial number is inverted.

[0016] The third feature of the present invention is the invention according to the first feature, The terminal further includes a learning unit that learns an overall image of the serial number in the correct position, an overall image of the serial number at an angle, the two screw marks, and images of each character to create a learned model.

[0017] According to a third aspect of the present invention, it is possible to improve the accuracy of automatic recognition of serial numbers attached to tires by creating a trained model by learning from an overall image of the serial number in the correct position, an overall image of the serial number at an angle, screw marks, and images of each character.

[0018] The fourth feature of the present invention is the invention according to the third feature, The learning unit provides a terminal that tags and learns the inverted image of each character.

[0019] According to a third feature of the present invention, by learning that the images of each detected character are inverted when the entire image of the detected serial number is upside down, it is possible to further improve the recognition accuracy of serial numbers attached to tires.

[0020] The fifth feature of the present invention is the invention according to the fourth feature, A second acquisition unit that acquires an image of the entire tire; A first estimation unit that estimates an overall image of the serial number of the tire in the acquired image based on the trained model and estimates two screw marks in the overall image; A second estimation unit that estimates an image of each character in the entire image of the serial number based on the trained model; A second determination unit that determines that the entire image of the serial number is an inverted image when an inverted character is present in the image of each of the characters estimated based on the trained model; The present invention provides a terminal further comprising:

[0021] According to a fifth aspect of the present invention, even if an image of the entire tire is taken from a distance using a fixed camera or the like, character recognition can be performed from the captured image using learning data. Therefore, when recognizing the tire serial number, there is no need to remove the tire from the vehicle or carry it from the storage location, and even more accurate character recognition is possible even with 2D images.

[0022] Although the present invention is in the category of terminals, similar actions and effects according to the category can be achieved in other categories such as systems. Effect of the Invention

[0023] According to the present invention, it is possible to provide a terminal, system, method and program for recognizing tire serial numbers that can read the tire serial number from an image by character recognition without removing the tire from the vehicle or carrying it from the storage location. [Brief description of the drawings]

[0024] [Figure 1] FIG. 1 is a diagram for explaining an overview of a terminal 1 for recognizing a serial number of a tire according to a first embodiment of the present invention. [Diagram 2] 1 is a configuration diagram of a terminal 1 for recognizing a serial number of a tire according to the present embodiment. [Diagram 3] 4 is a flowchart of a process executed by the terminal 1 for recognizing the serial number of a tire in the present embodiment. [Figure 4] 1A to 1C are diagrams for explaining a process for detecting an overall image of a serial number and screw marks executed by a terminal 1 for recognizing a tire serial number in this embodiment. [Diagram 5] 10A to 10C are diagrams for explaining an inclination correction process executed by the terminal 1 for recognizing the serial number of a tire according to the present embodiment. [Figure 6] 1A to 1C are diagrams for explaining a character image detection process executed by the terminal 1 for recognizing the serial number of a tire in this embodiment. [Figure 7]10 is a diagram for explaining an inversion determination process using a line connecting thread marks, which is executed by the terminal 1 for recognizing the serial number of a tire in this embodiment. FIG. [Figure 8] 1 is a diagram for explaining a serial number inversion process executed by a terminal 1 for recognizing the serial number of a tire in this embodiment. FIG. [Figure 9] 1 is a diagram for explaining a character recognition process executed by a terminal 1 for recognizing a serial number of a tire in this embodiment. FIG. [Figure 10] FIG. 1 is a diagram for explaining an overview of a terminal 1 for recognizing a serial number of a tire according to a second embodiment of the present invention. [Figure 11] 1 is a configuration diagram of a terminal 1 for recognizing a serial number of a tire according to the present embodiment. [Figure 12] 4 is a flowchart of a process executed by the terminal 1 for recognizing the serial number of a tire in the present embodiment. [Figure 13] FIG. 11 is a diagram for explaining a learning process executed by a terminal 1 for recognizing a serial number of a tire according to a second embodiment of the present invention. [Figure 14] FIG. 4 is a diagram for explaining another aspect of the first embodiment of the present invention. [Figure 15] FIG. 11 is a configuration diagram of another aspect of the present embodiment. [Figure 16] FIG. 11 is a diagram for explaining another aspect of the second embodiment of the present invention. [Figure 17] FIG. 11 is a configuration diagram of another aspect of the present embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0025] Hereinafter, the best mode for carrying out the present invention will be described with reference to the drawings. Note that this is merely an example, and the technical scope of the present invention is not limited to this example.

[0026] [First embodiment] [Terminal 1 Overview for Recognizing Tire Serial Numbers] An overview of a terminal 1 for recognizing a tire serial number according to a first embodiment of the present invention will be described with reference to Fig. 1. Fig. 1 is a diagram for explaining the overview of a terminal 1 for recognizing a tire serial number according to a first embodiment of the present invention.

[0027] The terminal 1 for recognizing the serial number of a tire is, for example, a mobile terminal such as a handheld terminal, a smartphone, or a tablet terminal, or a wearable terminal such as a head-mounted display such as smart glasses or a smart watch, and is equipped with an imaging device such as a camera that captures images such as color video and / or still images.

[0028] The terminal 1 may be realized, for example, by one terminal device, or may be realized by a plurality of terminal devices, or may be realized by a virtual device such as a cloud computer.

[0029] Next, an overview of the process executed by the terminal 1 to recognize the serial number of a tire will be described. First, the terminal 1 acquires an image 100 including an overall image of the serial number displayed on the tire (step S11). Specifically, the terminal 1 acquires an image 100 including an overall image 110 of the tire serial number photographed by the user using the terminal 1.

[0030] Next, the terminal 1 detects the overall image 110 of the serial number in the acquired image 100, and detects the two screw marks 120 in the overall image 110 (step S12). Specifically, the terminal 1 detects and extracts the overall image 110 of the tire serial number shown in the acquired image 100, and detects the two screw marks 120 included in the extracted image. There are no particular limitations on the method of detecting and extracting the overall image 110 of the serial number. Furthermore, there are no particular limitations on the method of detecting the two screw marks 120.

[0031] In step S12, the terminal 1 may re-learn the characteristic feature of the overall image 110 of the detected serial number and the characteristic feature of the two detected screw marks 120 by machine learning to update the trained model 300. This re-learning by machine learning may be performed at any time after step S12, or may be performed by the terminal 1 receiving an input from the user. Machine learning will be described later.

[0032] Next, the terminal 1 determines the inclination from the coordinates of the upper screw mark 120 and the lower screw mark 120 of the two detected screw marks 120, and corrects the inclination (step S13). Specifically, the terminal 1 determines the inclination of the detected overall image 110 of the serial number from the upper screw mark 120 and the lower screw mark 120 of the two screw marks 120, and corrects the overall image 110 of the serial number so that the two screw marks 120 are horizontal.

[0033] Next, the terminal 1 detects each character 130 in the overall image 110 of the corrected serial number (step S14). Specifically, the terminal 1 detects and extracts the image of each character 130 in the area 110 of the corrected serial number, character by character. The method of detecting and extracting each image 100 is not particularly limited.

[0034] In step S14, the terminal 1 may re-learn the image-specific feature amount of each detected character 130 by machine learning to update the trained model 300. This re-learning by machine learning may be performed at any time after step S14, or may be performed by the terminal 1 receiving an input from a user. Machine learning will be described later.

[0035] Next, the terminal 1 determines whether the overall image 110 of the serial number is inverted based on each of the detected characters 130 (step S15). Specifically, the terminal 1 performs character recognition on the image of each of the detected characters 130, and determines that the overall image 110 of the serial number is inverted if even one character is not recognized as a character. Furthermore, the terminal 1 may determine that the overall image 110 of the serial number is an inverted image if the coordinates of the center character in the image of each of the detected characters 130 are below the line connecting the two screw marks. The terminal 1 may also display each of the detected characters 130, and determine that the overall image 110 of the serial number is inverted by accepting input from the user if even one of the displayed characters is inverted.

[0036] In step S15, if the terminal 1 determines that the entire image 110 of the serial number is an inverted image, the terminal 1 may tag the image-specific features of each character 130 relearned in step S14 as inverted characters, and relearn using machine learning to update the trained model 300. This relearning using machine learning may be performed at any time after step S15, or may be performed by the terminal 1 receiving an input from a user. Machine learning will be described later.

[0037] Next, if it is determined that the entire image 110 of the serial number is inverted, the terminal 1 flips the entire image 110 upside down (step S16). If it is determined that the entire image 110 of the serial number is inverted, the terminal 1 rotates the entire image 110 by 180 degrees to obtain the entire image 111 of the serial number in the normal position. The normal position means that the entire image 111 of the serial number is in the normal position, and the characters of the serial number are not inverted.

[0038] After step S16, the terminal 1 may re-learn the feature amount specific to the overall image 110 of the serial number in the correct position by machine learning to update the trained model 300. This re-learning by machine learning may be performed at any time after step S16, or may be performed by the terminal 1 receiving an input from the user. Machine learning will be described later.

[0039] Next, the terminal 1 detects the character area 140 between the two screw marks 120 and recognizes the characters (step S17). Specifically, the terminal 1 detects the character area 140 between the two screw marks 120 and recognizes each character in the detected character area 140 as character data. The terminal 1 may output the recognized character data to its own display unit. The method of recognizing the character data is not particularly limited.

[0040] The above is an overview of the process executed by the terminal 1 to recognize the serial number of a tire.

[0041] [System configuration of Terminal 1 for recognizing tire serial numbers] The system configuration of the terminal 1 for recognizing the serial number of a tire according to this embodiment will be described with reference to FIG.

[0042] The terminal 1 may be realized, for example, by one terminal device, or may be realized by a plurality of terminal devices.

[0043] Terminal 1 is, for example, a mobile terminal such as a handheld terminal, a smartphone, or a tablet terminal, or a wearable terminal such as a head-mounted display such as smart glasses or a smart watch, and is equipped with an imaging device such as a camera that captures images such as color video and / or still images.

[0044] Terminal 1 is equipped with a control unit and a processing unit, such as a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a RAM (Random Access Memory), and a ROM (Read Only Memory). The control unit issues execution commands to the processing unit, imaging unit, communication unit, input unit, output unit, and memory unit, which will be described later, and the processing unit calculates data and determines the calculation results.

[0045] The imaging unit includes a device for capturing images such as moving images and / or still images. There are no particular limitations on the type of device as long as it can capture color images.

[0046] The terminal 1 includes a device as a communication unit for enabling communication with other terminals, devices, etc. The communication method may be wireless or wired.

[0047] The terminal 1 is assumed to have, as an input unit, functions necessary for a user to operate the terminal 1. Examples of input devices that can be provided include a liquid crystal display that provides a touch panel function, a keyboard, a mouse, a pen tablet, hardware buttons on the device, and a microphone for voice recognition. The present invention is not particularly limited in function depending on the input method.

[0048] The terminal 1 is assumed to have, as an output unit, functions necessary for a user to operate the terminal 1. Examples of output may include display such as projection on a display unit such as a liquid crystal display, a PC display, or a projector, and audio output. The present invention is not particularly limited in function by the output method.

[0049] The terminal 1 includes, as a storage unit, data storage such as a hard disk, a semiconductor memory, a recording medium, a memory card, etc. The data may be stored in a cloud service, a database, etc.

[0050] The control unit cooperates with the processing unit to realize an acquisition unit 10, a first detection unit 11, a correction unit 12, a second detection unit 13, a first determination unit 14, an inversion unit 15, a recognition unit 16, and a learning unit 17.

[0051] The above is the system configuration of the terminal 1 for recognizing the serial number of a tire.

[0052] [Tire serial number recognition processing] The tire serial number recognition process executed by the terminal 1 for recognizing the tire serial number will be described with reference to Fig. 3. Fig. 3 is a diagram showing a flowchart of the tire serial number recognition process executed by the terminal 1 for recognizing the tire serial number. As shown in Fig. 3, the tire serial number recognition process is made up of steps S31 to S37, which correspond to the above-mentioned steps S11 to S17.

[0053] First, the acquisition unit 10 of the terminal 1 acquires an image 100 including an overall image 110 of the serial number displayed on the tire (step S31). Specifically, the acquisition unit 10 acquires an image 100 including an overall image 110 of the tire's serial number that is photographed by a user using the terminal 1. The image may be a moving image or a still image as long as it is a color image.

[0054] Next, the first detection unit 11 of the terminal 1 detects the overall image 110 of the serial number in the acquired image 100, and detects two screw marks 120 in the overall image 110 (step S32). Specifically, the terminal 1 detects and extracts the overall image 110 of the tire serial number shown in the acquired image 100, and detects two screw marks 120 of a specific shape in the extracted image. There are no particular limitations on the method of detecting and extracting the overall image 110 of the serial number.

[0055] As shown in Fig. 4, the tire serial number is displayed between the two screw marks 120. In this specification, the shape of the two screw marks 120 is a round bump, but it may be any other shape. The method of detecting the two screw marks 120 is not particularly limited.

[0056] In step S32, the learning unit 17 of the terminal 1 may re-learn the characteristic feature of the overall image 110 of the detected serial number and the characteristic feature of the two detected screw marks 120 by machine learning to update the trained model 300. This re-learning by machine learning may be performed at any time after step S32, or may be performed by the input unit of the terminal 1 receiving input from the user. Machine learning will be described later.

[0057] Next, the correction unit 12 of the terminal 1 determines the inclination from the coordinates of the upper screw mark 120 and the lower screw mark 120 of the two screw marks 120, and corrects the inclination (step S33). Specifically, as shown in Fig. 5, the correction unit 12 obtains the angle of the upper screw mark with the lower screw mark 120 of the two screw marks 120 extracted in step S32 as the origin, and rotates the entire image 110 of the serial number in a clockwise or counterclockwise direction by the obtained angle, thereby correcting the entire image 110 of the serial number so that the two screw marks 120 are horizontal. As a result, the entire image 110 of the serial number in which the two screw marks 120 are in a horizontal position is obtained.

[0058] Next, the second detection unit 13 of the terminal 1 detects each character 130 in the overall image 110 of the corrected serial number (step S34). Specifically, as shown in Fig. 6, each character included in the overall image 110 of the corrected serial number is detected one by one, and each character 130 is extracted by surrounding it in a rectangular shape so that it covers the entirety, that is, a rectangular image including each character 130 is extracted. There are no particular limitations on the method of detecting and extracting characters. In this embodiment, each character is surrounded by a rectangle, but it may be surrounded by any other shape.

[0059] In step S34, the learning unit 17 of the terminal 1 may re-learn the image-specific feature amount of each detected character 130 by machine learning to update the trained model 300. This re-learning by machine learning may be performed at any time after step S34, or may be performed by the input unit of the terminal 1 receiving input from the user. Machine learning will be described later.

[0060] Next, the first determination unit 14 of the terminal 1 determines whether the overall image 110 of the serial number is inverted based on each of the detected characters 130 (step S35). Specifically, the first determination unit 14 performs character recognition on the image of each of the characters 130 extracted in step S14, and determines that the overall image 110 of the serial number is inverted if there is an image of a character that is not recognized as a character. Furthermore, the first determination unit 14 may determine that the overall image 110 of the serial number is an inverted image if the coordinates of the center character in the image of each of the detected characters 130 are below the line connecting the two screw marks. Also, the display unit of the terminal 1 may display each of the detected characters 130, and if even one of the displayed characters is inverted, the input unit of the terminal 1 may accept an input from the user, causing the first determination unit 14 to determine that the overall image 110 of the serial number is inverted.

[0061] In step S35, if the learning unit 17 of the terminal 1 determines that the entire image 110 of the serial number is an inverted image, the learning unit 17 may tag the image-specific feature of each character 130 relearned in step S34 as an inverted character, and perform relearning by machine learning to update the trained model 300. This relearning by machine learning may be performed at any time after step S35, or may be performed by the input unit of the terminal 1 accepting input from the user. Machine learning will be described later.

[0062] Next, when it is determined that the entire image 110 of the serial number is inverted, the inversion unit 15 of the terminal 1 inverts the entire image 110 upside down (step S36). Specifically, as shown in Fig. 8, when it is determined that the entire image 110 of the serial number is inverted, the inversion unit 15 rotates the entire image 110 of the serial number by 180° in a clockwise or counterclockwise direction with the center of the entire image 110 of the serial number as the origin, thereby correcting the entire image 110 of the serial number to the correct position. As a result, an entire image 111 of the serial number in the correct horizontal position is obtained, in which each character of the serial number is oriented in the correct direction.

[0063] After step S36, the learning unit 17 of the terminal 1 may re-learn the feature amount specific to the overall image 110 of the serial number in the correct position by machine learning to update the trained model 300. This re-learning by machine learning may be performed at any time after step S36, or may be performed by the input unit of the terminal 1 accepting input from the user. Machine learning will be described later.

[0064] Next, the recognition unit 16 of the terminal detects the character area 140 between the two screw marks 120 and performs character recognition (step S37). Specifically, as shown in Fig. 9, the recognition unit 16 detects the character area 140 between the two screw marks 120 and recognizes each character in the detected character area 140 as character data. The output unit of the terminal 1 may output the recognized character data to its own display unit. The method of recognizing the character data is not particularly limited.

[0065] This completes the tire serial number recognition process.

[0066] According to terminal 1 for recognizing tire serial numbers, character recognition is possible from an image of the tire serial number captured using terminal 1. This means that there is no need to remove the tire from the vehicle or carry it from the storage location when recognizing the tire serial number, and accurate character recognition is possible even with a 2D image.

[0067] In addition, by using the terminal 1 for recognizing the serial number of a tire, it is possible to determine that the entire serial number is inverted even if it is composed only of characters such as 0, 8, and S, whose inversion cannot be determined.

[0068] [Second embodiment] [Terminal 1 Overview for Recognizing Tire Serial Numbers] An overview of a terminal 1 for recognizing a tire serial number according to a second embodiment of the present invention will be described with reference to Fig. 10. Fig. 10 is a diagram for explaining an overview of a terminal 1 for recognizing a tire serial number according to one embodiment of the present invention. Note that the same functions and configurations as those in the first embodiment are given the same reference numerals, and descriptions thereof will be omitted. This embodiment differs from the first embodiment in that the terminal 1 for recognizing a tire serial number uses learned data.

[0069] [Terminal 1 Overview for Recognizing Tire Serial Numbers] An overview of a terminal 1 for recognizing a tire serial number according to a second embodiment of the present invention will be described with reference to Fig. 10. Fig. 10 is a diagram for explaining the overview of a terminal 1 for recognizing a tire serial number according to the second embodiment of the present invention.

[0070] The terminal 1 for recognizing the serial number of a tire, as in the first embodiment, is a terminal such as a mobile terminal such as a handheld terminal, a smartphone, or a tablet terminal, or a wearable terminal such as a head-mounted display such as smart glasses or a smart watch, and is equipped with an imaging device such as a camera that captures images such as color video and / or still images.

[0071] The terminal 1 may be realized, for example, by one terminal device, as in the first embodiment, or may be realized by a plurality of terminal devices, or may be realized by a virtual device such as a cloud computer.

[0072] Next, an overview of the process executed by the terminal 1 to recognize the serial number of a tire will be described. First, the terminal 1 acquires an image 200 of the entire tire (step S21). Specifically, the terminal 1 acquires an image 200 of the entire tire photographed by a user using the terminal 1. This step is the same as step S11 in the first embodiment.

[0073] Next, the terminal 1 estimates the overall image 110 of the tire serial number in the acquired image 200 based on the trained model 300, and estimates the two screw marks 120 in the overall image 110 (step S22). Specifically, the terminal 1 estimates and extracts the overall image 110 of the tire serial number shown in the acquired image 200 based on the trained model 300 created in advance by machine learning, and estimates the two screw marks 120 included in the extracted image. There is no particular restriction on the type or method of machine learning. There is no particular restriction on the method of estimating and extracting the overall image 110 of the serial number. In addition, there is no particular restriction on the method of estimating the two screw marks 120.

[0074] In step S22, the terminal 1 may re-learn the characteristic feature quantity of the overall image 110 of the inferred serial number by machine learning to update the trained model 300. This re-learning by machine learning may be performed at any time after step S22, or may be performed by the terminal 1 receiving an input from the user. Machine learning will be described later.

[0075] Next, the terminal 1 determines the inclination from the coordinates of the upper screw mark 120 and the lower screw mark 120 of the two screw marks 120, and corrects the inclination (step S23). This step is the same as step S13 in the first embodiment, except that the two screw marks 120 are estimated.

[0076] Next, the terminal 1 estimates an image of each character 130 in the overall image 110 of the corrected serial number based on the trained model 300 (step S24). Specifically, the terminal 1 estimates and extracts an image of each character 130 in the overall image 110 of the corrected serial number, character by character, based on the trained model 300. The method of detecting and extracting the image of each character 130 is not particularly limited.

[0077] In step S24, the terminal 1 may re-learn the image-specific feature amount of each of the estimated characters 130 by machine learning to update the trained model 300. This re-learning by machine learning may be performed at any time after step S24, or may be performed by the terminal 1 receiving an input from the user. Machine learning will be described later.

[0078] Next, the terminal 1 determines whether the overall image 110 of the serial number is inverted in the image of each of the predicted characters based on the trained model 300 (step S25). Specifically, the terminal 1 determines that the overall image 110 of the serial number is inverted based on the trained model 300 when the image is based on each of the trained characters 130 tagged as an inverted character.

[0079] In step S25, when the terminal 1 determines that the entire image 110 of the serial number is an inverted image, the terminal 1 may tag the image-specific features of each character 130 relearned in step S24 as inverted characters, and relearn using machine learning to update the trained model 300. This relearning using machine learning may be performed at any time after step S25, or may be performed by the input unit of the terminal 1 accepting input from the user. Machine learning will be described later.

[0080] Next, when it is determined that the entire image 110 of the serial number is inverted, the terminal 1 upside down the entire image 110 (step S26). This step is the same as step S16 in the first embodiment.

[0081] After step S26, the terminal 1 may re-learn the feature amount specific to the overall image 111 of the serial number in the correct horizontal position by machine learning, and update the trained model 300. This re-learning by machine learning may be performed at any time after step S26, or may be performed by the input unit of the terminal 1 receiving an input from the user. Machine learning will be described later.

[0082] Next, the terminal 1 detects the character area 140 between the two screw marks 120 and performs character recognition (step S27). This step is the same as step S17 in the first embodiment.

[0083] The above is an overview of the process executed by the terminal 1 to recognize the serial number of a tire.

[0084] [System configuration of Terminal 1 for recognizing tire serial numbers] The system configuration of the terminal 1 for recognizing the serial number of a tire according to this embodiment will be described with reference to FIG.

[0085] The terminal 1 may be realized, for example, by one terminal device, as in the first embodiment, or may be realized by a plurality of terminal devices.

[0086] As in the first embodiment, the terminal 1 is a terminal such as a mobile terminal such as a handheld terminal, a smartphone, or a tablet terminal, or a wearable terminal such as a head-mounted display such as smart glasses or a smart watch, and is equipped with an imaging device such as a camera that captures images such as color video and / or still images.

[0087] Similar to the first embodiment, the terminal 1 is equipped with a control unit and a processing unit, such as a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a RAM (Random Access Memory), and a ROM (Read Only Memory). The control unit issues execution commands to the processing unit, imaging unit, communication unit, input unit, output unit, and memory unit, which will be described later, and the processing unit calculates data and determines the calculation results.

[0088] The imaging unit includes a device for capturing images such as moving images and / or still images, as in the first embodiment, but is not particularly limited as long as it can capture color images.

[0089] As in the first embodiment, the terminal 1 includes, as a communication unit, a device for enabling communication with other terminals, devices, etc. The communication method may be wireless or wired.

[0090] As in the first embodiment, the terminal 1 is assumed to have, as an input unit, functions necessary for a user to operate the terminal 1. Examples of devices for implementing input include a liquid crystal display for implementing a touch panel function, a keyboard, a mouse, a pen tablet, hardware buttons on the device, and a microphone for performing voice recognition. The present invention is not particularly limited in function depending on the input method.

[0091] As in the first embodiment, the terminal 1 is an output unit having functions necessary for a user to operate the terminal 1. Examples of output may include display such as projection on a display unit such as a liquid crystal display, a PC display, or a projector, and audio output. The present invention is not particularly limited in function by the output method.

[0092] As in the first embodiment, the terminal 1 includes a storage unit such as a hard disk, a semiconductor memory, a recording medium, a memory card, etc. The data may be stored in a cloud service, a database, etc.

[0093] The control unit cooperates with the processing unit to realize an acquisition unit 10, a third acquisition unit 21, a first estimation unit 22, a correction unit 12, a second estimation unit 23, a second judgment unit 24, an inversion unit 15, a recognition unit 16, and a learning unit 17.

[0094] The above is the system configuration of the terminal 1 for recognizing the serial number of a tire.

[0095] [Tire serial number recognition processing] The tire serial number recognition process executed by the terminal 1 for recognizing the tire serial number will be described with reference to Fig. 12. Fig. 12 is a diagram showing a flowchart of the tire serial number recognition process executed by the terminal 1 for recognizing the tire serial number. As shown in Fig. 12, the tire serial number recognition process is made up of steps S41 to S47, which correspond to the above-mentioned steps S21 to S27.

[0096] First, the acquisition unit 10 of the terminal 1 acquires an image 200 of the entire tire (step S41). Specifically, the first acquisition unit 20 acquires an image 200 of the entire tire taken by a user using the terminal 1. The image may be a video or a still image as long as it is a color image. This step is the same as step S31 in the first embodiment.

[0097] Next, the first estimation unit 22 of the terminal 1 estimates the overall image 110 of the serial number of the tire in the acquired image 200 based on the trained model 300, and estimates the two screw marks 120 in the overall image 110 (step S42). Specifically, the first estimation unit 22 estimates and extracts the overall image 110 of the serial number of the tire shown in the acquired image 200 based on the overall image 110 of the serial number with an angle and the two screw marks 120, and estimates the two screw marks 120 included in the extracted image based on the two screw marks 120 that have been trained by machine learning. There is no particular limit to the type or method of machine learning. There is no particular limit to the method of estimating and extracting the overall image 110 of the serial number. There is also no particular limit to the method of estimating the two screw marks 120.

[0098] In step S42, the learning unit 17 of the terminal 1 may re-learn the estimated overall image 110 of the serial number with the characteristic feature of the overall image 110 of the angled serial number and the characteristic feature of the estimated two screw marks 120 by machine learning, thereby updating the trained model 300. This re-learning by machine learning may be performed at any time after step S42, or may be performed by the input unit of the terminal 1 accepting input from the user. Machine learning will be described later.

[0099] Next, the correction unit 12 of the terminal 1 determines the inclination from the coordinates of the upper screw mark 120 and the lower screw mark 120 of the two screw marks 120, and corrects the inclination (step S43). This step is the same as step S33 in the first embodiment, except that two screw marks 120 are estimated.

[0100] Next, the first estimation unit 22 of the terminal 1 detects each character 130 in the character string between the two screw marks 120 (step S44). Specifically, based on the trained model 300, as shown in FIG. 6, the first estimation unit 22 estimates each character included in the overall image 110 of the estimated serial number one by one, and extracts each character 130 by surrounding it in a rectangular shape so that the character 130 covers the entirety, that is, extracts a rectangular image including each character 130. There is no particular limitation on the method of estimating and extracting the image of each character 130. In this embodiment, each character is surrounded by a rectangle, but it may be surrounded by any other shape.

[0101] In step S44, the learning unit 17 of the terminal 1 may re-learn the image-specific feature amount of each inferred character 130 by machine learning to update the trained model 300. This re-learning by machine learning may be performed at any time after step S44, or may be performed by the input unit of the terminal 1 receiving an input from a user. Machine learning will be described later.

[0102] Next, the second determination unit 24 of the terminal 1 determines whether the overall image 110 of the serial number is inverted based on the trained model 300 (step S45). Specifically, when the image of each inferred character 130 that has been trained is tagged as an inverted character, the second determination unit 24 determines that the overall image 110 of the serial number is inverted.

[0103] In step S45, when the learning unit 17 of the terminal 1 determines that the entire image 110 of the serial number is an inverted image, the learning unit 17 may tag the image-specific features of each character 130 relearned in step S44 as inverted characters and relearn them by machine learning to update the trained model 300. Note that this relearning by machine learning may be performed at any time after step S45, or may be performed by the input unit of the terminal 1 accepting input from the user. Machine learning will be described later.

[0104] Next, when it is determined that the entire image 110 of the serial number is inverted, the inverting unit 15 of the terminal 1 inverts the entire image 110 upside down (step S46). This step is the same as step S36 in the first embodiment.

[0105] After step S46, the terminal 1 may re-learn the feature amount specific to the overall image 111 of the serial number in the correct horizontal position by machine learning, and update the trained model 300. This re-learning by machine learning may be performed at any time after step S46, or may be performed by the input unit of the terminal 1 receiving input from the user. Machine learning will be described later.

[0106] Next, the recognition unit 16 of the terminal detects the character area 140 between the two screw marks 120 and performs character recognition (step S47). This step is the same as step S37 in the first embodiment.

[0107] This completes the tire serial number recognition process.

[0108] [Learning process] The learning process executed by the terminal 1 for recognizing the serial number of a tire will be described with reference to Fig. 13. Fig. 13 is a diagram showing a flowchart of the learning process executed by the terminal 1 for recognizing the serial number of a tire.

[0109] The acquisition unit 10 of the terminal 1 acquires a learning image (step S51). Specifically, the acquisition unit 10 acquires an image of an overall image 111 of the serial number in the correct position photographed by the user using the terminal 1. The image may be a moving image or a still image as long as it is a color image. The learning image may be acquired from another terminal, computer, or device via data communication or the like.

[0110] The learning unit 17 of the terminal 1 analyzes the acquired learning image (step S52). Specifically, the learning unit 17 detects and analyzes the overall image 111 of the serial number in the correct position, the overall image 111 of the serial number at an angle, the two screw marks 120, and the images 100 of each character, which are captured in the learning image, and extracts characteristic features of each of these. The learning unit 17 may also execute this step by receiving input from a user via the input unit of the terminal 1.

[0111] For the upside-down image of each character image 100, the learning unit 17 of the terminal 1 creates an inverted image by inverting the image of each character 130 detected in step S52, extracts its feature amount, and tags it as an inverted character. The learning unit 17 may also execute this step by receiving an input from a user via the input unit of the terminal 1.

[0112] The learning unit 17 of the terminal 1 executes machine learning based on the feature amount extracted in step S51 and the feature amount of the upside-down image tagged as an inverted character in step S52 (step S53). Specifically, the learning unit 17 executes supervised learning as the machine learning, using the feature amount extracted in step S51 and the feature amount specific to the upside-down image tagged as an inverted character in step S52 as teachers. The feature amount extracted in step S51 and the feature amount of the upside-down image tagged as an inverted character in step S52 may be learned by deep learning, which is automatically defined and learned by a multi-layered neural network.

[0113] The learning unit 17 of the terminal 1 creates a trained model for recognizing tire serial numbers based on the learning results (step S54). Specifically, the learning unit 17 creates a trained model 300 for recognizing tire serial numbers that incorporates features specific to the overall image 111 of the serial number in the normal position, features specific to the overall image 111 of the serial number at an angle, features specific to the two screw marks 120, features specific to the image 100 of each character, and features specific to the upside-down image tagged as an inverted character in step S52.

[0114] The learning unit 17 of the terminal 1 stores the created trained model 300 in its own storage unit (step S56).

[0115] The above is the learning process.

[0116] Terminal 1 for recognizing tire serial numbers can perform character recognition from an image of the entire tire captured using terminal 1, so there is no need to remove the tire from the vehicle or carry it from the storage location when recognizing the tire serial number, and even more accurate character recognition is possible even with 2D images.

[0117] In addition, the terminal 1 for recognizing tire serial numbers can further improve the accuracy of automatic recognition of serial numbers attached to tires by learning the overall image of the serial number in the correct position, the overall image of the serial number at an angle, screw marks, and images of each character.

[0118] Furthermore, the terminal 1 for recognizing tire serial numbers can learn that when the entire image of the detected serial number is upside down, the images of each detected character are inverted, thereby making it possible to further improve the recognition accuracy of the serial numbers attached to tires.

[0119] [Another aspect of the first embodiment] [Overview of System 2 for Recognizing Tire Serial Numbers] An overview of a system 2 for recognizing a tire serial number, which is another aspect of the first embodiment of the present invention, will be described with reference to Fig. 14. Fig. 14 is a diagram for explaining an overview of a system 2 for recognizing a tire serial number, which is one aspect of the present invention. Note that the same functions and configurations as those of the above-mentioned first embodiment are given the same reference numerals, and descriptions thereof will be omitted. This aspect differs from the first embodiment in that the system 2 for recognizing a tire serial number is composed of a computer 3 and an imaging device 4.

[0120] The system 2 for recognizing tire serial numbers is comprised of the computer 3 and the imaging device 4, as described above.

[0121] The computer 3 may be, for example, a computer such as a desktop personal computer, a notebook computer, or a server, a mobile terminal such as a smartphone or a tablet terminal, or a wearable terminal such as a head-mounted display such as smart glasses or a smart watch.

[0122] The computer 3 may be realized, for example, as one terminal device, or may be realized as a plurality of terminal devices, or may be realized as a virtual device such as a cloud computer.

[0123] The imaging device 4 is, for example, a camera that captures images such as moving images and / or still images, and is not particularly limited as long as it can capture color images.

[0124] The computer 3 and the imaging device 4 are connected to each other via a public line network or the like so as to enable data communication, and transmit and receive necessary data and information.

[0125] Next, an overview of the process executed by the system 2 for recognizing the serial number of a tire will be described. First, the computer 3 acquires an image 100 including an overall image 110 of the serial number displayed on the tire (step S11). Specifically, the process is the same as that of the first embodiment described above, except that the computer 3 acquires an image 100 including an overall image 110 of the tire serial number captured by a user using the imaging device 4. The computer 3 may acquire the image captured by the imaging device 4 via a communication network using a communication unit of the computer 3, or may acquire the image by storing it in a storage unit using external media.

[0126] Next, the computer 3 detects the overall image 110 of the serial number in the acquired image 100, and detects two screw marks 120 in the overall image 110 (step S12). Specifically, this is the same as one aspect of the first embodiment described above.

[0127] Next, the terminal 1 determines the inclination from the coordinates of the upper screw mark 120 and the lower screw mark 120 of the two detected screw marks 120, and corrects the inclination (step S13). Specifically, this is the same as one aspect of the first embodiment described above.

[0128] Next, the terminal 1 detects each character 130 in the overall image 110 of the corrected serial number (step S14). Specifically, this is the same as in one aspect of the first embodiment described above.

[0129] Next, the terminal 1 determines whether the entire image 110 of the serial number is inverted based on each of the detected characters 130 (step S15). Specifically, this is the same as in one aspect of the first embodiment described above.

[0130] Next, when it is determined that the entire image 110 of the serial number is inverted, the terminal 1 upside down the entire image 110 (step S16). Specifically, this is the same as one aspect of the first embodiment described above.

[0131] Next, the terminal 1 detects the character area 140 between the two screw marks 120 and performs character recognition (step S17). Specifically, this is the same as one aspect of the above-mentioned first embodiment.

[0132] The above is an overview of the process executed by the terminal 1 to recognize the serial number of a tire.

[0133] [System configuration of System 2 for recognizing tire serial numbers] The system configuration of the system 2 for recognizing the serial number of a tire according to this embodiment will be described with reference to FIG.

[0134] The computer 3 may be, for example, a computer such as a desktop personal computer, a notebook computer, or a server, a mobile terminal such as a smartphone or a tablet terminal, or a wearable terminal such as a head-mounted display such as smart glasses or a smart watch.

[0135] The computer 3 may be realized, for example, as one terminal device, or may be realized as a plurality of terminal devices, or may be realized as a virtual device such as a cloud computer.

[0136] The computer 3 is equipped with a CPU (Central Processing Unit), GPU (Graphics Processing Unit), RAM (Random Access Memory), ROM (Read Only Memory), etc. as a control unit and a processing unit. The control unit issues execution commands to the processing unit, communication unit, input unit, output unit, and memory unit described below, and the processing unit calculates data and determines the calculation results, etc.

[0137] The computer 3 includes a device as a communication unit for enabling communication with other terminals, devices, etc. The communication method may be wireless or wired.

[0138] The imaging device 4 is, for example, a camera that captures images such as moving images and / or still images, and is not particularly limited as long as it can capture color images.

[0139] The computer 3 is an input unit that has functions necessary for a user to operate the terminal 1. Examples of input devices that can be provided include a liquid crystal display that provides a touch panel function, a keyboard, a mouse, a pen tablet, hardware buttons on the device, and a microphone for voice recognition. The present invention is not particularly limited in function depending on the input method.

[0140] The computer 3, as an output unit, is equipped with functions necessary for the user to operate the terminal 1. Examples of output may include display such as projection on a display unit such as a liquid crystal display, a PC display, or a projector, and audio output. The present invention is not particularly limited in function by the output method.

[0141] The computer 3 includes a data storage unit such as a hard disk, a semiconductor memory, a recording medium, a memory card, etc. The data may be stored in a cloud service, a database, etc.

[0142] The control unit cooperates with the processing unit to realize an acquisition unit 10, a first detection unit 11, a correction unit 12, a second detection unit 13, a first determination unit 14, an inversion unit 15, a recognition unit 16, and a learning unit 17.

[0143] The imaging device 4 is equipped with a control unit such as a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a RAM (Random Access Memory), and a ROM (Read Only Memory). The control unit issues execution commands to the processing unit, imaging unit, communication unit, input unit, output unit, and memory unit described below, and the processing unit calculates data and determines the calculation results.

[0144] The imaging device 4 is not particularly limited as long as it has a function required for capturing images such as moving images and / or still images and is capable of capturing color images.

[0145] The imaging device 4 includes a device as a communication unit for enabling communication with other terminals, devices, etc. The communication method may be wireless or wired.

[0146] The imaging device 4 is assumed to have, as an input unit, functions necessary for a user to operate the imaging device 4. As an example for implementing input, it is possible to have a liquid crystal display or the like that realizes a touch panel function. The present invention is not particularly limited in function depending on the input method.

[0147] The imaging device 4 is an output unit that has functions necessary for a user to operate the imaging device 4. A liquid crystal display or the like can be considered as an example of a means for realizing the output. The present invention is not particularly limited in function by the output method.

[0148] The imaging device 4 includes a storage unit for storing data, such as a hard disk, a semiconductor memory, a recording medium, a memory card, etc. The data may be stored in a cloud service, a database, etc.

[0149] The computer 3 and the imaging device 4 are connected to each other via a public line network or the like so as to enable data communication, and transmit and receive necessary data and information.

[0150] The above is the system configuration of system 2 for recognizing tire serial numbers.

[0151] [Tire serial number recognition processing] The tire serial number recognition process executed by the system 2 for recognizing the tire serial number is similar to that of the first embodiment described above, and is composed of steps S31 to S37 as shown in Fig. 3, which correspond to the above-mentioned steps S11 to S17. The only difference from the first embodiment described above is that the computer 3 acquires an image 100 capturing an overall image 110 of the tire serial number captured by the user using the imaging device 4, and therefore a description thereof will be omitted. The computer 3 may acquire the image captured by the imaging device 4 via a communication network using the communication unit of the computer 3, or may acquire the image by storing it in a storage unit using external media.

[0152] [Another aspect of the second embodiment] [Overview of System 2 for Recognizing Tire Serial Numbers] An overview of a system 2 for recognizing a tire serial number, which is another aspect of the second embodiment of the present invention, will be described with reference to Fig. 16. Fig. 16 is a diagram for explaining an overview of a system 2 for recognizing a tire serial number, which is one aspect of the present invention. Note that the same functions and configurations as those of the above-mentioned second embodiment are given the same reference numerals, and descriptions thereof will be omitted. This aspect differs from the second embodiment in that, like the above-mentioned first embodiment, the system 2 for recognizing a tire serial number is composed of a computer 3 and an imaging device 4.

[0153] The system 2 for recognizing the serial number of a tire is composed of a computer 3 and an imaging device 4, similar to another aspect of the first embodiment.

[0154] The computer 3 may be, for example, a computer such as a desktop personal computer, a notebook computer, or a server, a mobile terminal such as a smartphone or a tablet terminal, or a wearable terminal such as a head-mounted display such as smart glasses or a smart watch.

[0155] The computer 3 may be realized, for example, as one terminal device, or may be realized as a plurality of terminal devices, or may be realized as a virtual device such as a cloud computer.

[0156] The imaging device 4 is, for example, a camera that captures images such as moving images and / or still images, and is not particularly limited as long as it can capture color images.

[0157] The computer 3 and the imaging device 4 are connected to each other via a public line network or the like so as to enable data communication, and transmit and receive necessary data and information.

[0158] Next, an overview of the process executed by the system 2 for recognizing the serial number of a tire will be described. First, the system 2 acquires an image 200 of the entire tire (step S21). Specifically, the system 2 is similar to one aspect of the second embodiment described above, except that the computer 3 acquires an image 100 that includes an entire image 110 of the tire's serial number, which is captured by the user using the imaging device 4.

[0159] Next, the system 2 estimates an overall image 110 of the tire serial number in the acquired image 200 based on the trained model 300, and estimates two screw marks 120 in the overall image 110 (step S22). Specifically, this is the same as one aspect of the second embodiment described above.

[0160] Next, the system 2 determines the inclination from the coordinates of the upper screw mark 120 and the lower screw mark 120 of the two screw marks 120, and corrects the inclination (step S23). Specifically, this is the same as one aspect of the second embodiment described above.

[0161] Next, the system 2 estimates an image of each character in the corrected overall image 110 of the serial number based on the trained model 300 (step S24). Specifically, this is similar to one aspect of the second embodiment described above.

[0162] Next, the system 2 determines whether the overall image 110 of the serial number is inverted in the image of each of the estimated characters based on the trained model 300 (step S25). Specifically, this is the same as one aspect of the second embodiment described above.

[0163] Next, when it is determined that the entire image 110 of the serial number is inverted, the system 2 upturns the entire image 110 upside down (step S26). Specifically, this is the same as one aspect of the second embodiment described above.

[0164] Next, the system 2 detects the character area 140 between the two screw marks 120 and performs character recognition (step S27). Specifically, this is the same as one aspect of the second embodiment described above.

[0165] The above is an overview of the process executed by the terminal 1 to recognize the serial number of a tire.

[0166] [System configuration of System 2 for recognizing tire serial numbers] As shown in FIG. 17, the system configuration of a system 2 for recognizing tire serial numbers in another aspect of this embodiment is the same as that of the other aspect of the first embodiment, except that a control unit of a computer 3 cooperates with a processing unit to realize an acquisition unit 10, a third acquisition unit 21, a first estimation unit 22, a correction unit 12, a first estimation unit 22, a second judgment unit 24, a reversal unit 15, a recognition unit 16, and a learning unit 17, and therefore a description thereof will be omitted. This article explains:

[0167] [Tire serial number recognition processing] The tire serial number recognition process executed by the system 2 for recognizing the tire serial number is similar to that of the second embodiment described above, and is composed of steps S41 to S47 as shown in Fig. 12, which correspond to the above-mentioned steps S21 to S27. The only difference from the second embodiment described above is that the computer 3 acquires an image 100 showing an overall image 110 of the tire serial number photographed by the user using the imaging device 4, and therefore a description thereof will be omitted.

[0168] [Learning process] The learning process executed by the terminal 1 for recognizing the tire serial number is different from that of the second embodiment described above only in that the computer 3 acquires an image 100 capturing an overall image 110 of the tire serial number captured by the user using the imaging device 4, and therefore a description thereof will be omitted. The computer 3 may acquire the image captured by the imaging device 4 via a communication network using a communication unit of the computer 3, or may acquire the image by storing it in a memory unit using external media.

[0169] According to the system 2 of another aspect of the first and second embodiments, the system can be realized by a computer 3 such as a desktop personal computer, a notebook computer, a server, or the like, and an imaging device 4 such as a camera that captures images such as color video and / or still images. Therefore, by installing the imaging device 4 at a fixed location, the computer 3 can obtain an image including an overall image 110 of the tire serial number captured when a tire is placed on or passes by that location, and recognize the tire serial number, thereby saving the user the effort and time of having to go to a fixed location.

[0170] The above-mentioned means and functions are realized by a computer (including a CPU, an information processing device, and various terminals) reading and executing a predetermined program. The program is provided, for example, in a form provided from one or more terminals via a network (cloud service, SaaS: Software as a Service). The program is also provided, for example, in a form recorded on a computer-readable recording medium. In this case, the computer reads the program from the recording medium, transfers it to an internal recording device or an external recording device, records it, and executes it. The program may also be recorded in advance on a recording device (recording medium) such as a magnetic disk, optical disk, or magneto-optical disk, and provided to the terminal from the recording device via a communication line.

[0171] Although the embodiments of the present invention have been described above, the present invention is not limited to these embodiments. Furthermore, the effects described in the embodiments of the present invention are merely a list of the most preferable effects resulting from the present invention, and the effects of the present invention are not limited to those described in the embodiments of the present invention. [Explanation of symbols]

[0172] REFERENCE SIGNS LIST 1 terminal, 2 system, 3 computer, 4 imaging device, 10 acquisition unit, 11 first detection unit, 12 correction unit, 13 second detection unit, 14 first judgment unit, 15 inversion unit, 16 recognition unit, 17 learning unit, 21 third acquisition unit, 22 first estimation unit, 23 second estimation unit, 24 second judgment unit, 100 200 image, 110 overall image of serial number, 120 screw marks, 130 characters, 300 trained model

Claims

1. A terminal for recognizing a serial number of a tire, A first acquisition unit that acquires an image including an entire image of the serial number displayed on the tire; a first detection unit that detects an entire image of the serial number in the acquired image and detects two screw marks in the entire image; a correction unit that determines an inclination from coordinates of an upper screw mark and a lower screw mark of the two screw marks and corrects the inclination; a second detector for detecting each character within an overall image of the corrected serial number; a first determination unit that determines whether an entire image of the serial number is inverted based on each of the detected characters; an inversion unit that inverts the entire image of the serial number upside down when it is determined that the entire image of the serial number is inverted; a recognition unit that detects a character area between the two screw marks and recognizes the characters; A terminal comprising:

2. The terminal according to claim 1 , wherein the first determination unit determines that the entire image of the serial number is an inverted image based on the detected images of each character and a line connecting the two screw marks.

3. The terminal of claim 1, further comprising a learning unit that learns an overall image of the serial number in the correct position, an overall image of the serial number at an angle, the two screw marks, and images of each character to create a learned model.

4. The terminal according to claim 3 , wherein the learning unit tags and learns the inverted image of each character.

5. A second acquisition unit that acquires an image of the entire tire; A first estimation unit that estimates an overall image of the serial number of the tire in the acquired image based on the trained model and estimates two screw marks in the overall image; A second estimation unit that estimates an image of each character in an overall image of the serial number based on the trained model; a second determination unit that determines that the entire image of the serial number is an inverted image when an inverted character is present in the image of each of the characters estimated based on the trained model; The terminal of claim 4 further comprising:

6. A system for recognizing serial numbers of tires, comprising: A first acquisition unit that acquires an image including an entire image of the serial number displayed on the tire; a first detection unit that detects an entire image of the serial number in the acquired image and detects two screw marks in the entire image; a correction unit that determines an inclination from coordinates of an upper screw mark and a lower screw mark of the two screw marks and corrects the inclination; a second detector for detecting each character within an overall image of the corrected serial number; a first determination unit that determines whether an entire image of the serial number is inverted based on each of the detected characters; an inversion unit that inverts the entire image of the serial number upside down when it is determined that the entire image of the serial number is inverted; a recognition unit that detects a character area between the two screw marks and recognizes the characters; A system comprising:

7. 1. A method implemented by a system for recognizing serial numbers of tires, comprising: obtaining an image including a full view of the serial number displayed on the tire; Detecting an overall image of the serial number in the acquired image, and detecting two screw marks in the overall image; determining a tilt from coordinates of an upper screw mark and a lower screw mark of the two screw marks, and correcting the tilt; Detecting each character within the corrected image of the serial number; determining whether the serial number is inverted based on each of the detected characters; if it is determined that the entire image of the serial number is inverted, flipping the entire image upside down; detecting a character area between the two screw marks and performing character recognition; A method for providing the above.

8. On the computer, obtaining an image including a complete view of the serial number displayed on the tire; detecting an overall image of the serial number in the acquired image, and detecting two screw marks in the overall image; determining a tilt from coordinates of an upper screw mark and a lower screw mark of the two screw marks, and correcting the tilt; detecting each character within the corrected image of the serial number; determining whether the serial number is inverted based on each of the detected characters; if it is determined that the entire image of the serial number is inverted, flipping the entire image upside down; detecting a character area between the two screw marks and performing character recognition; A computer-readable program that causes a computer to execute

Images