Tire damage evaluation device, tire damage evaluation method, and tire damage evaluation program

Abstract

This tire damage evaluation device comprises a processing unit that acquires a tread image obtained by imaging a tread section of a tire, and quantifies the damage state of a surface region of the tread section on the basis of the tread image.

Description

Tire damage assessment device, tire damage assessment method, and tire damage assessment program

[0001] The present disclosure relates to a tire damage assessment device, a tire damage assessment method, and a tire damage assessment program.

[0002] Conventionally, the appearance and damage of tires have been visually inspected by humans, and evaluation of tire surface conditions from tire images has been adopted in tire appearance inspections, for example (see, for example, Patent Document 1).

[0003] JP 2021-42988 A

[0004] However, inspection methods involving human visual inspection of tire appearance are heavily influenced by experience and know-how, lacking objectivity and remaining limited to qualitative evaluation. Furthermore, conventional image analysis methods only detect whether a tire is good or bad on the tire production line, and are unable to quantitatively evaluate the surface condition of the tire tread after it has been in use.

[0005] Therefore, the purpose of the present disclosure, which has been made with these points in mind, is to provide a tire damage evaluation device, a tire damage evaluation method, and a tire damage evaluation program that can quantify and evaluate the surface condition of the tread portion of a tire.

[0006] (1) A tire damage assessment device according to one aspect of the present disclosure includes a processing unit that acquires a tread image of a tire tread portion and quantifies the damage state of a surface region of the tread portion based on the tread image.

[0007] (2) A tire damage assessment device according to one embodiment of the present disclosure is the tire damage assessment device described in (1) above, wherein the tread portion is configured to include a plurality of layers arranged in the radial direction of the tire, and the processing unit preferably generates a segmented image in which the surface area of ​​the tread portion included in the tread image is segmented by the layers exposed in the surface area.

[0008] (3) A tire damage assessment device according to one embodiment of the present disclosure is the tire damage assessment device described in (2) above, and it is preferable that the multiple layers include an outer layer rubber that forms the tread surface of the tread portion, a layer of metal cords arranged radially inward of the outer layer rubber, and an inner layer rubber located between the outer layer rubber and the layer of metal cords.

[0009] (4) A tire damage assessment device according to one embodiment of the present disclosure is the tire damage assessment device described in (2) or (3) above, and it is preferable that the processing unit inputs tread images of a plurality of sample tires and executes a process of inputting the tread images of the tire to a first trained model trained using first training data that outputs segmented images dividing the surface area of ​​the sample tires included in the tread images by the layers exposed on the surface, and then outputs the segmented images of the tire.

[0010] (5) A tire damage assessment device according to one embodiment of the present disclosure is the tire damage assessment device described in any of (2) to (4) above, wherein the processing unit preferably divides the segmented image of the tire into a plurality of strip-shaped regions extending in the width direction of the tire, aligns the positions of the plurality of strip-shaped regions so that the widthwise positions of the tread portion within the strip-shaped regions coincide, and generates a histogram by accumulating the pixels of the plurality of strip-shaped regions for at least any of the layers exposed to the surface region for each position in the width direction.

[0011] (6) A tire damage assessment device according to one embodiment of the present disclosure is the tire damage assessment device described in any one of (1) to (5) above, and the processing unit is preferably configured to extract from the tread image at least any of a groove portion indicating a groove in the tread portion, a local damage portion indicating localized damage to the surface region of the tread portion, and a repaired portion indicating a location where past damage to the tire has been repaired.

[0012] (7) A tire damage assessment device according to one embodiment of the present disclosure is the tire damage assessment device described in any one of (1) to (6) above, and it is preferable that the processing unit inputs tread images of a plurality of sample tires and outputs images of local damage areas of the sample tires included in the tread images into a second trained model trained using second training data, the second training model outputting the images of the local damage areas of the tire.

[0013] (8) A tire damage assessment device according to one embodiment of the present disclosure is the tire damage assessment device described in (7) above, and it is preferable that the processing unit calculates the ratio of the area of ​​the localized damage portion to the area of ​​the surface region of the tread portion.

[0014] (9) A tire damage assessment device according to one embodiment of the present disclosure is the tire damage assessment device described in (7) or (8) above, and it is preferable that the processing unit calculates the center of gravity position of the area of ​​the localized damage portion.

[0015] (10) A tire damage assessment device according to one embodiment of the present disclosure is a tire damage assessment device according to any one of (2) to (5) above, wherein the processing unit may quantify, based on the tread image, the damage state of the surface region of the tire at a portion where the outermost layer of the plurality of layers of the tire is exposed.

[0016] (11) A tire damage assessment device according to one embodiment of the present disclosure is the tire damage assessment device described in (10) above, wherein the processing unit preferably extracts an image of an area encompassing the surface area of ​​the tread portion from the tread image, divides the image into images of multiple sub-areas, and quantifies the damage state of the surface area for each of the images of the sub-areas.

[0017] (12) A tire damage assessment device according to one embodiment of the present disclosure is the tire damage assessment device described in (10) or (11) above, and it is preferable that the processing unit inputs tread images of a plurality of sample tires to a third trained model trained using third training data that outputs information quantifying the damage state of the surface region of a portion where the outermost layer of the tire is exposed among the plurality of layers included in the tread image, and inputs the tread images of the tire to output information quantifying the damage state of the surface region.

[0018] (13) A tire damage assessment device according to one embodiment of the present disclosure is the tire damage assessment device described in (12) above, and it is preferable that the damage state of the surface region in the third training data is quantified based on at least one of the number of scratches contained in the surface region and the depth of the scratches.

[0019] (14) A tire damage assessment device according to one embodiment of the present disclosure is a tire damage assessment device described in any of (1) to (13) above, wherein the damage state includes the state of chips and scratches on the tire, and the processing unit preferably estimates the wear characteristics of the tire based on information on the quantified state of the chips and scratches.

[0020] (15) A tire damage assessment device according to one embodiment of the present disclosure is a tire damage assessment device according to any one of (1) to (14) above, and preferably includes at least one of a display unit that displays quantified information on the damage state of the surface region of the tread portion, and a communication unit that transmits the information to an external device.

[0021] (16) A tire damage assessment method according to one aspect of the present disclosure is a tire damage assessment method executed by an information processing device, and includes acquiring a tread image of a tire tread portion and quantifying a damage state of a surface region of the tread portion based on the tread image.

[0022] (17) A tire damage evaluation program according to one embodiment of the present disclosure causes a computer to perform a process of acquiring a tread image of a tire tread portion, and a process of quantifying the damage state of the surface region of the tread portion based on the tread image.

[0023] According to the present disclosure, it is possible to quantify and evaluate the damage state of the surface region of the tread portion of a tire.

[0024] 10 is a block diagram showing a schematic configuration of a tire damage assessment system including a tire damage assessment device according to an embodiment of the present disclosure. FIG. 11 is a cross-sectional view showing an example of a tire to be assessed by the tire damage assessment system of FIG. 1. FIG. 12 is a flow diagram showing a method for generating a first trained model that segments the surface region of the tread portion based on a plurality of sample tire images. FIG. 13 is a diagram showing an example of an image of the tread portion of a sample tire. FIG. 14 is a diagram showing an example of a segmented image generated from the image of the tread portion. FIG. 15 is a diagram explaining a method for generating a histogram from the segmented image. FIG. 16 is a diagram showing an example of a groove portion, a localized damaged portion, and a repaired portion of a tire. FIG. 17 is a flow diagram showing a method for generating a third trained model that quantifies the damage level of the surface region of the tread portion based on a plurality of sample tire images. FIG. 18 is a diagram showing an example of a tire tread image captured by an imaging unit. FIG. 19 is a diagram showing an example of a process for determining a rectangular region executed on the tread image of FIG. 9. FIG. 11 is a diagram showing an example of masking the damaged region in the rectangular region image of FIG. 10 and segmenting it into sub-regions. FIG. 12 is a diagram showing an image in which a numerical value quantifying the damage state is assigned to each sub-region of the tread image of FIG. 11. FIG. 13 is a flow diagram explaining a tire damage assessment method according to an embodiment of the present disclosure. FIG. 14 is a block diagram showing a schematic configuration of a tire damage assessment system according to another embodiment of the present disclosure.

[0025] Tires used in vehicles operating in special environments, such as mines, are susceptible to damage, including chipping and scratches, on the tread surface when driven on roads other than paved roads. Small damage to tires can lead to rapid tire wear. By quantitatively understanding the relationship between tire damage and the progression of tire wear, it is expected that the relationship between tire damage and tread pattern can be clarified and utilized in tire design. Furthermore, it is expected that the appropriate timing for tire rotation or replacement can be estimated based on the state of tire damage, and information can be provided to tire users.

[0026] In this application, "damage" means that a normal state is at least partially lost. "Damage" to a tire surface includes wear, chips, scratches, etc. on the tire surface. Wear on a tire surface means that the tire surface has been worn down and scraped away due to friction with the ground. Chips on a tire surface mean that part of the tire surface is missing. Scratches on a tire surface mean that the tire has scratches, cuts, etc. In this application, "localized damage" means chips, gouges, scratches, etc. that occur in a relatively small area, and whose location and range can be identified by an operator from an image of the tire surface.

[0027] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings.

[0028] (Tire Damage Assessment System) As shown in FIG. 1 , a tire damage assessment system 1 according to an embodiment of the present disclosure includes a tire damage assessment device 10 and an imaging device 20. FIG. 2 shows a cross-sectional view of an example of a tire 30 to be assessed by the tire damage assessment system 1. In the tire damage assessment system 1, the tire damage assessment device 10 analyzes and quantifies the state of damage in the surface region of the tread portion 35 based on an image of the tread portion 35 of the tire 30 captured by the imaging device 20. The tread portion 35 refers to the portion of the tire 30 that directly contacts the road. Hereinafter, the captured image of the tread portion 35 will be referred to as a tread image.

[0029] The tire damage assessment device 10 is an information processing device, such as a server computer, a workstation, or a PC (Personal Computer). The tire damage assessment device 10 may be installed at a user's site or at a remote location via a network. The components and functions of the tire damage assessment device 10 described below may be distributed across multiple different pieces of hardware. The tire damage assessment device 10 may also be a small information processing device such as a portable information terminal or a smartphone. In the embodiment described below, the tire damage assessment device 10 is described as a stationary computer. The tire damage assessment device 10 acquires tread images from the imaging device 20 and performs processing including image processing.

[0030] The imaging device 20 includes an optical system and an imaging element. The imaging element includes a charge-coupled device (CCD) image sensor (charge-coupled device image sensor) and a complementary metal-oxide semiconductor (CMOS) image sensor. The imaging element converts an image formed on a light-receiving surface by the optical system into an electrical signal. In the following embodiment, the imaging device 20 includes a single set of optical systems and imaging elements, and images the tread portion 35 of the tire 30 from one direction while the tire 30 is removed from the vehicle and laid flat. However, the imaging device 20 may include multiple sets of optical systems and imaging elements and be configured to image the tread portion 35 from multiple directions. The imaging device 20 may be configured to image the tire 30 from all directions, 360 degrees, along the circumferential direction of the tire 30. The imaging device 20 may be a digital camera or a portable information terminal equipped with a camera. The imaging device 20 may also be a device dedicated to the tire damage assessment system 1.

[0031] (Configuration of Tire Damage Assessment Device) In one embodiment, the tire damage assessment device 10 includes a communication unit 11 , an input unit 12 , an output unit 13 , a processing unit 14 , and a storage unit 15 .

[0032] The communication unit 11 provides a communication function for communicating with devices external to the tire damage assessment device 10. The communication unit 11 includes a communication module that connects to a communication network such as the Internet. The communication unit 11 may include a communication module that supports mobile communication standards such as 4G (4th Generation) and 5G (5th Generation). The communication unit 11 may include a communication module that supports a wired LAN standard (1000BASE-T, for example). The communication unit 11 may include a communication module that supports a wireless LAN standard (IEEE802.11, for example). The communication unit 11 may include a communication module that supports short-range communication such as Bluetooth and BLE.

[0033] The communication unit 11 may be configured to be connected to the imaging device 20 via wired or wireless communication and to receive tread images captured by the imaging device 20. The communication unit 11 may be configured to be able to send and receive information to and from an information processing device for machine learning (described later) via a network such as the Internet.

[0034] The input unit 12 includes one or more input interfaces for the tire damage assessment device 10. The input interfaces include, for example, physical keys, capacitance keys, a pointing device, a touch screen integrated with a display, a microphone, and a camera. The input unit 12 includes an input interface that acquires information from a portable storage medium such as a memory card. The tire damage assessment device 10 may acquire tread images captured by the imaging device 20 via the storage medium.

[0035] The output unit 13 includes one or more output interfaces. Examples of the output interfaces include a display, a printer, and a speaker. Examples of the display that can be used include an LCD (Liquid Crystal Display) or an organic EL (Electro Luminescence) display. In other words, the output unit 13 includes a display unit. The input unit 12 may include an output interface that writes information to a portable storage medium such as a memory card. The output unit 13 can output information obtained by the operation of the tire damage assessment device 10 to a user.

[0036] The processing unit 14 includes one or more processors. The processor may be a general-purpose processor such as a CPU (Central Processing Unit) that executes a program that defines a control procedure, or may be a dedicated processor specialized for a specific process. The processing unit 14 performs arithmetic processing to execute each function of the tire damage assessment device 10 and controls each component of the tire damage assessment device 10.

[0037] The processing unit 14 may be able to perform various types of image processing on the tread image acquired from the imaging device 20. The processing unit 14 can evaluate the damage state of the surface region of the tread portion 35 by using at least one of first to third trained models stored in the storage unit 15, which will be described later.

[0038] The storage unit 15 includes one or more memories. The memories may be, for example, semiconductor memory, magnetic memory, or optical memory, but are not limited to these, and may be any type of memory. The semiconductor memories may include volatile memories such as RAM (Random Access Memory) and non-volatile memories such as ROM (Read Only Memory) and flash memory. The magnetic memories may include, for example, hard disks and magnetic tapes. The optical memories may include, for example, CDs (Compact Discs), DVDs (Digital Versatile Discs), and BDs (Blu-ray (registered trademark) Discs).

[0039] The storage unit 15 stores information necessary for the processing unit 14 of the tire damage assessment device 10 to perform processing, information generated as a result of processing by the processing unit 14, etc. For example, the storage unit 15 stores tread images captured by the imaging device 20, images obtained by processing the tread images, and information resulting from an assessment of the damage state of the surface region of the tire 30 from the tread images. The storage unit 15 may store a tire damage assessment program executed by the processing unit 14 of the tire damage assessment device 10, and first to third trained models. The tire damage assessment program may be provided in a form stored on a non-transitory storage medium such as a CD, a DVD, or a USB (Universal Serial Bus) memory.

[0040] (Example of the Structure of a Tire to be Evaluated) An example of a tire 30 to be evaluated by the tire damage evaluation device 10 will be described with reference to the cross-sectional view of Figure 2. In one embodiment, the tire 30 is a tire mounted on a vehicle used in special environments such as mines, and a special vehicle for work and construction. When used in harsh environments, the tire 30 is susceptible to various types of damage. Furthermore, the tire 30 is often used to the end of its life despite sustaining damage.

[0041] 2 has a pair of bead portions 31 and a pair of sidewall portions 32, a tread portion 35 connected to both sidewall portions 32 and having an outer layer rubber 33 and an inner layer rubber 34 in that order from the outside in the tire radial direction, a carcass 36 extending in a toroidal shape between the pair of bead portions 31 to reinforce the bead portions 31, sidewall portions 32, and tread portion 35, and a metal cord layer 37 arranged radially outward of a crown portion of the carcass 36 and radially inward of the inner layer rubber 34. That is, the outer layer rubber 33, inner layer rubber 34, and metal cord layer 37 are arranged in this order from the outside in the radial direction of the tire 30.

[0042] Here, the outer layer rubber 33 of the tire 30 shown in Figure 2 is also generally called "cap rubber." The inner layer rubber 34 is also generally called "base rubber." Furthermore, the metal cord layer 37 is also generally called "belt."

[0043] The carcass 36 of the tire 30 shown in Figure 2 is composed of one carcass ply made of a plurality of parallel-arranged cords covered with a coating rubber. The carcass 36 also comprises a main body portion extending in a toroidal shape between the bead cores 38 embedded in the bead portions 31, and turn-up portions wound up radially outward from the inner side toward the outer side in the tire width direction around each bead core 38. The number and structure of the plies of the carcass 36 are not limited to this.

[0044] The metal cord layer (belt) 37 of the tire 30 shown in Fig. 2 is composed of two belt layers. The number of belt layers constituting the metal cord layer 37 is not limited to this, and the number of belt layers may be three or more. Here, the belt layer is usually composed of a rubberized layer of metal cords (preferably steel cords) extending at an angle with respect to the tire equatorial plane. The two belt layers are laminated so that the metal cords constituting the belt layers cross each other with the tire equatorial plane in between, thereby constituting the metal cord layer 37.

[0045] The tire 30 includes a tread portion 35 including an outer layer rubber 33 that forms the tread surface of the tread portion 35, a metal cord layer 37 disposed radially inward of the outer layer rubber 33, and an inner layer rubber 34 positioned between the outer layer rubber 33 and the metal cord layer 37. A tread pattern is formed on the surface of the outer layer rubber 33. The rubber layer of the tread portion 35 of the tire 30 is not limited to a two-layer structure consisting of the outer layer rubber 33 and the inner layer rubber 34, but may have a single-layer structure or a layer structure of three or more layers. The tire damage assessment device 10 of the present disclosure can be suitably applied to assessing damage in a tire 30 having a tread portion 35 with such a configuration.

[0046] (Method for Segmenting Tire Surface Region) The processing unit 14 of the tire damage assessment device 10 acquires a tread image including an image of the tread portion 35 of the tire 30 captured by the imaging device 20 via the communication unit 11 or the input unit 12. The processing unit 14 generates a segmented image in which the surface region of the tread portion 35 included in the acquired tread image is segmented by the layers exposed in the surface region. The processing unit 14 may segment the surface region into a plurality of regions including a first region 41 covered by the outer rubber layer 33, a second region 42 in which the inner rubber layer 34 is exposed, and a third region 43 in which the metal cord layer 37 is exposed (see FIGS. 4 and 5 ).

[0047] Various image processing methods can be used to generate the segment image. In one embodiment, machine learning can be used to generate the segment image. The processing unit 14 may be configured to input the tread image to a first trained model that has undergone machine learning in an information processing device for machine learning, and output a segment image of the tire 30. The information processing device for machine learning may be the same computer as the tire damage assessment device 10, or may be a different computer. In the latter case, the information processing device for machine learning may be configured to be able to communicate with the tire damage assessment device 10.

[0048] A method for generating a first trained model will be described based on the flow diagram of Figure 3.

[0049] First, an operator takes tread images of multiple tires (step S101). The multiple tires that have been photographed are called sample tires. Sample tires may include tires that are still in use and tires that have reached the end of their lifespan. Many images of sample tires under different conditions are taken and stored.

[0050] The surface area of ​​the tread portion 35 of the captured tread image of each sample tire is divided by the layer exposed on the surface area on the screen of a work computer (step S102). The worker may be a specialized engineer who can determine whether the exposed area is the outer rubber layer 33, the inner rubber layer 34, or the metal cord layer 37 by looking at the image of the sample tire.

[0051] Here, the fact that a layer other than the outermost layer of the tire 30 is exposed on the tire surface means that the upper layer has been worn away due to vehicle travel. Therefore, for example, the area where the inner rubber layer 34 or the metal cord layer 37 is exposed is included in the damaged area. It can be said that the area where the metal cord layer 37 is exposed is more severely damaged than the area where the inner rubber layer 34 is exposed.

[0052] FIG. 4 shows an example of a tread image 40 of a sample tire and how an operator segments the surface area on the tread image 40. As shown in FIG. 4 , the operator uses a GUI (Graphical User Interface) on a computer to sequentially specify boundaries where different layers are exposed on an image of the tread portion 35 of the sample tire. Examples of boundaries where different layers are exposed include the boundary between a first region 41 where the outer rubber layer 33 is located on the outermost surface and a second region 42 where the inner rubber layer 34 is exposed, and the boundary between the second region 42 and a third region 43 where the metal cord layer 37 is exposed. The operator specifies boundaries to surround damaged regions, such as the second region 42 and the third region 43, respectively, to generate a segmented image in which the surface area of ​​the tread image 40 is segmented. In this manner, the first region 41, the second region 42, and the third region 43 are segmented on the tread image 40 of the sample tire. The work computer may be any computer, and may be the same computer as the tire damage assessment device 10 and / or the information processing device for machine learning.

[0053] Next, first training data for machine learning is generated (step S103), with the tread image 40 of the sample tire as input and data of segmented images obtained by segmenting the surface region of the sample tire included in the tread image 40 as output. The first training data includes a large amount of data generated from a large number of sample tires.

[0054] An information processing device for machine learning performs machine learning using the first training data generated in step S103 (step S104). The information processing device for machine learning includes a function approximator capable of learning input / output relationships, and generates a first trained model by learning the first training data (step S105). The first trained model is also referred to as a region segmentation model. The function approximator can perform machine learning using a known machine learning algorithm, such as a neural network or deep learning. When a tread image 40 of a tire 30 is input, the first trained model outputs a segmented image that segments the surface region of the tread portion 35 of the tire 30. By learning data on tread images 40 of more sample tires, the accuracy of machine learning can be improved, and the accuracy of segmenting the tire surface region can be improved.

[0055] The first trained model generated according to the flow diagram of Fig. 3 is stored in the memory unit 15 of the tire damage assessment device 10. The processing unit 14 inputs the tread image 40 of the tire 30 captured by the imaging device 20 into the first trained model read from the memory unit 15, and generates a segmented image by outputting a segmented image that segments the surface area of ​​the tread portion 35 of the tire 30. The segmented image may be displayed on the display, which is the output unit 13. This visualizes the state of damage caused by wear in the tread portion 35 of the tire 30.

[0056] FIG. 5 shows an example of a segmented image generated from an image of the tread portion 35 by segmenting the surface region. In FIG. 5, a first region 41 in which the outer rubber layer 33 is exposed in the surface region, a second region 42 in which the inner rubber layer 34 is exposed, and a third region 43 in which the metal cord layer 37 is exposed are shown. Each region may be displayed in a different color. The size of each region allows the user to visually grasp the degree of damage to the surface region of the tire 30 quantitatively to some extent. The processing unit 14 may digitize the area ratio, which is the ratio of the area of ​​each of the first region 41, the second region 42, and the third region 43 to the entire surface region of the tread portion 35, as a ratio of pixels in the image.

[0057] The processing unit 14 can further process the segmented image to grasp the positions in the width direction of the tire 30 where the inner rubber layer 34 and the metal cord layer 37 are largely exposed. For this reason, as shown in Fig. 6, the processing unit 14 divides the segmented image of the tire 30 into a plurality of rectangular regions 45 that are narrow in the radial direction of the tire 30 and extend long in the width direction of the tire 30. The term "rectangular" can be rephrased as "band-like" or "strip-like."

[0058] The processing unit 14 aligns the positions of the multiple strip-shaped regions 45 so that the positions of the tread portions 35 in the strip-shaped regions 45 in the width direction of the tire 30 are consistent. Figure 6 shows the strip-shaped regions 45 before alignment. The tread image 40 of the tire 30 may be outwardly displaced from the center of the image depending on the angle at which it is captured. Also, the width of the tire 30 may appear narrower on the outside than in the center of the image. The processing unit 14 aligns the center positions of the strip-shaped regions 45 and aligns the lengths of the strip-shaped regions 45 in the width direction of the tire 30, i.e., the vertical direction in Figure 6. This operation aligns the positions of the strip-shaped regions 45 in the width direction of the tire 30. This operation is called normalization.

[0059] Next, the processing unit 14 generates a histogram by integrating pixels corresponding to one or more of the first region 41, the second region 42, and the third region 43 for each position in the width direction of the tire 30 on the image formed by arranging the normalized strip-shaped regions 45. In the example of FIG. 6 , the histograms for the second region 42 and the third region 43 are shown on the right side. This allows the user to grasp the positions and sizes of the most exposed areas of the inner rubber layer 34 and the metal cord layer 37 in the width direction of the tire 30. In other words, the user can grasp the positions and sizes of areas of the surface region of the tread portion 35 that are extensively damaged due to wear or the like.

[0060] (Extraction of Locally Damaged Areas on Tire Surface) In the above explanation using Figures 3 to 6, the tire damage assessment device 10 generates segmented images that segment the exposed areas of the outer rubber layer 33, inner rubber layer 34, and metal cord layer 37 on the surface region of the tire 30. In addition to these, the tire damage assessment device 10 may be configured to extract groove portions 51 that indicate tread grooves, local damaged areas 52 that include local damage such as chips, gouges, and cuts, and repaired areas 53 that indicate areas that have been repaired in the past, as shown in Figure 7. Figure 7 shows an enlarged view of the local damaged area 52 included in a small region 54.

[0061] The groove portion 51, the locally damaged portion 52, and the repaired portion 53 may be extracted using a trained model generated by machine learning similar to that described using the flow diagram of Fig. 3. In particular, even a small locally damaged portion 52 of the tire 30 affects the progress of wear of the tire 30.

[0062] The procedure for extracting the local damaged area 52 will be described. Similar to the flow diagram of FIG. 3 , an operator extracts the local damaged area 52 on the tread images 40 of multiple sample tires and records the area on the tread image 40. For example, the operator performs a process of filling in chips on the tread image 40. This makes it possible to generate second training data in which the tread image 40 is input and the area of ​​the local damaged area 52 on the tread image 40 is output. A second trained model can be generated by performing machine learning using the second training data on an information processing device for machine learning. The tire damage assessment device 10 can store the second trained model in the memory unit 15. The processing unit 14 can execute a process of inputting the tread image 40 of the tire 30 into the second trained model read from the memory unit 15 and outputting the local damaged area 52 of the tire 30. Increasing the number of sample tires used to generate the second training data can improve the accuracy of extracting the local damaged area 52.

[0063] When the processing unit 14 extracts the locally damaged areas 52, it calculates the center of gravity (G ｘ , G ｙ) to calculate the area ratio. This allows the processing unit 14 to present the location and area of ​​the local damaged area 52 to the user. The processing unit 14 may further calculate an area ratio, which is the ratio of the total area of ​​the local damaged area 52 to the area of ​​the entire surface area of ​​the tread portion 35. This allows the processing unit 14 to quantify the size of the local damaged area 52.

[0064] In the above description, the second trained model is configured to input the tread image 40 and output information about the local damaged portion 52. The second trained model may be configured to output the groove portion 51 and / or the repaired portion 53 in addition to or instead of the local damaged portion 52. Information about the groove portion 51 and the repaired portion 53 may be used as appropriate to evaluate the damage state of the tire.

[0065] The first trained model and the second trained model may be generated as a single integrated trained model rather than as separate trained models. When a user inputs a tread image 40 into the tire damage assessment device 10, the processing unit 14 may be configured to generate a segmented image and extract minute local damage areas 52 in the first region 41 on the segmented image. By continuously monitoring the local damage areas 52, the user can monitor the process by which minute local damage areas 52 lead to significant wear of the tire 30. The user can observe the relationship between the shape of the tread pattern of the tread portion 35 and the progress of wear of the tire 30, and use the observation results in designing the tread pattern of the tire 30.

[0066] (Detection of Minor Damage Levels) The surface of the tire 30 may have many small scratches that are difficult to identify individually, even in areas where the inner rubber layer 34 or metal cord layer 37 is not exposed, i.e., areas of the outer rubber layer 33. In one embodiment, the tire damage assessment device 10 can quantify relatively minor damage, such as scratches, in the tread portion 35 of the tire 30. The damage state of the surface region of the tire 30 can be determined by analyzing the tread image 40 through image processing. In one embodiment, machine learning can be used to determine the damage state.

[0067] 8 is a flow diagram illustrating a method for generating a third trained model for evaluating the damage state of the tire 30 by machine learning based on the tread image 40. The processing of the flow diagram in FIG. 8 may be executed by an information processing device for machine learning. This information processing device for machine learning may be the same computer as or a different computer from the information processing device for machine learning used to generate the first trained model and / or the second trained model.

[0068] First, the information processing device for machine learning acquires tread images 40 of a plurality of sample tires captured by an operator (step S201). Fig. 9 is a diagram showing the acquired tread images 40 of the sample tires. It is desirable that the tread images of the plurality of sample tires are captured at similar angles and magnifications so as to include the entire tire.

[0069] Next, a rectangular region 55 including the tread portion 35 is extracted from the tread image 40 of the sample tire (step S202). The rectangular region 55 may be extracted as the smallest rectangle that encompasses the tread portion 35 in the image of the sample tire. In one embodiment, to identify the tread portion 35 of the sample tire, the tread image 40 may be input to a first trained model, which may output a segmented image. By using the first trained model, the tread portion 35 including a first region 41, a second region 42, and a third region 43 (not shown in FIG. 10 ) is extracted from the tread image 40 of the sample tire, as shown in FIG. 10 . In the example of FIG. 10 , the surface region of the sample tire is mainly the first region 41 where the outer rubber layer 33 is exposed, and a portion of the surface region is the second region 42 where the inner rubber layer 34 is exposed. The rectangular region 55 is determined to encompass the first region 41.

[0070] Next, the machine learning information processing device divides the extracted rectangular region 55 vertically and horizontally into n x m subregions 57. For example, in the example of FIG. 11 , the partitioned region is divided into eight 2 x 4 subregions 57. n and m can be set to any natural number. In one embodiment, the machine learning information processing device can apply a mask 56 to exclude regions of the rectangular region 55 that have significant damage from subsequent processing. The regions of significant damage include, for example, the second region 42 and the third region 43, which are determined to expose the inner rubber layer 34 or the metal cord layer 37 when analyzed using the first trained model. In another embodiment, the regions of significant damage may include regions from which localized damage 52 is extracted when analyzed using the second trained model.

[0071] In this way, for each divided subregion 57, the worker determines the damage level according to predetermined criteria and quantifies it (step S204). The damage level may be determined on a five-level scale from 0 to 4, as shown in FIG. 12 . For example, damage level 0 indicates slight damage, and damage level 4 indicates severe damage. The predetermined criteria may include criteria related to the number of scratches within the subregion 57 or per unit area, and the depth of the scratches. Because the surface region of the tread portion 35 of the sample tire has many minute scratches, the worker does not extract individual scratches, unlike the method for generating the first and second trained models. The machine learning information processing device acquires data that quantifies the damage level determined by the worker.

[0072] The third training data is generated by creating and accumulating a large amount of data in which the tread image 40 is input and the damage level of each sub-region 57 of the tread image 40 is quantified as output (step S205).

[0073] The machine learning information processing device performs machine learning using the third training data generated in step S205 (step S206). The machine learning information processing device includes a function approximator capable of learning input / output relationships, and generates a third trained model by learning the third training data (step S207). The third trained model can also be called a flaw prediction model. When a tread image 40 of a tire 30 is input, the third trained model outputs a numerical value indicating the damage state of each subregion 57 obtained by dividing the surface region of the tread portion 35 of the tire 30. This makes it possible to objectively evaluate the damage state of the surface region of the tire 30, even for scratches and other damage that are difficult to determine by visual inspection of the surface region. Increasing the number of sample tires used to generate the third training data can improve the accuracy of evaluating local conditions.

[0074] The third trained model generated according to the flow chart of FIG. 8 is stored in the memory unit 15 of the tire damage assessment device 10. The processing unit 14 inputs the tread image 40 of the tire 30 captured by the imaging device 20 into the third trained model and can output information quantifying the damage state of each subregion 57 into which the surface region of the tread portion 35 is divided. In this way, the tire damage assessment device 10 can evaluate the damage state of the surface region according to the position of the subregion 57 into which the surface region of the tread portion 35 is divided. The processing unit 14 may perform this process on the first region 41 portion where the outer rubber layer 33, which is the outermost layer of the tread portion 35 of the tire 30, is exposed. Note that the quantified information on the damage state output by the third trained model can be output as a numerical value more precise than the damage level determined by the operator in step S204. For example, if the operator determines the damage state using natural numbers from 0 to 4 in step S204, the quantified information on the damage state output by the third trained model may include decimals between 0 and 4. The processing unit 14 may convert the information quantifying the damage state, which includes decimals, into integer values ​​ranging from 1 to 100, for example.

[0075] (Tire Damage Assessment Process) Next, an example of a series of processes of a tire damage assessment method executed by the processing unit 14 of the tire damage assessment device 10 that has acquired an image of the tire 30 will be described with reference to the flowchart of Fig. 13. The tire damage assessment device 10 may execute all of the processes in the flowchart of Fig. 13, or may execute only some of the processes.

[0076] First, the processing unit 14 acquires a tread image 40, which is an image of the tread portion 35 of the tire 30, via the communication unit 11 or the input unit 12. The processing unit 14 performs preprocessing on the tread image 40, which is necessary for the following processing (step S301). The preprocessing includes processing to standardize the size of the acquired image. Therefore, in the preprocessing, the processing unit 14 may compress or expand the image.

[0077] Next, the processing unit 14 inputs the tread image 40 into the first trained model and the second trained model, and segments the tread image 40 into segment images including a first region 41 to a third region 43 (step S302). The segmentation in step S302 includes extracting the local damaged area 52. As a result, the processing unit 14 can generate a segment image as shown in FIG. 5 and extract the local damaged area 52 as shown in FIG. 7. The processing unit 14 can color-code each area of ​​the segment image and display it on a display, which is a display unit included in the output unit 13. The processing unit 14 stores information on each area and information on the local damaged area 52 in the memory unit 15.

[0078] The processing unit 14 performs post-processing on the image of the second region 42, which is the exposed region of the inner layer rubber 34 (step S303). For example, the processing unit 14 can generate a histogram showing the distribution of pixels in the second region 42 in the width direction of the tire 30, as described with reference to FIG. 6. Furthermore, for example, the processing unit 14 calculates the size and / or area ratio of the second region 42 to the surface region of the tread portion 35 of the tire 30. The processing unit 14 outputs the position, size and / or area ratio of the second region 42 to the surface region of the tread portion 35 of the tire 30 via the output unit 13, or stores them in the storage unit 15.

[0079] The processing unit 14 performs post-processing on the image of the third region 43, which is the exposed region of the metal cord layer 17, in the same manner as the image of the second region 42 (step S304). The sizes of the second region 42 and the third region 43 quantitatively indicate the area where the surface of the tire 30 is damaged over a wide area.

[0080] The processing unit 14 performs post-processing on the image of the local damaged portion 52, such as a chip, output by inputting the tread image 40 into the second trained model (step S305). As described with reference to FIG. 7 , the post-processing of the local damaged portion 52 may include calculating the position of the center of gravity of the local damaged portion 52 and calculating the area ratio of the area of ​​the local damaged portion 52 to the area of ​​the surface region of the tread portion 35 of the tire 30. The processing unit 14 stores information on the position of the center of gravity and the area ratio in the storage unit 15.

[0081] Next, the processing unit 14 generates a mask 56 that covers the second region 42, the third region 43, and the local damaged area 52 on the tread image 40 (step S306). This mask 56 is similar to the mask 56 that was applied to the tread image 40 of the sample tire when generating the third training data. The mask 56 is used to exclude the second region 42, the third region 43, and the local damaged area 52, which are severely damaged, from the areas to be evaluated in the process of determining relatively minor damage due to scratches or the like, which will be described below.

[0082] 10 , the processing unit 14 extracts a rectangular region 55 encompassing the tread portion 35 of the tire 30 from the tread image 40 (step S307). The processing unit 14 can recognize the extent of the tread portion 35 by generating the divided regions in step S302. The processing unit 14 may also apply the mask 56 generated in step S306 to the rectangular region 55. As a result, regions other than the first region 41 where the outer layer rubber 33 is exposed on the surface are masked.

[0083] The processing unit 14 evaluates the damage state of the surface region of the tire 30 (step S308). Specifically, the processing unit 14 inputs the image of the rectangular region 55 extracted in step S307 into the third trained model, and obtains, as output, a numerical value representing the damage state of each subregion 57. The processing unit 14 can display the output numerical value representing the damage state of each subregion 57 on a display unit included in the output unit 13 and store it in the storage unit 15.

[0084] As described above, the tire damage assessment device 10 can output the following assessment result information that quantifies the damage state of the surface region of the tread portion 35 according to the procedure in the flowchart of Fig. 13: (1) Images showing the exposed state of each of the multiple layers of the tread portion 35 due to wear of the tread portion 35 of the tire 30, and quantitative data such as the area of ​​the exposed region of each layer; (2) Images showing localized damaged areas 52 such as chips that have occurred in the tread portion 35 of the tire 30, the number of localized damaged areas 52, the position of the center of gravity of each localized damaged area 52, and the area ratio of the localized damaged areas 52 to the tread portion 35; and (3) An index that quantifies minor damaged areas such as scratches that have occurred in the surface region of the outer rubber layer 33, the outermost layer of the tread portion 35. In this way, the tire damage assessment device 10 disclosed herein can quantify and assess the surface state of the tread portion 35 of the tire 30. The tire damage assessment device 10 may display information about the assessment results on the display unit of the output unit 13 and / or transmit the information to an external device via the communication unit 11. This allows the tire damage assessment device 10 to provide the user with the damage state of the surface region of the tread portion 35 in a manner that is easy for the user to use.

[0085] The processing unit 14 of the tire damage assessment device 10 can further estimate the wear characteristics of the tire 30 based on the output information on localized damage 52 such as chips and information on the state of damage due to scratches and the like. By using the tire damage assessment device 10, the user can analyze the relationship between the material and / or tread pattern shape of the tire 30 and tire wear. Furthermore, by using the information output by the tire damage assessment device 10, the user can predict the lifespan of the tire 30.

[0086] (Another Configuration Example of Tire Damage Assessment System) The tire damage assessment system 1 shown in Fig. 1 is composed of a tire damage assessment device 10 and an imaging device 20. Fig. 14 shows another configuration example of the tire damage assessment system 1. In the embodiment of Fig. 14, the tire damage assessment device 10 is placed in a location remote from the user, for example, on the cloud. The tire damage assessment device 10 is configured to be able to communicate with a user's portable information terminal 60 and an information processing device 70 via a network 80 such as the Internet.

[0087] The portable information terminal 60 includes, for example, a smartphone, a tablet terminal, etc. The information processing device 70 is a computer such as a PC. Similar to the tire damage assessment device 10, the portable information terminal 60 and the information processing device 70 may include components such as a communication unit, an input unit, an output unit, a processing unit, and a storage unit. The portable information terminal 60 may further include an imaging unit.

[0088] The portable information terminal 60 may be configured to transmit the tread image 40 of the tire 30 captured by the imaging unit to the tire damage assessment device 10. Therefore, the portable information terminal 60 functions as the imaging device 20 of the tire damage assessment system 1 of FIG. 1. The portable information terminal 60 may also acquire and execute an application compatible with the tire damage assessment system 1. A user of the portable information terminal 60 may capture an image of the tire 30 by following the instructions of the application and transmit the image to the tire damage assessment device 10.

[0089] The information processing device 70 may acquire the evaluation results from the tire damage assessment device 10 and display them on a display, which is an output unit. The information processing device 70 may store the evaluation results from the tire damage assessment device 10 in a memory unit. The information processing device 70 may perform further analysis using the evaluation results stored in the memory unit. The information processing device 70 may analyze and evaluate damage to the tire 30 while transmitting and receiving information to and from the tire damage assessment device 10 using a dedicated application stored in the memory unit.

[0090] Although the embodiments of the present disclosure have been described based on the drawings and examples, it should be noted that those skilled in the art would easily be able to make various modifications or alterations based on the present disclosure. Therefore, it should be noted that these modifications and alterations are included within the scope of the present disclosure. For example, the functions included in each component or step can be rearranged so as not to cause logical inconsistencies, and multiple components or steps can be combined or divided into one. The embodiments of the present disclosure can also be realized as a method, a program executed by a processor included in an apparatus, or a storage medium on which a program is recorded. It should be understood that these are also included within the scope of the present disclosure.

[0091] [Contribution to the United Nations-led Sustainable Development Goals (SDGs)] The SDGs have been proposed to realize a sustainable society. One embodiment of the present invention is thought to be a technology that can contribute to "No. 9 - Build resilient infrastructure for industry, innovation and other goals."

[0092] REFERENCE SIGNS LIST 10 Tire damage evaluation device 11 Communication unit 12 Input unit 13 Output unit (display unit) 14 Processing unit 15 Memory unit 20 Imaging device 30 Tire 31 Bead portion 32 Sidewall portion 33 Outer layer rubber 34 Inner layer rubber 35 Tread portion 36 Carcass 37 Metal cord layer 38 Bead core 40 Tread image 41 First region 42 Second region 43 Third region 45 Strip-shaped region 51 Groove portion 52 Locally damaged portion 53 Repaired portion 54 Small region 55 Rectangular region 56 Mask 57 Lower region 60 Portable information terminal 70 Information processing device 80 Network

Claims

1. A tire damage evaluation device comprising a processing unit that acquires a tread image of a tire tread portion and quantifies the damage state of the surface region of the tread portion based on the tread image.

2. A tire damage assessment device as described in claim 1, wherein the tread portion is configured to include a plurality of layers arranged in the radial direction of the tire, and the processing unit generates a divided image in which the surface area of ​​the tread portion included in the tread image is divided by the layers exposed in the surface area.

3. A tire damage evaluation device as described in claim 2, wherein the multiple layers include an outer layer of rubber that constitutes the contact surface of the tread portion, a layer of metal cords arranged radially inward of the outer layer of rubber, and an inner layer of rubber located between the outer layer of rubber and the layer of metal cords.

4. The tire damage evaluation device described in claim 2, wherein the processing unit inputs tread images of a plurality of sample tires and executes a process of inputting the tread images of the tires into a first trained model trained using first teacher data that outputs a partitioned image in which the surface area of ​​the sample tires contained in the tread images is partitioned by the layers exposed on the surface, and outputting the partitioned image of the tire.

5. A tire damage evaluation device as described in claim 2, wherein the processing unit divides the section image of the tire into a plurality of rectangular regions extending in the width direction of the tire, aligns the positions of the plurality of rectangular regions so that the widthwise positions of the tread portion within the rectangular regions coincide, and generates a histogram by accumulating the pixels of the plurality of rectangular regions for at least any of the layers exposed to the surface region for each position in the width direction.

6. The tire damage evaluation device of claim 1, wherein the processing unit is configured to extract from the tread image at least any of a groove portion indicating a groove in the tread portion, a local damage portion indicating localized damage to the surface region of the tread portion, and a repaired portion indicating an area where damage that occurred in the past in the tire has been repaired.

7. The tire damage evaluation device described in claim 1, wherein the processing unit executes a process of inputting the tread images of the tire into a second trained model trained using second teacher data that takes as input tread images of a plurality of sample tires and outputs images of local damage areas showing local damage in the tread areas of the sample tires contained in the tread images, and outputting the images of the local damage areas of the tire.

8. The tire damage evaluation device according to claim 7, wherein the processing unit calculates a ratio of an area of ​​the localized damaged portion to an area of ​​the surface region of the tread portion.

9. The tire damage assessment device according to claim 7, wherein the processing unit calculates a center of gravity position of the area of ​​the localized damage portion.

10. A tire damage assessment device as described in claim 2, wherein the processing unit quantifies the damage state of the surface region of the tire where the outermost layer of the plurality of layers of the tire is exposed based on the tread image.

11. A tire damage evaluation device as described in claim 10, wherein the processing unit extracts an image of an area including the surface area of ​​the tread portion from the tread image, divides the image into images of a plurality of sub-areas, and quantifies the damage state of the surface area for each of the images of the sub-areas.

12. The tire damage evaluation device described in claim 10, wherein the processing unit inputs tread images of a plurality of sample tires and executes a process of inputting the tread images of the tire into a third trained model trained using third teacher data that outputs information quantifying the damage state of the surface region of a portion where the outermost layer of the tire is exposed among the plurality of layers included in the tread image, and outputting information quantifying the damage state of the surface region.

13. A tire damage evaluation device as described in claim 12, wherein the damage state of the surface area in the third teaching data is quantified based on at least one of the number of scratches contained in the surface area and the depth of the scratches.

14. A tire damage assessment device as described in claim 1, wherein the damage state includes the state of chips and scratches in the tire, and the processing unit estimates the wear characteristics of the tire based on information on the quantified state of the chips and scratches.

15. A tire damage evaluation device as described in claim 1, comprising at least one of a display unit that displays quantified information on the damage state of the surface region of the tread portion, and a communication unit that transmits the information to an external device.

16. A tire damage assessment method executed by an information processing device, comprising: acquiring a tread image of a tire tread portion; and quantifying a damage state of a surface region of the tread portion based on the tread image.

17. A tire damage evaluation program that causes a computer to execute a process of acquiring a tread image by capturing an image of a tire tread portion, and a process of quantifying the damage state of the surface region of the tread portion based on the tread image.

Images