Apparatus and method for inspecting composite material, program for inspecting composite material, and computer-readable storage medium having program recorded thereon

Abstract

The present invention makes it possible to inspect the fiber arrangement state of a fiber composite material more accurately than conventional technologies even when a measuring instrument having low resolution is used. The present invention also makes it possible to inspect defects in the fiber composite material more easily and accurately than conventional technologies on the basis of images showing the fiber arrangement state of the fiber composite material. By executing image processing for sequentially enlarging the high-density image region for each of steps, the number of steps until the high-density image region is absorbed by another high-density image region and disappears is measured. According to whether or not the measured number of steps is equal to or greater than a prescribed threshold, it is determined whether or not a high-density pixel region has a high-density pixel region having a low density, and the fiber cross-section position is detected on the basis of the determination result.

Description

Composite material inspection device and method, composite material inspection program, and computer-readable storage medium on which the program is recorded

[0001] The present invention relates to an inspection device and method for a composite material, an inspection program for a composite material, and a computer-readable storage medium on which the program is recorded.

[0002] Known composite materials include carbon-fiber reinforced plastic (CFRP), a composite of resin and glass fiber (GFRP), and a material combining a SiC matrix (base material) with SiC fiber. These composite materials have lightweight and high-strength properties, with a specific strength per weight that is approximately 10 times that of steel. For this reason, these composite materials are widely used in various industrial fields as structural materials for flying objects, aircraft, automobiles, and other applications.

[0003] For example, carbon fiber reinforced resin structural materials are made by stacking thin plates (for example, up to 0.2 mm thick) in which carbon fibers (CF) with a diameter of 5 to 7 μm are aligned in one direction and embedded in resin, with the carbon fibers in each layer being sequentially shifted so that their orientation is 0°, 45°, 90°, etc., and then shaping and processing them by heat treatment to integrate them. Inspecting whether the carbon fibers in carbon fiber reinforced resin are regularly aligned is a very important matter in order to ensure the mechanical strength of the entire structural member.

[0004] Conventionally, as a method for non-destructively inspecting structural members made of composite materials, there has been known a material evaluation method for checking large defects (e.g., defects on the order of mm) in manufactured structural members by non-destructive observation such as ultrasonic detection. For example, Patent Document 1 discloses a laser ultrasonic testing device for inspecting the surface of an object.

[0005] Furthermore, composite materials with embedded optical fibers have been known as sensors for ultrasonic inspection. For example, Patent Document 2 discloses a method for manufacturing a material with a built-in optical fiber sensor. Furthermore, Patent Document 3 discloses an ultrasonic inspection system including an optical fiber sensor embedded in the surface of a composite material.

[0006] Furthermore, methods for analyzing the structure of materials using persistent homology have been known for some time. For example, Patent Document 4 discloses an invention in which imaging mass spectrometry is performed on each sample, imaging data is collected for each sample, persistent homology processing is performed based on the imaging data to obtain a persistence diagram, and differences between samples are analyzed based on the persistence diagram.

[0007] Furthermore, structural observation methods using X-rays, particularly X-ray computed tomography (CT), have been known for some time. For example, Non-Patent Document 1 discloses the results of real-time three-dimensional observation of the progression of damage in a carbon fiber reinforced resin composite tube.

[0008] Japanese Patent Application Laid-Open No. 2021-193372 International Publication No. 2008 / 123285 Japanese Patent Application Laid-Open No. 2017-173191 Japanese Patent Application Laid-Open No. 2023-157085

[0009] Y. Wang, SC Garcea and PJ Withers, 7.6 Computed Tomography of Composites, Comprehensive Composite Materials II, 2018.

[0010] A first object of the present invention is to enable inspection of the arrangement of fibers or particles of a composite material with higher accuracy than conventional techniques, even when using a measuring instrument with low spatial resolution.

[0011] A second object of the present invention is to enable easier and more accurate inspection of defects in composite materials based on images showing the arrangement of fibers or particles in the composite material, compared to conventional techniques.

[0012] The first aspect is an inspection device for a composite material, which includes an inspection object image acquisition unit that acquires an inspection object image of the composite material; a pixel region division unit that divides the acquired inspection object image into a low-density pixel region and a high-density pixel region; a measurement unit that measures the number of steps until the high-density image region is absorbed into another high-density image region by performing image processing that sequentially expands the high-density image region for each step; a judgment unit that judges whether a low-density high-density pixel region exists within the high-density pixel region depending on whether the measured number of steps is equal to or greater than a predetermined threshold; and a fiber cross section / particle cross section position detection unit that detects the cross section positions of fibers or particles based on the judgment result of the judgment unit.

[0013] A second aspect is a composite material inspection device according to the first aspect, wherein the determination unit further determines, based on the number of measured steps, that a high-density pixel region and another high-density pixel region correspond to a fiber or particle cross section and a base material, respectively.

[0014] The third aspect is a composite material inspection device in the first aspect, which is equipped with a homogenization processing unit that homogenizes pixel values ​​in areas where a high-density pixel area and another high-density pixel area exist within the same fiber or particle cross section.

[0015] A fourth aspect is a composite material inspection device that includes an inspection object image acquisition unit that acquires an inspection object image of the composite material; a fiber cross section / particle cross section position acquisition unit that acquires the position indicating each fiber or particle cross section in the acquired inspection object image; an arrangement feature value acquisition unit that acquires an arrangement feature value that indicates the characteristics of the arrangement of each fiber or particle cross section; an arrangement state quantity measurement unit that measures an arrangement state quantity corresponding to the deviation of the acquired arrangement feature value from a reference value; and a defect location detection unit that associates the arrangement state quantity with the fiber or particle cross section position to detect locations that indicate defects in the composite material.

[0016] A fifth aspect is an inspection device for a composite material, comprising: an inspection object image acquisition unit that acquires an inspection object image of the composite material; a fiber cross section / particle cross section position acquisition unit that acquires a position indicating each fiber or particle cross section in the acquired inspection object image; an arrangement state quantity acquisition unit that acquires an arrangement state quantity indicating the arrangement state of each fiber or particle cross section by performing image processing to enlarge each of the fiber or particle cross sections; and a defect location detection unit that detects a location indicating a defect in the composite material by correlating the acquired arrangement state quantity with the fiber or particle cross section position.

[0017] This is a composite material inspection device equipped with the above.

[0018] A sixth aspect is a method for inspecting a composite material, the method including: an inspection object image acquisition step for acquiring an inspection object image of the composite material; a pixel region division step for dividing the acquired inspection object image into a low-density pixel region and a high-density pixel region; a measurement step for measuring the number of steps until the high-density image region is absorbed into another high-density image region by performing image processing that sequentially enlarges the high-density image region for each step; a determination step for determining whether a low-density high-density pixel region exists within the high-density pixel region depending on whether the measured number of steps is equal to or greater than a predetermined threshold; and a fiber cross section / particle cross section position detection step for detecting the cross section positions of the fibers or particles based on the determination result.

[0019] The seventh aspect is a method for inspecting a composite material, which includes an inspection object image acquisition step for acquiring an inspection object image of the composite material; a fiber cross section / particle cross section position acquisition step for acquiring positions indicating each fiber or particle cross section in the acquired inspection object image; an arrangement feature value acquisition step for acquiring arrangement feature values ​​indicating characteristics of the arrangement of each fiber or particle cross section; an arrangement state quantity measurement step for measuring an arrangement state quantity corresponding to a deviation of the acquired arrangement feature value from a reference value; and a defect location detection step for detecting locations indicating defects in the composite material by correlating the arrangement state quantity with the fiber or particle cross section position.

[0020] The eighth aspect is a method for inspecting a composite material, which includes an inspection object image acquisition step for acquiring an inspection object image of the composite material; a fiber cross section / particle cross section position acquisition step for acquiring positions indicating each fiber or particle cross section in the acquired inspection object image; an arrangement state quantity acquisition step for acquiring arrangement state quantities indicating the arrangement state of each fiber or particle cross section by performing image processing to enlarge each fiber or particle cross section; and a defect location detection step for detecting locations indicating defects in the composite material by correlating the acquired arrangement state quantities with fiber or particle cross section positions.

[0021] A ninth aspect is a composite material inspection program for causing a computer to execute an inspection object image acquisition process for acquiring an inspection object image of a composite material; a pixel region division process for dividing the acquired inspection object image into a low-density pixel region and a high-density pixel region; a measurement process for measuring the number of steps until a high-density image region is absorbed into another high-density image region by performing image processing for sequentially enlarging the high-density image region at each step; a determination process for determining whether a low-density high-density pixel region exists within the high-density pixel region depending on whether the measured number of steps is equal to or greater than a predetermined threshold; and a fiber cross section / particle cross section position detection process for detecting the cross section positions of fibers or particles based on the determination result.

[0022] In a tenth aspect, a computer is provided with: an inspection object image acquisition process for acquiring an inspection object image of a composite material; a fiber cross section / particle cross section position acquisition process for acquiring positions indicating cross sections of each fiber or particle in the acquired inspection object image; and an array feature value acquisition process for acquiring array feature values ​​indicating the characteristics of the array of each fiber or particle cross section.

[0023] This is a composite material inspection program for executing an arrangement state quantity measurement process that measures an arrangement state quantity corresponding to the deviation of the acquired arrangement feature value from a reference value, and a defect location detection process that associates the arrangement state quantity with the cross-sectional position of a fiber or particle to detect locations that indicate defects in the composite material.

[0024] The eleventh aspect is a composite material inspection program for causing a computer to execute an inspection object image acquisition process for acquiring an inspection object image of a composite material; a fiber cross section / particle cross section position acquisition process for acquiring the position indicating each fiber or particle cross section in the acquired inspection object image; an arrangement state quantity acquisition process for acquiring an arrangement state quantity indicating the arrangement state of each fiber or particle cross section by performing image processing to enlarge each fiber or particle cross section; and a defect location detection process for detecting locations indicating defects in the composite material by correlating the acquired arrangement state quantity with the fiber or particle cross section position.

[0025] A twelfth aspect is a computer-readable storage medium having recorded thereon a composite material inspection program for causing a computer to execute the following: an inspection object image acquisition process for acquiring an inspection object image of a composite material; a pixel area division process for dividing the acquired inspection object image into a low-density pixel area and a high-density pixel area; a measurement process for measuring the number of steps until a high-density image area is absorbed into another high-density image area by performing image processing that sequentially expands the high-density image area at each step; a determination process for determining whether a low-density high-density pixel area exists within the high-density pixel area depending on whether the measured number of steps is equal to or greater than a predetermined threshold; and a fiber cross section / particle cross section position detection process for detecting the cross section positions of fibers or particles based on the determination result.

[0026] In a thirteenth aspect, a computer is provided with: an inspection object image acquisition process for acquiring an inspection object image of a composite material; a fiber cross section / particle cross section position acquisition process for acquiring positions indicating cross sections of each fiber or particle in the acquired inspection object image; and an array feature value acquisition process for acquiring array feature values ​​indicating the characteristics of the array of each fiber or particle cross section.

[0027] This is a computer-readable storage medium that records a composite material inspection program for executing an arrangement state quantity measurement process that measures an arrangement state quantity corresponding to the deviation of an acquired arrangement feature value from a reference value, and a defect location detection process that associates the arrangement state quantity with the cross-sectional position of a fiber or particle to detect locations that indicate defects in the composite material.

[0028] A fourteenth aspect is a computer-readable storage medium having recorded thereon a composite material inspection program for causing a computer to execute an inspection object image acquisition process for acquiring an inspection object image of a composite material; a fiber cross section / particle cross section position acquisition process for acquiring the position indicating each fiber or particle cross section in the acquired inspection object image; an arrangement state quantity acquisition process for acquiring an arrangement state quantity indicating the arrangement state of each fiber or particle cross section by performing image processing to enlarge each fiber or particle cross section; and a defect location detection process for matching the acquired arrangement state quantity with the fiber or particle cross section position to detect locations indicating defects in the composite material.

[0029] According to the first to third, sixth, ninth and twelfth aspects, even when a measuring instrument with low resolution is used, the arrangement state of fibers or particles of a composite material can be inspected with higher accuracy than conventional techniques.

[0030] Furthermore, according to the fourth, fifth, seventh, eighth, tenth, eleventh, thirteenth and fourteenth aspects, defects in composite materials can be inspected more easily and with higher accuracy than with conventional techniques, based on images showing the arrangement of fibers or particles in the composite material.

[0031] FIG. 1 is a block diagram showing the functional configuration of a fiber composite material inspection device according to a first embodiment. FIG. 2 is a diagram illustrating the measurement processing performed by the measurement unit. FIG. 3 is a block diagram showing the functional configuration of a fiber composite material inspection device according to a second embodiment. FIG. 4 is a diagram illustrating the results of associating each coordinate point of the persistence diagram shown in FIG. 17(A) with the cross-sectional positions of fibers in the original inspection target image by inverse analysis. FIG. 5 is a diagram illustrating a system configuration in which a fiber composite material inspection device according to an embodiment is configured on a network. FIG. 6 is a diagram illustrating the hardware configuration of an inspection computer terminal, a management computer terminal, and a server according to an embodiment. FIG. 7A is a flowchart showing a fiber composite material inspection program according to the first embodiment. FIG. 7B is a flowchart showing a fiber composite material inspection program according to a second embodiment. FIGS. 8A(A), (B), and (C) are diagrams illustrating the image processing procedure for an inspection target image of a fiber composite material. FIGS. 8B(A), (B), and (C) are diagrams illustrating the image processing procedure for an inspection target image of a fiber composite material. FIGS. 8C(A), (B), and (C) are diagrams illustrating the image processing procedure for an inspection target image of a fiber composite material. FIG. 9 is a diagram illustrating an example of smoothing processing. FIG. 10 is a diagram illustrating measurement processing performed by the measurement unit. FIG. 11 is a diagram illustrating the relationship between the number of steps until a hole disappears and the frequency. FIG. 12 is a diagram plotting coordinate points at which holes appear and disappear, with the horizontal axis representing the number of steps at which a hole is generated and the vertical axis representing the number of steps at which a hole disappears. FIG. 13(A) is a diagram illustrating the geometric relationship when adjacent fibers are arranged in a triangle, and FIG. 13(B) is a diagram illustrating the geometric relationship when adjacent fibers are arranged in a square. FIGS. 14(A) and (B) are diagrams illustrating changes in the cross-sectional shape of each fiber arranged in an equilateral triangle shown in FIG. 13(A). FIG. 15(A) is a diagram illustrating an example of an inspection target image, and FIG. 15(B) is a diagram illustrating the alignment state quantities of each corresponding region. FIGS. 16(A) and (B) are diagrams illustrating an example of alignment state quantity acquisition processing. FIG. 17A is a diagram illustrating an example of a persistence diagram, and FIGS. 17B and 17C are diagrams illustrating the results of inverse analysis based on the persistence diagram.

[0032] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of an inspection device and method for a composite material, an inspection program for a composite material, and a computer-readable storage medium storing the program according to the present invention will be described with reference to the drawings.

[0033] (Outline of the embodiment)

[0034] The composite material to which this embodiment is applied is a material in which materials with different properties are combined and used as a single material. Composite materials include fiber composite materials composed of fibers and a matrix material, and particle composite materials composed of particles and a matrix material. In the following embodiment, a fiber composite material will be described as an example of a composite material. A fiber composite material is composed of a fiber reinforcing material such as carbon fiber, SiC (silicon carbide) fiber, or glass, and a matrix material (resin, SiC, etc.) that serves as a matrix material to support the fiber reinforcing material. The types and combinations of the fiber reinforcing material and matrix material can be selected arbitrarily.

[0035] Composite materials are lighter and stronger than existing metallic materials, making them suitable for use in industrial fields where lightweight, high-strength structures are required.

[0036] However, because composite materials are made by combining materials with different properties (fiber material and matrix material), differences in the properties of each material (mechanical strength, elastic modulus, etc.) and the existence of interfaces between different materials can cause defects such as cracks to occur under severe loading conditions, deteriorating the mechanical properties of the entire material. Therefore, observing (monitoring) defects such as cracks is essential for the appropriate manufacturing and use of composite materials.

[0037] Hereinafter, carbon-reinforced plastic (CFRP) will be used as an example for explanation, but the present embodiment is not limited to this.

[0038] Identifying the location of carbon fibers is an important factor that determines the mechanical strength and fatigue of the entire composite material, such as carbon fiber reinforced plastic.

[0039] Ultrasonic inspection has been known as a non-destructive, three-dimensional material evaluation method for manufactured carbon fiber reinforced plastic structural materials, but ultrasonic inspection can only detect large defects (e.g., defects on the order of mm) in manufactured structural members.

[0040] On the other hand, by using X-ray techniques, particularly X-CT, it is possible to observe three-dimensional microstructures with a spatial resolution of about several micrometers.

[0041] However, the method using X-CT has the problem that even though it can observe with a spatial resolution of about several μm, the resolution is insufficient to identify the position of carbon fibers. In other words, a general high-spatial resolution X-CT device (spatial resolution of several μm) has insufficient spatial resolution to identify carbon fibers with a diameter of about 5 to 6 μm.

[0042] There was no three-dimensional non-destructive observation method that could detect the arrangement of small carbon fibers over a wide field of view.

[0043] Even if it were possible to detect the position of carbon fibers in some way, there was no method for evaluating the variation in the three-dimensional carbon fiber arrangement. This posed a challenge in quality control of carbon fiber reinforced resins, optimizing manufacturing conditions, and predicting damage and deterioration during use. In other words, even if it became possible to observe carbon fibers only in cross section using an electron microscope, there was no method for quantifying the carbon fiber arrangement and determining whether it was good or bad.

[0044] This embodiment was developed after extensive study to solve these two problems. To solve the above-mentioned problems, the present embodiment was conceived by focusing on the arrangement of carbon fibers and using an analytical method based on persistent homology, which is one of the topological data analysis methods, based on mathematical knowledge and analysis, rather than black-box methods such as simple image processing or machine learning. This embodiment can automatically quantify the disorder in the arrangement of carbon fibers, and furthermore, can identify the locations of the disordered arrangement in a composite material member by inverse analysis, and has the characteristics of being understandable and readable by humans.

[0045] This embodiment contributes to quality control of carbon fiber reinforced resin composite materials, optimization of manufacturing conditions, and prediction of breakage and deterioration during use.

[0046] (Device configuration)

[0047] Hereinafter, an embodiment of an inspection device for a fiber composite material according to the present invention will be described with reference to the drawings.

[0048] FIG. 1 is a block diagram showing the functional configuration of a fiber composite material inspection device 100 according to the first embodiment.

[0049] The fiber composite material inspection device 100 of the first embodiment is configured to include an inspection object image acquisition unit 110, a pixel region division unit 120, a measurement unit 130, a determination unit 140, a homogenization processing unit 150, and a fiber cross-section position detection unit 160.

[0050] FIG. 3 is a block diagram showing the functional configuration of a fiber composite material inspection device 100 according to the second embodiment.

[0051] The fiber composite material inspection device 100 of the second embodiment is configured to include an inspection object image acquisition unit 110, a fiber cross-section position acquisition unit 170, an array feature value acquisition unit 180, an array state quantity measurement unit 190, and a defect location detection unit 200.

[0052] FIG. 5 is a diagram illustrating a system configuration in which the fiber composite material inspection device 100 according to the embodiment is configured as a network.

[0053] The functions of Figures 1 and 3 can be realized by a combination of an inspection-side computer terminal 10, a management-side computer terminal 20, a server 30, and a network 40 that communicatively connects the inspection-side computer terminal 10, the management-side computer terminal 20, and the server 30, as shown in Figure 5.

[0054] The server 30 stores an inspection program for fiber composite materials so that it can be accessed from outside.

[0055] The management computer terminal 20 is provided on the system administrator side. The system administrator has the authority to access the server 30 via the management computer terminal 20 and to create, edit, update, etc., an inspection program for a fiber composite material.

[0056] The inspection-side computer terminal 10 is provided on the side of a user who inspects a carbon fiber reinforced resin composite material. The user accesses the server 30 via the inspection-side computer terminal 10 and executes an inspection program for a fiber composite material to inspect the carbon fiber reinforced resin composite material.

[0057] 1 and 3 can also be realized by a single computer terminal, for example, the inspection-side computer terminal 10. By installing a fiber composite material inspection program in a single computer terminal, the functions of the fiber composite material inspection device 100 can be realized by a single computer terminal, for example, the inspection-side computer terminal 10 alone.

[0058] The network 40 is configured by the Internet, an intranet, etc. The server 30 is configured by a server device as a specific computer or a virtual server built on a cloud computing service, etc.

[0059] FIG. 6 is a diagram illustrating the hardware configuration of the inspection-side computer terminal 10, the management-side computer terminal 20, and the server 30 according to the embodiment, and is a diagram illustrating the hardware configuration for realizing the functional configurations of FIGS.

[0060] As shown in FIG. 6 , the inspection-side computer terminal 10, the management-side computer terminal 20, and the server 30 are each configured such that a CPU (Central Processing Unit) 11, a ROM (Read Only Memory) 12, a RAM (Random Access Memory) 13, a storage 14, an input device 16, a display device 17, a communication interface 18, and an external storage device 19 are connected to each other via a system bus 15 so that they can communicate with each other.

[0061] The CPU 11 is a central processing unit that executes various programs and controls each device connected to the system bus 15. That is, the CPU 11 reads a program from the ROM 12 or the storage 14 and executes the program using the RAM 13 as a work area. The CPU 11 controls each device connected to the system bus 15 and performs various arithmetic processing in accordance with the program recorded in the ROM 12 or the storage 14.

[0062] The ROM 12 or storage 14 stores various programs, including a BIOS (Basic Input / Output System) and an OS (Operating System), which are control programs executed by the CPU 11, and a fiber composite material inspection program that can be read and executed by a computer to realize this embodiment, as well as various necessary data.

[0063] The RAM 13 functions as a main memory, work area, etc. for the CPU 11 and temporarily stores programs or data as a work area. The storage 14 is configured with an HDD (Hard Disk Drive) or an SSD (Solid State Drive) and stores various programs including the BIOS and OS, and various data.

[0064] The input device 16 includes a pointing device such as a mouse, a keyboard, a reading device such as a scanner, and is used to input various types of information.

[0065] The display device 17 is, for example, an organic EL display or a liquid crystal display, and displays various information including information showing the execution results of the fiber composite material inspection program. The display device 17 may be a touch panel type and function as the input device 16.

[0066] The communication interface 18 is an interface for communicating with other devices, and uses standards such as Ethernet (registered trademark), FDDI, Wi-Fi (registered trademark), etc. The communication interface 18 connects to a network 40 and controls the transmission and reception of data.

[0067] The external storage device 19 is composed of various types of memory cards such as USB memory, HDD, SSD, and other external storage media that can be detachably connected.

[0068] (First embodiment)

[0069] The inspection side computer terminal 10 is located in the design / manufacturing department, quality control department, etc. of an institution or company (e.g., a manufacturer) that manufactures, uses, or inspects composite material carbon fiber reinforced resin (hereinafter referred to as fiber composite material).

[0070] The X-ray CT device 50 acquires an inspection object image of a fiber composite material. The X-ray CT device 50 corresponds to the inspection object image acquisition unit 110. The X-ray CT device 50 and the inspection-side computer terminal 10 are connected to each other by wire or wirelessly so that data can be transmitted and received. The inspection object image acquired by the X-ray CT device 50 is imported into the inspection-side computer terminal 10.

[0071] 7A is a flowchart showing the fiber composite material inspection program PB1 of Embodiment 1. Each process of the fiber composite material inspection program PB1 corresponds to each step of the fiber composite material inspection method of Embodiment 1.

[0072] The fiber composite material inspection program PB1 is stored in the ROM 12, storage 14, or external storage device 19 of the server 30, the inspection-side computer terminal 10, or the management-side computer terminal 20 so that it can be processed by the inspection-side computer terminal 10 or the management-side computer terminal 20. The ROM 12, storage 14, or external storage device 19 is an example of a computer-readable storage medium on which the fiber composite material inspection program PB1 is recorded. The fiber composite material inspection programs PB2, PB3, and PB4 shown in Figures 7B, 7C, and 7D are also similarly recorded in computer-readable storage media such as the ROM 12, storage 14, or external storage device 19.

[0073] Each process performed by the fiber composite material inspection device 100 of the first embodiment will be described below with reference to the flowchart shown in FIG. 7A.

[0074] (Inspection object image acquisition process S11)

[0075] The X-ray CT device 50 acquires an image of the fiber composite material to be inspected.

[0076] 8A, 8B, and 8C are diagrams for explaining the image processing procedure for an inspection image of a fiber composite material.

[0077] FIG. 8A (A) shows an inspection object image 300 of a fiber composite material captured by the X-ray CT device 50.

[0078] FIG. 8A(B) is an inspection object image 301 showing an enlarged portion of the inspection object image 300.

[0079] The X-ray CT device 50 can capture images with a typical high spatial resolution (spatial resolution is 1 μm). In the inspection object image 301, pixel values ​​(brightness, luminance) increase according to the density. In areas with high density, pixel values ​​increase, and the brightness in the inspection object image 301 increases, approaching white. Conversely, in areas with low density, pixel values ​​decrease, and the brightness in the inspection object image 301 decreases, approaching black. Note that pixel values ​​are represented by digital data with, for example, 256 gradations, and various image processing operations are performed by arithmetic processing of digital signals indicating pixel values.

[0080] Ideally, the carbon fiber F area will have a high density, a high pixel value, and appear white in the image. The resin M area will have a low density, a low pixel value, and appear gray in the image. Areas where defects such as cracks have occurred will have an even lower density, an even lower pixel value, and appear black in the image.

[0081] However, depending on the type of fiber composite material and noise in the X-ray CT measurement and image reconstruction calculation, the density difference between the carbon fiber F and the matrix resin M may be small.

[0082] Furthermore, depending on the type of fiber composite material, a black region with a lower density and lower pixel value appears in the inner part of the carbon fiber F than in the outer part. In other words, the pixel value may be slightly lower in the inner part of the carbon fiber F than in the outer part.

[0083] This, combined with the limited resolution of the X-ray CT device 50 (spatial resolution is 1 μm), results in a poor S / N ratio for the measurement data, and at the stage when the inspection target image 301 is acquired, there is a risk that the position (cross-sectional center position) P of the carbon fiber F in the image cannot be identified.

[0084] Therefore, in the following, in order to identify the position of the carbon fiber F, image processing is performed to recognize the outer and inner portions of the carbon fiber F together as the carbon fiber F. The inner portion of the carbon fiber F is assumed to have an area where the density is slightly lower than that of the outer portion, and focusing on the characteristic of composite materials that the cross-sectional shape of the carbon fiber F is approximately circular or elliptical, image processing and geometric techniques are used to identify the position P of the center of the carbon fiber F with a high degree of accuracy that exceeds the resolution of the X-CT device 50.

[0085] Next, image processing is performed to increase the contrast of the inspection object image 301 .

[0086] FIG. 8B(A) shows an inspection object image 302 obtained by increasing the contrast of the inspection object image 300 .

[0087] In the inspection target image 302, the density of the inner portion of the carbon fiber F is slightly lower than that of the outer portion, and irregular holes are observed in the center of the carbon fiber F. Therefore, even if the contrast of the image is simply increased as has been done in the past, there is a risk that the position P of the carbon fiber F in the image cannot be identified.

[0088] Next, a smoothing process is executed to smooth the inspection target image 302 .

[0089] FIG. 8B(B) shows an inspection object image 303 obtained by smoothing the inspection object image 302 .

[0090] The smoothing process is performed to eliminate variations in pixel values ​​of the pixels in the inspection target image 302 and to facilitate subsequent image processing.

[0091] 9 is a diagram illustrating an example of the smoothing process. As shown in Fig. 9, the measured pixel value for each pixel G is updated to a pixel value weighted according to the average value of the pixel values ​​of the surrounding pixels G1, G2, G3, G4, G5, G6, G7, and G8, thereby obtaining a smoothed inspection target image 303.

[0092] (Pixel region division process S12)

[0093] The pixel region dividing unit 120 divides the acquired inspection target image 303 into a low-density pixel region GL and a high-density pixel region GU. Here, the high-density pixel region GU is either a "high-density pixel region distributed in the inner part of the fiber cross section," a "high-density pixel region present in the outer part of the fiber cross section," or a "high-density pixel region distributed in the matrix resin M."

[0094] For example, the pixel value of each pixel in the inspection target image 303 can be compared with a predetermined threshold value, and the image can be divided into a low-density pixel region GL and a high-density pixel region GU depending on whether the pixel value is below the predetermined threshold value.

[0095] (Measurement process S13)

[0096] The measurement unit 130 performs image processing to sequentially enlarge the high density image region GU for each step, thereby measuring the number of brightness steps Ns until the high density image region GU2 is absorbed into another high density image region GU1 and disappears.

[0097] The inner portion of the fiber F has a high-density region (called high-density image region GU2) that is slightly lower in density than the outer portion (called high-density image region GU1) of the fiber F. Therefore, in order to make the inner portion of the fiber F have the same pixel value as the outer portion of the fiber F, a measurement process S13 is executed to determine the inner portion of the fiber F.

[0098] FIG. 10 is a diagram illustrating the measurement process S13 performed by the measurement unit 130.

[0099] FIG. 10A is a diagram showing the relationship between position and pixel value (brightness, luminance), and is a diagram showing the density distribution of two carbon fibers F and a matrix resin M.

[0100] Figure 10(B) shows the distribution of low-density pixel regions GL (shown in hatched areas) and high-density pixel regions GU (shown in white) binarized by a specified threshold value Ns when the specified threshold value Ns is large (Ns = Ns1).

[0101] Figure 10(C) shows the distribution of low-density pixel regions GL (shown in hatching) and high-density pixel regions GU (shown in white) binarized by a specified threshold value Ns when the specified threshold value Ns is smaller (Ns = Ns2) compared to Figure 10(B).

[0102] Figure 10(D) shows the distribution of low-density pixel regions GL (shown in hatched areas) and high-density pixel regions GU (shown in white) binarized by a specified threshold value Ns when the specified threshold value Ns is smaller (Ns = Ns3) compared to Figure 10(C).

[0103] 10(C), an image divided into a low-density pixel region GL and a high-density pixel region GU by the pixel region division process S12 is defined as an original image 402. The original image 402 includes, for example, two independent high-density pixel regions GU1 and GU2 and a low-density pixel region GL. The high-density image region GU1 corresponds to the outer portion of the fiber F. The high-density pixel region GU2 corresponds to the inner portion of the fiber F. The inner portion of the fiber F is a high-density region with a slightly lower density than the outer portion of the fiber F.

[0104] It is generally difficult to determine by image processing whether the high-density pixel regions GU2 are distributed within the cross section of the same fiber F or whether they are from the cross sections of different fibers F, but this embodiment makes it possible to reliably perform this determination. The algorithm to be applied will be described below.

[0105] To make the low-density region inside the fiber F the same pixel value as the outer region, for example, the Watershed algorithm can be applied. The Watershed algorithm likens the image's brightness gradient to a topographical map of mountains and valleys, imagining water flowing through them and identifying the watersheds (walls) that store water as "contours." When a grayscale image is treated as a topography (high pixel values ​​represent peaks, low pixel values ​​represent valleys), and the entire image is filled with water (when the water level is raised), the surrounding areas are filled first, followed by the independent valleys (minima). The method that focuses on the connection of valleys is called the sublevel-set method, and the method that focuses on the connection of mountains is called the superlevel-set method. Figure 10 illustrates the processing method using the superlevel-set method, in which the threshold brightness is changed in descending steps.

[0106] The low-density pixel region GL and the low-density high-density pixel region GU2 are regarded as holes h in the high-density pixel region GU1, and the high-density pixel region GU is expanded step by step. This makes it easy to determine whether the low-density high-density pixel region GU2 is within the high-density pixel region GU1. By lowering a predetermined threshold value step by step and sequentially expanding the high-density pixel region GU, images 401, 402, and 403 are sequentially acquired. This corresponds to calculating a first-order persistent homology in which the high-density pixel region GU surrounds the low-density pixel region GL based on the super-level set method. By lowering the predetermined threshold value for each brightness step, the number of holes h nh (0, 2, 0) for the brightness step number Ns (Ns1, Ns2, Ns3) is sequentially counted. A change in the number of holes h nh from 0 to 2 indicates that a figure has appeared in two locations within the image in which the high-density image region GU1 surrounds the low-density high-density pixel region GU2. The number of brightness steps Ns at this time is counted as, for example, 200. The subsequent change in the number of holes h, nh, from 2 to 0 means that the low-density high-density image region GU2 is integrated into another high-density image region GU1, and the holes disappear. The number of brightness steps Ns at this time is counted as, for example, 160.

[0107] (Determination process S14)

[0108] The judgment unit 140 judges whether the low-density high-density pixel area GU2 is located within the cross section of the fiber F or outside the fiber F, depending on whether the number of brightness steps (e.g., 160) when the low-density high-density image area GU2 is absorbed into the surrounding high-density image area GU1 and the hole disappears is greater than or equal to a predetermined threshold value Nst (e.g., 95).

[0109] That is, the number of brightness steps Ns until the low-density high-density image region GU2 is absorbed and disappears into another high-density image region GU1 is a large value equal to or greater than a predetermined threshold value Nst when the hole h is present inside the fiber, and is a small value below the predetermined threshold value Nst when the hole h is present outside the fiber. For example, the region P present inside the carbon fiber F observed in Figure 8B (B) corresponds to the former, and the region M present outside the carbon fiber F corresponds to the latter.

[0110] 11 shows the relationship between the number of brightness steps Ns until the hole h disappears and the frequency. As shown in Fig. 11, a predetermined threshold value Nst (e.g., Nst = 95) is set for the number of brightness steps Ns until the hole h disappears, and when the number of brightness steps Ns is greater than the predetermined threshold value Nst, it can be determined that the corresponding low-density high-density image region GU2 is present in the inner part of the carbon fiber F.

[0111] (Equalization process S15)

[0112] As described above, the determination process S14 identifies high-density image regions GU2 present in the inner part of the fiber F. The homogenization processing unit 150 homogenizes the pixel values ​​of the cross section of the fiber F by setting the pixel values ​​of the inner part of the fiber, where the high-density pixel regions GU2 identified by the determination process S14 are scattered, to the same value as the outer part of the fiber.

[0113] As shown in Figure 8B (C), a process is performed in which a white disk is placed so that the low-density area on the inner side of the carbon fiber F has the same pixel value as the outer side, and an image 304 is obtained in which the entire cross section of the carbon fiber F has an almost uniform pixel value.

[0114] Next, a smoothing process similar to that shown in Figure 9 is performed again, and the pixel values ​​of the cross-sectional region of the fiber F are made uniform, and a smoothed inspection object image 305 is obtained, as shown in Figure 8C (A).

[0115] Even after the homogenization process, high-density ghost regions that appear slightly white may be observed in the areas between the carbon fibers F that should be resin M, as shown by the arrows in Figure 8C (A), for example.

[0116] In such a case, the determination process S14 may be executed again to exclude the ghost region (resin portion M). That is, by executing the same determination process S14, if the number of steps Ns until the hole h disappears is equal to or less than a predetermined threshold value Nst, it can be determined that the corresponding low-density high-density image region exists inside the carbon fiber F, that is, it is a ghost.

[0117] (Fiber cross section position detection process S16)

[0118] The fiber cross-sectional position detection unit 160 detects the cross-sectional position P of the fiber F based on the determination result.

[0119] FIG. 8C (C) shows the inspection object image 307 after the fiber cross section position detection process S16 has been executed.

[0120] In the image 305 that has undergone the homogenization process, circular or nearly circular regions that are high-density regions GU are determined to be fiber F regions, and their center positions are detected as death points of the first-order persistent homology of the sublevel-set. As described above, the sublevel-set method focuses on the connection of valleys (minima). Specifically, in the opposite way to what was explained in Figure 10, the predetermined threshold value Th is increased to search for the center position of the fiber or particle (indicated by an x).

[0121] This measurement process will be described with reference to FIG.

[0122] FIG. 2A is a diagram showing the relationship between position and pixel value (brightness, luminance), and is a diagram showing the density distribution of two carbon fibers F and a matrix resin M.

[0123] Figure 2(B) shows the distribution of low-density pixel regions GL (shown in hatched areas) and high-density pixel regions GU (shown in white) binarized by a specified threshold value Th when the specified threshold value Th is small (Th = Th1).

[0124] Figure 2(C) shows the distribution of low-density pixel regions GL (shown in hatched areas) and high-density pixel regions GU (shown in white) binarized by a specified threshold value Th when the specified threshold value Th is larger (Th = Th2) compared to Figure 2(B).

[0125] The predetermined threshold value Th is increased step by step to sequentially expand the low-density pixel region GL, thereby sequentially acquiring images 411 and 412. The positions (indicated by x marks) where the high-density image region GU is absorbed and disappears can be detected as the center positions P1 and P2 of each fiber F.

[0126] In this manner, the central positions P1, P2, P3, P4, P5, P6 and P7 of each fiber F in FIG. 8C (C) are identified.

[0127] The center positions P1, P2, P3, P4, P5, P6, and P7 of the fiber F are each represented by a coordinate position (X, Y) on the inspection target image 305.

[0128] FIG. 8C(B) is a diagram showing a comparative example, showing an image 306 in which P' derived from the ghost-like resin portion is not removed.

[0129] If the fiber cross section position detection process S16 is performed on the image 306 as is, P' resulting from the ghost-like resin portion may be erroneously detected as the center position of the fiber F.

[0130] The determining unit 140 further determines whether the high-density pixel region GU in FIG. 2 corresponds to the cross section of the fiber F or the base material M, based on the number of steps measured.

[0131] For example, as shown in Figure 8B (B), a high-density ghost region that appears slightly white is observed in the area between carbon fibers F that should be resin M. This ghost region occurs due to noise and low S / N in the X-ray CT measurement and image reconstruction calculation.

[0132] Therefore, in order to avoid mistakenly determining that the ghost areas between the carbon fibers F, which should actually be determined to be resin M, are actually fiber F areas, a separate threshold value is set and a determination process is performed.

[0133] To perform this determination process, threshold values ​​Nbt and Nst2 are set for the number of brightness steps Nb at which the hole h appears and the number of brightness steps Ns at which the hole h disappears, respectively.

[0134] 12 plots coordinate points (Nb, Ns) where holes h appear and disappear, with the horizontal axis representing the number of brightness steps Nb at which holes h appear and disappear, and the vertical axis representing the number of brightness steps Ns at which holes h disappear. The shading of the plotted points in the figure corresponds to the frequency of the coordinate points (Nb, Ns).

[0135] In Figure 12, range FA indicates the range of coordinate points (Nb, Ns) where the number of brightness steps Nb at which a hole h appears and the number of brightness steps Ns at which a hole h disappears are equal to or greater than threshold values ​​Nbt and Nst2, respectively. Range FA corresponds to the region of only fiber F, excluding the ghost region (resin portion M). The ghost region (resin portion M) is included outside range FA.

[0136] The determination unit 140 determines that the high-density pixel region GU2 and the other high-density pixel region GU1 each correspond to a cross section of the fiber F when the number of brightness steps Nb and the number of brightness steps Ns are greater than or equal to the threshold values ​​Nbt and Nst2, respectively.

[0137] When at least one of the number of brightness steps Nb and the number of brightness steps Ns is smaller than the threshold values ​​Nbt and Nst2, the determination unit 140 determines that the high-density pixel region GU2 and the other high-density pixel region GU1 correspond to the fiber F and the base material M (ghost region), respectively. In this way, the determination unit 140 can appropriately remove the base material M (ghost region) and accurately identify the cross section of the fiber F.

[0138] When the ghost region (resin portion M) is removed by the judgment process S14 and the fiber cross-section position detection process S16 is then executed, the center positions P1, P2, P3, P4, P5, P6, and P7 of the fibers F can be accurately detected without any false detection, as shown in Figure 8C (C).

[0139] 8A(C) shows an inspection target image 310, which is an entire image corresponding to the inspection target image 300 and has been subjected to the fiber cross section position detection process S16. Data on the center positions P (P1, P2, P3, P4, P5, P6, P7, ...) of each fiber F in the image is associated with the inspection target image 310.

[0140] As described above, according to the first embodiment, even if the inspection object image 300 is acquired using a general high spatial resolution X-CT device 50 (spatial resolution is several μm), the position P of each carbon fiber F having a diameter of approximately 5 to 6 μm can be accurately detected.

[0141] The above has described a case where the position P of the fiber F is detected for a carbon fiber reinforced resin composite material in which high-density fiber F exists in a low-density matrix M. However, the position of the fiber can also be accurately detected in a manner similar to the first embodiment for composite materials in which the matrix and fiber densities are close to each other, such as a composite material in which the matrix has a high density and the fiber has a low density, or a composite material in which the fiber is made of SiC and the matrix is ​​made of SiC.

[0142] Second Embodiment

[0143] 7C is a flowchart showing the fiber composite material inspection program PB3 of the second embodiment. Each process of the fiber composite material inspection program PB3 corresponds to each step of the fiber composite material inspection method of the second embodiment.

[0144] In the second embodiment of the fiber composite material inspection program PB3, an inspection object image acquisition process S31, a fiber cross section position acquisition process S32, an array feature value acquisition process S33, an array state quantity measurement process S34, and a defect location detection section process S35 are executed.

[0145] The inspection object image acquisition unit 110, the fiber cross section position acquisition unit 170, the array feature value acquisition unit 180, the array state quantity measurement unit 190, and the defect location detection unit 200 respectively execute an inspection object image acquisition process S31, a fiber cross section position acquisition process S32, an array feature value acquisition process S33, an array state quantity measurement process S34, and a defect location detection unit process S35.

[0146] (Inspection object image acquisition process S31)

[0147] The inspection object image acquisition unit 110 acquires an inspection object image of a fiber composite material.

[0148] The inspection object image acquisition process S31 can be performed in the same manner as the inspection object image acquisition process S11 in the first embodiment. Note that the inspection object image of the fiber composite material may be acquired by any image acquisition method other than the image acquisition method shown in the first embodiment.

[0149] (Fiber cross section position acquisition process S32)

[0150] The fiber cross section position acquisition unit 170 acquires the position P indicating the cross section of each fiber F from the acquired inspection target image.

[0151] The fiber cross section position acquisition process S32 can be performed in the same manner as the pixel area division process S12, the measurement process S13, the determination process S14, the uniformization process S15, and the fiber cross section position detection process S16 of the first embodiment. The fiber cross section position acquisition process S32 may also be performed in the same manner as the pixel area division process S22, the relative distance measurement process S23, the determination process S24, the uniformization process S25, and the fiber cross section position detection process S26 of the second embodiment.

[0152] Note that the fiber cross-section positions may be acquired by any method other than the method for acquiring the fiber cross-section positions shown in the first embodiment.

[0153] A comparative example will be described below, followed by a description of the array characteristic value acquisition process S33, the array state quantity measurement process S34, and the defect location detection process S35 of the second embodiment.

[0154] (Comparative Example)

[0155] A comparative example of the second embodiment will be described.

[0156] If the positions P of the fibers F in the fiber composite material can be identified and quantified as in the first embodiment, a trained model can be constructed by machine learning the relationship between the numerical information on the distribution of fiber positions (a set of positions (X, Y)) and the mechanical strength of the fiber composite material. Therefore, by inputting the distribution of fiber positions into the trained model, the mechanical strength of the fiber composite material can be output, and vice versa. For example, it is possible to determine the fiber distribution required to achieve a desired mechanical strength (e.g., a structure with appropriate fracture strength and ductility strength), and this can be used as a guideline for designing fiber composite materials.

[0157] However, in machine learning methods, the optimization procedure is often unclear, and there are often problems in ensuring the technical validity and reliability of the optimal solution. The second embodiment solves these problems.

[0158] (Array characteristic value acquisition process S33)

[0159] The arrangement characteristic value acquisition process S33 acquires an arrangement characteristic value that indicates the characteristics of the arrangement of the cross section of each fiber F.

[0160] 13A and 13B are diagrams illustrating the arrangement characteristic values ​​that indicate the characteristics of the arrangement of the cross sections of adjacent fibers F.

[0161] 13A shows the geometric relationship when adjacent fibers F are arranged in a triangle. In the triangle with each fiber F as a vertex, half the length of the longest side LN1 is N b1 Let the radius of the circumscribed circle be N d1 Let N b1 and N d1 If the ratio of is √3 to 2, each fiber F is arranged in an equilateral triangle, which can be evaluated as an ideal arrangement of the fibers F. However, b1 and N d1 If the ratio deviates from the reference value of √3:2, the fibers F are arranged in a manner that is deviated from the equilateral triangle, and it can be evaluated that the arrangement is in a disordered state according to the amount of deviation.

[0162] where N b1 and N d1Using this, the sequence feature value score3 is defined as in the following equation (1).

[0163]

[0164] ...(1)

[0165] That is, when the fiber F is in an ideal arrangement state,

[0166] N b1 / N d1 = √3 / 2 = sin(π / 3)

[0167] The sequence characteristic value score3 shown in the above formula (1) becomes 1.

[0168] However, when each fiber F is in a disordered arrangement, N b1 / N d1 is a value greater than the reference value sin(π / 3). When it is a right-angled triangle, it takes the maximum value of 1, and the sequence feature value score3 shown in the above formula (1) becomes 0. In this way, score3 takes a value ranging from 0 to 1 depending on the amount of deviation.

[0169] Similarly, Figure 13(B) shows the geometric relationship when adjacent fibers F are arranged in a quadrangle. The longest side of the quadrangle is represented by LN1, and the shorter of the diagonals of the quadrangle is represented by LN2. The quadrangle with each fiber F as a vertex is divided into two triangles by the diagonal LN2, and attention is paid to the triangle containing the side LN1. Half the length of the side LN1 is N b1 , the radius of the circumscribing circle of the triangle is N d1 Then, N b1 and N d1 If the ratio of is 1 to √2, each fiber F is arranged in a square, which can be evaluated as an ideal arrangement of the fibers F. However, b1 and N d1 If the ratio of the fibers F deviates from the standard value of 1 to √2, the fibers F are arranged in a displaced manner from the square, and it can be evaluated that the arrangement is in a disordered state according to the amount of deviation. b2 is the longest side N b1 Anything with a length equal to or greater than this is no longer considered a quadrilateral, but two independent triangles.

[0170] Here, similarly to the above formula (1), the sequence characteristic value score4 is defined as in the following formula (2).

[0171]

[0172] ...(2)

[0173] That is, when the fiber F is in an ideal arrangement state,

[0174] N b1 / N d1 = 1 / √2 = sin(π / 4), and N b1 / N b2 = 1 / √2 = sin(π / 4),

[0175] Therefore, score4 as the sequence characteristic value shown in the above formula (2) becomes 1.

[0176] However, when each fiber F is in a disordered arrangement, N b1 / N d1 MoN b1 / N b2 is also greater than the reference value sin(π / 4). b1 = N b2 In this case, the sequence characteristic value score4 shown in the above formula (2) is 0. In this way, score4 takes a value in the range from 0 to 1 depending on the deviation amount. When each fiber F in FIG. 13(B) forms a square, that is, when N b1 >N b2 When this is the case, of the two triangles generated by dividing along the shorter diagonal line LN2, the array feature value score3 of the triangle containing the side LN1 is defined as follows:

[0177]

[0178] ...(3)

[0179] The array feature value score3 of the remaining triangles is defined in the same way as in the above-mentioned formula (1).

[0180] The array feature value acquisition process S33 acquires array feature values ​​score3,i or score4s,i for each location i of the inspection target image 310 shown in Figure 8A (C), for example, as shown in the following equations (4), (5), or (6).

[0181]

[0182] ...(4)

[0183]

[0184] ...(5)

[0185]

[0186] ...(6)

[0187] The above formulas (4) and (5) are both array feature values ​​for a triangle. If the triangle is part of a quadrangle and includes the shorter diagonal and the longest side of the quadrangle, formula (5) is applied; otherwise, formula (4) is applied.

[0188] (Array state quantity measurement process S34)

[0189] The array state quantity measurement unit 190 measures the array state quantity score1 or score2 according to the acquired array feature value score3,i or score4,i.

[0190] 15A illustrates an example of an inspection target image 500. The inspection target image 500 is divided into, for example, nine areas AR1 to AR9.

[0191] For each of the regions AR1 to AR9, the array state quantity score1 is calculated by the following equation (7).

[0192]

[0193] ... (7)

[0194] In the above formula (7), n is the number of triangular fiber F arrangements in each of the regions AR1 to AR9. The notation "or" in the right-hand side numerator conforms to the selection between formulas (4) and (5).

[0195] The smaller the arrangement state quantity score1, the more the fibers F are arranged deviating from the ideal equilateral triangle, and the more the fibers F are evaluated to be arranged in a disordered manner.

[0196] Similarly, for each of the regions AR1 to AR9, the array state quantity score2 is calculated using the following equation (8).

[0197]

[0198] ...(8)

[0199] The smaller the arrangement state quantity score2, the more the fibers F are arranged that they deviate from an ideal square, and the more the fibers F are evaluated to be arranged in a disordered manner.

[0200] (Defective part detection unit processing S35)

[0201] The defect detecting unit 200 detects a defect in the fiber composite material by associating the alignment state quantity score1 or score2 with the cross-sectional position P of the fiber F.

[0202] FIG. 15B illustrates the array state quantity score2 calculated for each of the nine areas AR1 to AR9 of the inspection target image 500.

[0203] Of the regions AR1 to AR9 in the inspection target image 500, the alignment state quantity score2 in region AR8 is the smallest, and it can be quantitatively confirmed that the alignment of the fibers F is greatly disturbed.

[0204] It is possible to estimate a location indicating a defect in the fiber composite material by associating the coordinate position P (X, Y) of the fiber F with the arrangement state quantity score1 or score2. For example, as shown in Fig. 15(A) , the coordinate position P (X, Y) of the fiber F in the inspection target image 500 that shows the smallest value of the arrangement state quantity score1 or score2 can be estimated as the location P where a crack has occurred.

[0205] (Example of the second embodiment)

[0206] Next, an example of the above-mentioned sequence characteristic value acquisition process S33, sequence state quantity measurement process S34, and defect location detection process S35 will be described below. In this example, topological data analysis is applied to a set of positions of each fiber F.

[0207] In this embodiment, the cross section of each fiber F is regarded as a circle with a radius r, and an alignment state quantity indicating the alignment state of the cross section of each fiber F is obtained by performing image processing to enlarge r.

[0208] Figures 16(A) and (B) provide a simple explanation of how to check the first-order persistent homology of point cloud data on a two-dimensional plane. A quadrangle (b1, d1) contains a triangle (b2, d2). When there is such a relationship, the quadrangle is called the "parent" and the triangle is called the "child."

[0209] FIG. 16A is a diagram showing the transition of the arrangement state of the fibers F when the horizontal axis represents the radius r of the cross section of the fibers F.

[0210] The upper part of Figure 16(A) shows the change in shape of the fiber cross section when the radius r of the cross section of four adjacent fibers F in the inspection target image 310 shown in Figure 8(C), for example, is gradually enlarged.

[0211] The lower part of Figure 16(A) shows the changes in the creation and disappearance of holes h surrounded by the fiber cross sections when the radius r of the cross sections of four adjacent fibers F in the inspection target image 310 shown in Figure 8A(C) is gradually enlarged, for example.

[0212] It is assumed that the radius r of the cross section of four adjacent fibers F in the inspection target image 310 is 0 in the initial state.

[0213] 16(A), when the cross-sectional radius r of fiber F is gradually enlarged, holes h are generated when the fiber cross-sectional radius r is b1 or b2, and the holes h disappear when the fiber cross-sectional radius r is d1 or d2. That is, a quadrangle h is generated when b1, a triangle h is generated when b2, the triangle h disappears when d2, and another triangle h disappears when d1.

[0214] For each hole h, a pair (b, d) is obtained, which is the fiber cross-sectional radius b when the hole h is generated and the fiber cross-sectional radius d when the hole h disappears. The pair (b, d) defines the alignment state quantity of the fibers F. The alignment state of the fibers F can be evaluated according to the alignment state quantity (b, d).

[0215] FIG. 16B shows a persistence diagram.

[0216] The persistence diagram is a graph in which the horizontal axis represents the fiber cross-sectional radius b when the hole h is generated and the vertical axis represents the fiber cross-sectional radius d when the hole h disappears, and the alignment state quantities (b, d) obtained for each hole h are plotted.

[0217] That is, when the fibers F are positioned close to each other, holes h are generated at a small b value and disappear at a small d value. Conversely, when the fibers F are positioned farther apart, holes h are generated at a large b value and disappear at a large d value.

[0218] 14(A) and (B) show the change in the cross-sectional shape of each fiber F arranged in a triangle as shown in FIG. 13(A).

[0219] As shown in FIG. 14(A), the radius of the cross section of the fiber F is half the length of the longest side of the triangle, that is, the length N shown in FIG. b1 When this occurs, a hole h is generated.

[0220] As shown in FIG. 14(B), the radius of the cross section of the fiber F is the radius of the circumscribing circle of the triangle, that is, the length N shown in FIG. d1 When this occurs, the hole h disappears.

[0221] If the fibers F are arranged in an ideal equilateral triangle, the alignment state quantities (b, d) are uniquely determined, and even when the allowable range of the actual fiber alignment is taken into consideration, the alignment state quantities (b, d) are considered to fall within a predetermined allowable range. The same applies to the square-shaped fiber alignment shown in Figure 13(B).

[0222] Returning to Fig. 16(B), if the fibers F are arranged in an ideal state or in a state close to the ideal state, the arrangement state quantities (b, d) will fall within a predetermined region on the persistence diagram shown in Fig. 16(B). However, if there is a defect in the fiber arrangement on the inspection target image 310, such as if the fibers F are arranged in a disordered manner, the arrangement state quantities (b, d) will exist in a region outside the predetermined region on the persistence diagram.

[0223] Figure 17(A) shows the results of performing the same processing as in Figures 16(A) and (B) on the entire fiber array of the inspection target image 310 shown in Figure 8A(C), for example, to obtain the array state quantities (b, d) and plotting them on a persistence diagram.

[0224] 17A, the horizontal axis represents the number of brightness steps Nb until the hole h is generated, and the vertical axis represents the number of brightness steps Ns until the hole h disappears. The fiber cross-sectional radius r is gradually enlarged for each step.

[0225] The shading of the plotted points in FIG. 17(A) corresponds to the frequency of the array state quantity (b, d).

[0226] For example, region IA shown in Figure 17(A) indicates the allowable range when the size of the fibers F is taken into consideration. If the alignment state quantities (b, d) exist in the overlapping portion of region IA and region Q1 (shown in box Q1), the alignment state of the corresponding fibers F is determined to be an ideal triangular alignment state (equilateral triangle) or an alignment state H1 close to the ideal alignment state, as shown in Figure 17(B). Furthermore, if the alignment state quantities (b, d) exist in the overlapping portion of region IA and region Q2 (shown in box Q2), the alignment state of the fibers F is determined to be an ideal quadrangular alignment state (square) or an alignment state H2 close to the ideal alignment state, as shown in Figure 17(C).

[0227] It should be noted that the equilateral triangle as the "child" of the quadrangle cannot be determined solely from the position of the array state quantity (b, d).

[0228] However, if the arrangement state quantity (b, d) exists at a coordinate point outside the area IA, the arrangement state of each fiber F is determined to be a disordered arrangement state, which may be defective or has a defect.

[0229] The values ​​of Nb1, Nd1, Nb2, and Nd2 shown in Fig. 13 correspond to b1, d1, b2, and d2 described in Fig. 16(A). Therefore, by using information obtained from the persistent homology of point clouds, including parent-child relationships, in this example, the method described in the second embodiment can be easily applied to an actually measured image.

[0230] The defect detecting unit 200 associates the acquired alignment state quantity (b, d) with the cross-sectional position P of the fiber F to detect a location indicating a defect in the fiber composite material.

[0231] In topological data analysis using persistent homology, each coordinate point (b, d) in the persistence diagram can be associated with the cross-sectional position P of the fiber F in the original image by inverse analysis. Fig. 4 illustrates the result of associating each coordinate point (b, d) in the persistence diagram shown in Fig. 17(A) with the cross-sectional position P of the fiber F in the original inspection target image 700 by inverse analysis.

[0232] For example, if the alignment state quantity (b, d) of fiber F is within region IA of the persistence diagram shown in Figure 17 (A), then a display Fi (shown by a ◆ in the figure) indicating an ideal alignment state is associated with the corresponding position Pi of fiber F, as shown in Figure 4.

[0233] However, if the alignment state quantity (b, d) of fiber F is outside the region IA of the persistence diagram shown in Figure 17(A), a display Fd (indicated by * in the figure) indicating a disordered alignment state is associated with the position Pd of the corresponding fiber F, as shown in Figure 4.

[0234] 4 is displayed on the display screen of the display device 17 of the inspection-side computer terminal 10. Therefore, the user can easily check on the display screen of the display device 17 the locations of fibers Fd with a disordered arrangement where cracks are likely to occur (indicated by * in the figure) and the boundary locations between fibers Fd (indicated by * in the figure) and fibers Fi (indicated by ◆ in the figure) where there is a large change in arrangement.

[0235] In the above embodiments, a fiber composite material containing fibers, a matrix, and another fiber has been described as a composite material. However, the present disclosure can also be applied to a particle composite material containing particles and a matrix. That is, the fibers in the embodiments can be replaced with particles. The particle shape can be any shape, such as spherical. The particle size is also not important. In addition, while the present disclosure uses an X-ray CT scanner to acquire an image of the composite material to be inspected, any means or method can be used to acquire an image of the composite material to be inspected, as long as it can acquire a cross-section of the fibers or particles of the composite material. For example, a scanning electron microscope (SEM) or a transmission electron microscope (TEM) can also be used to acquire an image of the composite material to be inspected. Even when using a scanning electron microscope (SEM) or a transmission electron microscope (TEM), the fibers or particles in the microstructure of the composite material can be measured, inspected, and evaluated in the same manner as when using an X-ray CT scanner. While specific embodiments have been described above, the described embodiments are merely examples and do not limit the scope of the invention. The novel methods and apparatus described herein may be embodied in a variety of other forms. Furthermore, various omissions, substitutions, and modifications may be made in the methods and apparatus described herein without departing from the spirit of the invention. The appended claims and their equivalents encompass such various forms and modifications as falling within the scope and spirit of the invention. For example, although a machine learning approach is described as a comparative example of an embodiment, the application of machine learning is not excluded from the scope of the invention.

[0236] The present disclosure is suitable for the manufacturing, molding, quality control, optimization of manufacturing conditions, and prediction of fracture and deterioration during use of composite materials.

[0237] REFERENCE SIGNS LIST 100 Fiber composite material inspection device 110 Inspection object image acquisition unit 120 Pixel area division unit 130 Measurement unit 135 Relative distance measurement unit 140 Determination unit 150 Fiber cross section position detection unit 160 Homogenization processing unit 170 Fiber cross section position acquisition unit 180 Arrangement feature value acquisition unit 185 Arrangement state quantity acquisition unit 190 Arrangement state quantity measurement unit 200 Defect location detection unit

Claims

1. A composite material inspection device comprising: an inspection object image acquisition unit that acquires an inspection object image of a composite material; a pixel area division unit that divides the acquired inspection object image into a low-density pixel area and a high-density pixel area; a measurement unit that measures the number of steps until a high-density image area is absorbed into another high-density image area by performing image processing that sequentially expands the high-density image area for each step; a determination unit that determines whether a low-density high-density pixel area exists within a high-density pixel area depending on whether the measured number of steps is equal to or greater than a predetermined threshold; and a fiber cross section / particle cross section position detection unit that detects the cross section positions of fibers or particles based on the determination result of the determination unit.

2. The composite material inspection device according to claim 1, wherein the determination unit further determines that the high-density pixel region and the other high-density pixel region correspond to a fiber or particle cross section and a base material, respectively, based on the number of measured steps.

3. The composite material inspection device according to claim 1, further comprising a homogenization processing unit that homogenizes pixel values ​​in areas where a high-density pixel region and another high-density pixel region exist within the same fiber or particle cross section.

4. A composite material inspection device comprising: an inspection object image acquisition unit that acquires an inspection object image of the composite material; a fiber cross section / particle cross section position acquisition unit that acquires the position indicating each fiber or particle cross section in the acquired inspection object image; an arrangement characteristic value acquisition unit that acquires an arrangement characteristic value that indicates the arrangement characteristic of each fiber or particle cross section; an arrangement state quantity measurement unit that measures an arrangement state quantity according to the deviation of the acquired arrangement characteristic value from a reference value; and a defect location detection unit that associates the arrangement state quantity with the fiber or particle cross section position to detect locations indicating defects in the composite material.

5. A composite material inspection device comprising: an inspection object image acquisition unit that acquires an inspection object image of the composite material; a fiber cross section / particle cross section position acquisition unit that acquires the position indicating each fiber or particle cross section in the acquired inspection object image; an arrangement state quantity acquisition unit that acquires an arrangement state quantity indicating the arrangement state of each fiber or particle cross section by performing image processing to enlarge each fiber or particle cross section; and a defect location detection unit that detects locations indicating defects in the composite material by correlating the acquired arrangement state quantity with the fiber or particle cross section position.

6. A method for inspecting composite materials, comprising: an inspection object image acquisition step for acquiring an inspection object image of the composite material; a pixel area division step for dividing the acquired inspection object image into a low-density pixel area and a high-density pixel area; a measurement step for measuring the number of steps until a high-density image area is absorbed into another high-density image area by performing image processing to sequentially expand the high-density image area for each step; a determination step for determining whether a low-density high-density pixel area exists within a high-density pixel area depending on whether the measured number of steps is equal to or greater than a predetermined threshold; and a fiber cross section / particle cross section position detection step for detecting the cross section positions of fibers or particles based on the determination result.

7. A method for inspecting a composite material, comprising: an inspection object image acquisition step for acquiring an inspection object image of the composite material; a fiber cross section / particle cross section position acquisition step for acquiring the position indicating each fiber or particle cross section in the acquired inspection object image; an arrangement feature value acquisition step for acquiring an arrangement feature value indicating the arrangement characteristics of each fiber or particle cross section; an arrangement state quantity measurement step for measuring an arrangement state quantity corresponding to the deviation of the acquired arrangement feature value from a reference value; and a defect location detection step for detecting locations indicating defects in the composite material by associating the arrangement state quantity with the fiber or particle cross section position.

8. A method for inspecting a composite material, comprising: an inspection object image acquisition step for acquiring an inspection object image of the composite material; a fiber cross section / particle cross section position acquisition step for acquiring the position indicating each fiber or particle cross section in the acquired inspection object image; an arrangement state quantity acquisition step for acquiring an arrangement state quantity indicating the arrangement state of each fiber or particle cross section by performing image processing to enlarge each fiber or particle cross section; and a defect location detection step for detecting locations indicating defects in the composite material by correlating the acquired arrangement state quantity with the fiber or particle cross section position.

9. A composite material inspection program for causing a computer to execute the following: an inspection object image acquisition process for acquiring an inspection object image of a composite material; a pixel area division process for dividing the acquired inspection object image into a low-density pixel area and a high-density pixel area; a measurement process for measuring the number of steps until a high-density image area is absorbed into another high-density image area by performing image processing to sequentially expand the high-density image area at each step; a determination process for determining whether a low-density high-density pixel area exists within a high-density pixel area depending on whether the measured number of steps is equal to or greater than a predetermined threshold; and a fiber cross section / particle cross section position detection process for detecting the cross section positions of fibers or particles based on the determination result.

10. A composite material inspection program for causing a computer to execute the following: an inspection object image acquisition process for acquiring an inspection object image of a composite material; a fiber cross section / particle cross section position acquisition process for acquiring the position indicating each fiber or particle cross section in the acquired inspection object image; an array feature value acquisition process for acquiring array feature values ​​indicating the characteristics of the array of each fiber or particle cross section; an array state quantity measurement process for measuring an array state quantity corresponding to the deviation of the acquired array feature value from a reference value; and a defect location detection process for detecting locations indicating defects in the composite material by associating the array state quantity with the fiber or particle cross section position.

11. A composite material inspection program for causing a computer to execute the following: an inspection object image acquisition process for acquiring an inspection object image of a composite material; a fiber cross section / particle cross section position acquisition process for acquiring the position indicating each fiber or particle cross section in the acquired inspection object image; an arrangement state quantity acquisition process for acquiring an arrangement state quantity indicating the arrangement state of each fiber or particle cross section by performing image processing to enlarge each fiber or particle cross section; and a defect location detection process for detecting locations indicating defects in the composite material by correlating the acquired arrangement state quantity with the fiber or particle cross section position.

12. A computer-readable storage medium having recorded thereon a composite material inspection program for causing a computer to execute: an inspection object image acquisition process for acquiring an inspection object image of a composite material; a pixel area division process for dividing the acquired inspection object image into a low-density pixel area and a high-density pixel area; a measurement process for measuring the number of steps until a high-density image area is absorbed into another high-density image area by performing image processing to sequentially expand the high-density image area at each step; a determination process for determining whether a low-density high-density pixel area exists within a high-density pixel area depending on whether the measured number of steps is equal to or greater than a predetermined threshold; and a fiber cross section / particle cross section position detection process for detecting the cross section positions of fibers or particles based on the determination results.

13. A computer-readable storage medium having recorded thereon a composite material inspection program for causing a computer to execute the following: an inspection object image acquisition process for acquiring an inspection object image of a composite material; a fiber cross section / particle cross section position acquisition process for acquiring the position indicating each fiber or particle cross section in the acquired inspection object image; an array feature value acquisition process for acquiring array feature values ​​indicating the characteristics of the array of each fiber or particle cross section; an array state quantity measurement process for measuring an array state quantity corresponding to the deviation of the acquired array feature value from a reference value; and a defect location detection process for correlating the array state quantity with the fiber or particle cross section position to detect locations indicating defects in the composite material.

14. A computer-readable storage medium having recorded thereon a composite material inspection program for causing a computer to execute: an inspection object image acquisition process for acquiring an inspection object image of a composite material; a fiber cross section / particle cross section position acquisition process for acquiring the position indicating each fiber or particle cross section in the acquired inspection object image; an arrangement state quantity acquisition process for acquiring an arrangement state quantity indicating the arrangement state of each fiber or particle cross section by performing image processing to enlarge each fiber or particle cross section; and a defect location detection process for detecting locations indicating defects in the composite material by correlating the acquired arrangement state quantity with the fiber or particle cross section position.

Images