Fracture surface image analysis device and fracture surface image analysis method
By segmenting and processing the fracture surface image, using the machine learning model to derive the crack travel direction and starting point position, the problem of insufficient analytical accuracy in the prior art is solved, and high-precision fracture surface image analysis is achieved.
Patent Information
- Application Number
- CN202411900780.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2024-01-05
- Filing Date
- 2024-12-23
- Publication Date
- 2025-07-08
AI Technical Summary
In the prior art, when analyzing the image of the crack surface of a component, it is difficult to determine the starting point of the crack with high accuracy, resulting in insufficient analytical accuracy.
The image analysis device of the fracture surface is used to segment the fracture surface images through image processing technology, and multiple crack loss travel directions are derived, and the crack loss starting point position is derived based on these directions. The analytical accuracy is improved by using machine learning models and deep learning technology.
It realizes high-precision analysis of the crack starting point of the rupture surface, and supports more accurate component design changes and fault analysis.
Smart Images
Figure CN120278942A_ABST
Abstract
Description
Technical Field
[0001] Embodiments of the present invention relate to a fracture surface image analysis device and a fracture surface image analysis method. Background Art
[0002] For example, a component is analyzed based on an image of a fracture surface of the component or the like. Higher-precision analysis is desired.
[0003] Prior Art Documents:
[0004] Patent Documents:
[0005] Patent Document 1: Japanese Patent No. 6789460 Summary of the Invention
[0006] Problems to be Solved by the Invention:
[0007] Embodiments provide a fracture surface image analysis device and a fracture surface image analysis method capable of performing highly accurate analysis.
[0008] Means for Solving the Problems:
[0009] According to an embodiment, a fracture surface image analysis device includes: an acquisition unit configured to acquire a first image including a fracture surface of a component; and a processing unit configured to perform image analysis on the first image. The image analysis includes a first process and a second process. The first process includes: deriving a plurality of crack propagation directions in the fracture surface. One of the plurality of crack propagation directions corresponds to one of a plurality of positions included in the fracture surface. The second process includes: deriving a crack start position in the fracture surface based on at least a part of the plurality of crack propagation directions. Brief Description of the Drawings
[0010] Figure 1 is a flowchart illustrating the operation of the fracture surface image analysis device according to the first embodiment.
[0011] Figure 2 is a schematic diagram illustrating the fracture surface image analysis device according to the first embodiment.
[0012] Figure 3 is a schematic diagram illustrating a part of the operation of the fracture surface image analysis device according to the first embodiment.
[0013] Figure 4 is a schematic diagram illustrating a part of the operation of the fracture surface image analysis device according to the first embodiment.
[0014] Figure 5 of (a) to Figure 5(f) is a schematic diagram illustrating the operation of the fracture surface image analysis device according to the first embodiment.
[0015] Figure 6 (a) of Figure 6 and (b) of are schematic diagrams illustrating the operation of the fracture surface image analysis device according to the first embodiment.
[0016] Figure 7 is a flowchart illustrating the operation of the fracture surface image analysis device according to the first embodiment.
[0017] Figure 8 is a schematic diagram illustrating the operation of the fracture surface image analysis device according to the first embodiment.
[0018] Figure 9 (a) of Figure 9 and (b) of are schematic diagrams illustrating the operation of the fracture surface image analysis device according to the first embodiment.
[0019] Figure 10 (a) of Figure 10 and (b) of are schematic diagrams illustrating the operation of the fracture surface image analysis device according to the first embodiment.
[0020] Figure 11 is a schematic diagram illustrating the operation of the fracture surface image analysis device according to the first embodiment.
[0021] Explanation of reference numerals:
[0022] 50: Camera device, 70: Processing unit, 70C: Reference coordinate, 70a, 70b: First and second processing parts, 71: Divider, 71D: First image, 71P: Small block area, 72: Model selector, 72A, 72B: First and second models, 72M: Model, 72a: Damage mode, 72b: Material, 72c: Shape, 72d: Molding condition, 74: Small block extractor, 75: Acquisition unit, 79a: GUI, 79b: Display, 79c: Input unit, 79d: Memory, 81: Component, 110: Fracture surface image analysis device, 210: Analysis device, Ar1, Ar2, As1: Arrows, CN1: Confidence level, D1 to D3: First to third directions, Dr1: Damage propagation direction, Dr2: Damage propagation direction after extraction, Dra1: Averaged damage propagation direction, E1 to E3: First to third endpoints, Ln1: Straight line, Ln2: Extension line, S1 to S3: First to third starting points, SP1: Damage starting point position, SPL1 to SPL3: First to third specimens. Detailed implementation manners
[0023] The following describes each embodiment of the present invention with reference to the drawings.
[0024] The drawings are schematic or conceptual, and the relationships between the thickness and width of each part, the ratio of the sizes between components, etc. are not necessarily the same as in reality. Even when representing the same part, the mutual dimensions or ratios may sometimes be shown differently according to the drawings.
[0025] In the specification and each figure of the present application, for elements that are the same as those described in the previously described drawings, the same reference signs are attached and detailed descriptions are appropriately omitted.
[0026] (First Embodiment)
[0027] Figure 1 It is a flowchart illustrating the operation of the fracture surface image analysis device according to the first embodiment.
[0028] Figure 2 It is a schematic diagram illustrating the fracture surface image analysis device according to the first embodiment.
[0029] As Figure 2 shown, the fracture surface image analysis device 110 according to the embodiment includes an acquisition unit 75 and a processing unit 70. The acquisition unit 75 is configured to acquire a first image 71D. The acquisition unit 75 may be, for example, an interface or the like. The processing unit 70 is configured to perform image analysis on the first image 71D. The fracture surface image analysis device 110 may be, for example, an image processing device.
[0030] The processing unit 70 may be, for example, an electronic circuit. The processing unit 70 may be, for example, a computer or the like. The processing unit 70 may include, for example, a CPU (Central Processing Unit) or the like. The processing unit 70 may include, for example, a GPU (Graphics Processing Unit) or the like.
[0031] As Figure 2 shown, the fracture surface image analysis device 110 may also include a GUI (Graphical User Interface) 79a. The fracture surface image analysis device 110 may also include a display 79b. The content as the object may be displayed on the display 79b through the GUI 79a. The fracture surface image analysis device 110 may also include an input unit 79c and a memory 79d, etc. The input unit 79c may include, for example, at least any one of a keyboard, a mouse, a touch panel, and a voice input device, etc. The memory 79d can store at least a part of the data used for processing or at least a part of the processing result.
[0032] The first image 71D obtained by the acquisition unit 75 includes the fracture surface of the component 81. The first image 71D can also be obtained, for example, from an imaging device 50 that images the fracture surface of the component 81. The location where the imaging device 50 is provided can be different from the location where the fracture surface image analysis device 110 is provided. The location where the imaging device 50 is provided can also be the same as the location where the fracture surface image analysis device 110 is provided. The analysis device 210 according to the embodiment can also include the fracture surface image analysis device 110 and the imaging device 50. The imaging device 50 can also be regarded as being included in the fracture surface image analysis device 110. Information related to the first image 71D can be supplied to the acquisition unit 75 by any method such as wired or wireless. For example, it can also be that information related to the first image 71D is stored in an arbitrary memory (which can be the memory 79d, etc.), and the stored information is supplied to the acquisition unit 75.
[0033] The first image 71D can also include, for example, a micrograph (such as an SEM: Scanning Electron Microscope image) of the fracture surface of the component 81. The component 81 can include, for example, at least one of resin and metal. The component 81 can also be a component included in various devices, etc. Components sometimes crack. By analyzing the fractured fracture surface, the cause of the crack can be determined, etc. The analysis of the fracture surface can also be carried out in the development stage, design stage, manufacturing stage, or after-sales stage of various devices.
[0034] As Figure 1 shown, the image processing performed by the processing unit 70 includes a first process (step S110) and a second process (step S120). The first process includes: deriving (for example, inferring) a plurality of crack propagation directions in the fracture surface. One of the plurality of crack propagation directions corresponds to one of the plurality of positions included in the fracture surface. As described later, the first image can also be divided into a plurality of small patch regions. Derive the crack propagation directions in each of the plurality of small patch regions. The plurality of small patch regions can be, for example, rectangles.
[0035] The second process includes: deriving the crack starting position in the fracture surface based on at least a part of the plurality of crack propagation directions. For example, by deriving the crack starting position, the mechanism of the crack can be derived with high precision. For example, by obtaining information related to the crack starting position, it is easy to obtain guidelines for design changes of the component 81 (parts, etc.) as the object.
[0036] For example, there is a reference example of inferring a crack pattern, etc. based on an image of the fracture surface. The pattern of the crack is, for example, ductile fracture, fatigue fracture, brittle fracture, solvent crack, or intergranular fracture, etc. In this reference example, even if the crack pattern can be determined, the position of the starting point of the crack cannot be derived. Therefore, the accuracy of the analysis is low.
[0037] In contrast, in an embodiment, the propagation direction of the crack is derived. That is, the propagation of the crack (the time change of the crack position) in the component 81 as the object is inferred. Thereby, the mechanism of the crack in the component 81 as the object can be grasped more accurately. According to the embodiment, an image analysis device capable of performing highly accurate analysis can be provided.
[0038] In the embodiment, the crack starting position is further derived. The crack starting position corresponds to, for example, the starting point of the occurrence of the crack in the component 81 as the object. By determining the crack starting position, the mechanism of the crack in the component 81 as the object can be derived with higher accuracy. Regarding the component 81 as the object, it is easy to perform a design change with higher accuracy. According to the embodiment, an image analysis device capable of performing more accurate analysis can be provided.
[0039] For example, as described later, one of the multiple crack propagation directions can be represented by an angle θ. The angle θ is 0 degrees or more and less than 360 degrees. One of the multiple crack propagation directions can also be represented by sin θ and cos θ. By processing the information (value group) related to the multiple crack propagation directions, the position that is the starting point of the multiple crack propagation directions can be derived. The derived position becomes the crack starting position.
[0040] As Figure 1 shown, in the embodiment, at least a part obtained by the analysis process can also be displayed through the GUI (step S130).
[0041] As Figure 2 shown, the processing unit 70 may also include a plurality of processing parts (the first processing part 70a, the second processing part 70b, etc.). The plurality of processing parts may correspond to a plurality of models, for example. The number of the plurality of processing parts is arbitrary.
[0042] In the embodiment, the first process may include: deriving one crack propagation direction in the fracture surface. The second process may include: deriving the crack starting position in the fracture surface based on the crack propagation direction. In the second process, for example, other information (for example, known information about the crack propagation direction and the crack starting position, etc.) may be used to derive the crack starting position.
[0043] Hereinafter, an example of the first process will be described.
[0044] Figure 3 is a schematic diagram illustrating a part of the operation of the fracture surface image analysis device according to the first embodiment.
[0045] As Figure 3As shown, for example, the first image 71D obtained by the acquisition unit 75 is supplied to the processing unit 70. For example, the processing unit 70 includes a splitter 71 and a first processing section 70a. The splitter 71 divides the first image 71D into a plurality of small block regions 71P. Information related to the images included in each of the plurality of small block regions 71P obtained by the division is supplied to the first processing section 70a. In the first processing section 70a, the crack propagation direction Dr1 in each of the plurality of small block regions 71P is derived.
[0046] The first processing section 70a derives the crack propagation direction Dr1, for example, through a processing model based on deep learning. The first processing section 70a performs processing based on machine learning, for example. For example, the processing based on machine learning is performed using one of the plurality of models 72M. The plurality of models 72M are, for example, models completed by machine learning.
[0047] The first processing section 70a corresponds to a regression processing unit, for example. The regression processing unit is configured to perform processing based on machine learning that is related to a plurality of teacher images including the fracture surface of the component 81 and a plurality of crack propagation directions Dr1. The first processing includes the processing performed by such a regression processing unit (the first processing section 70a).
[0048] As Figure 3 shown, for example, at least any one of information related to the crack pattern 72a of the fracture surface, information related to the material 72b of the component 81, information related to the shape 72c of the component 81, and information related to the molding conditions 72d of the component 81 is supplied to the processing unit 70. The processing unit 70 includes a model selector 72. The above-mentioned information is supplied to the model selector 72.
[0049] The model selector 72 selects one of the plurality of models 72M to be used in the processing based on the crack pattern 72a, material 72b, shape 72c, molding conditions 72d, etc. Information related to the selected model 72M is supplied to the first processing section 70a. The first processing section 70a uses the selected model 72M to derive the crack propagation direction Dr1 in each of the plurality of small block regions 71P.
[0050] Like this, the processing unit 70 may also include multiple models 72M. The multiple models 72M include a first model 72A, a second model 72B, and so on. For example, in one action (the first action), the processing unit 70 uses one of the multiple models 72M (the first model 72A) to perform the first processing. In another action (the second action), the processing unit 70 uses another one of the multiple models 72M (the second model 72B) to perform the first processing. Between one of the multiple models 72M and another one of the multiple models 72M, at least one of the crack damage pattern 72a of the fracture surface, the material 72b of the component 81, the shape 72c of the component 81, and the molding condition 72d of the component 81 is different.
[0051] Based on the difference in at least one of the crack damage pattern 72a of the fracture surface, the material 72b of the component 81, the shape 72c of the component 81, and the molding condition 72d of the component 81, an appropriate one of the multiple models 72M is selected. By using the appropriate model, more accurate processing can be performed.
[0052] The first processing part 70a uses the selected one of the multiple models 72M to derive crack propagation directions Dr1 respectively related to the multiple small block regions 71P. Like this, the first processing includes: for one of the multiple small block regions 71P obtained by dividing the first image 71D acquired by the acquisition unit 75, deriving one of the multiple crack propagation directions Dr1.
[0053] The processing unit 70 outputs at least a part of the derived multiple crack propagation directions Dr1.
[0054] The processing unit 70 may also output a confidence level CN1 respectively related to the multiple crack propagation directions Dr1. Like this, the first processing may also include: deriving a confidence level CN1 respectively related to the derived multiple crack propagation directions Dr1. As described later, the processing unit 70 may also perform the second processing according to the confidence level CN1. The confidence level CN1 regarding one of the multiple crack propagation directions Dr1 may be derived from two or more directions among the multiple crack propagation directions Dr1.
[0055] Hereinafter, an example of the second processing will be described.
[0056] Figure 4 It is a schematic diagram illustrating a part of the operation of the fracture surface image analysis device according to the first embodiment.
[0057] Figure 4 It shows an example of the second processing. As Figure 4As shown, the processing unit 70 may also include a patch extractor 74 and a second processing section 70b. For example, a plurality of crack propagation directions Dr1 and confidence levels CN1 are supplied to the patch extractor 74. Based on the confidence level CN1, the patch extractor 74 extracts at least a part of the plurality of patch regions 71P used when deriving the crack starting position SP1. Alternatively, based on the confidence level CN1, the patch extractor 74 extracts at least a part of the plurality of patch regions 71P not used when deriving the crack starting position SP1. Through the operation of the patch extractor 74, a part of the plurality of crack propagation directions Dr1 used when deriving the crack starting position SP1 (the extracted crack propagation direction Dr2) is extracted.
[0058] For example, there are a plurality of crack propagation directions Dr1 (extracted crack propagation direction Dr2) having a confidence level CN1 equal to or higher than a determined reference value. The second processing section 70b uses the extracted crack propagation direction Dr2 to derive the crack starting position SP1.
[0059] On the other hand, for example, the confidence level CN1 of at least one direction among the plurality of crack propagation directions Dr1 is less than the determined reference value. The second processing section 70b does not use at least one of such a plurality of crack propagation directions Dr1 and derives the crack starting position SP1. The confidence level CN1 corresponding to the at least one direction not used among the plurality of crack propagation directions Dr1 is less than the determined reference value.
[0060] By not using the data with a low confidence level CN1, the crack starting position SP1 can be derived with higher accuracy.
[0061] When deriving the crack starting position SP1, the determined coordinates (reference coordinates 70C) may also be used. As already described, for example, a first image 71D is obtained from an imaging device 50 that images the fracture surface of the component 81 (refer to Figure 2 ). The reference coordinates 70C are, for example, the reference coordinates at the time of imaging by the imaging device 50. The reference coordinates 70C at the time of imaging by the imaging device 50 may also be, for example, the coordinates set for the workbench on which the imaging device 50 is provided. The second processing may also include: using the common coordinates (reference coordinates 70C) related to the plurality of crack propagation directions Dr1 to derive the crack starting position SP1. By processing the plurality of crack propagation directions Dr1 using the common coordinates (reference coordinates 70C), the crack starting position SP1 can be derived with high accuracy.
[0062] Figure 5 of (a) to Figure 5 of (f) are schematic diagrams illustrating the operation of the fracture surface image analysis device according to the first embodiment.
[0063] Figure 5 of (a) andFigure 5 (b) relates to the first specimen SPL1. Figure 5 (c) and Figure 5 (d) relate to the second specimen SPL2. Figure 5 (e) and Figure 5 (f) relate to the third specimen SPL3. Figure 5 (a), Figure 5 (c) and Figure 5 (e) correspond to the first image 71D of these specimens (fracture surfaces). Figure 5 (b) shows a plurality of crack propagation directions Dr1 derived from the first image 71D of Figure 5 (a). Figure 5 (d) shows a plurality of crack propagation directions Dr1 derived from the first image 71D of Figure 5 (c). Figure 5 (f) shows a plurality of crack propagation directions Dr1 derived from the first image 71D of Figure 5 (e).
[0064] For example, the first image 71D is divided into a plurality of small block regions 71P (see Figure 3 ). In each of the divided small block regions 71P, one of the plurality of crack propagation directions Dr1 is derived by the first processing section 70a based on the machine learning model.
[0065] In this example, the first specimen SPL1 corresponds to ductile fracture. The second specimen SPL2 corresponds to fatigue fracture. The third specimen SPL3 corresponds to brittle fracture. These different types of crack patterns correspond to the crack pattern 72a (see Figure 3 ). According to the difference in the crack pattern, one of the plurality of models 72M is selected, and a plurality of crack propagation directions Dr1 are derived.
[0066] Figure 6 (a) and Figure 6 (b) are schematic diagrams illustrating the operation of the fracture surface image analysis device according to the first embodiment.
[0067] Figure 6 (a) illustrates the first image 71D. Figure 6 (b) illustrates a plurality of crack propagation directions Dr1 derived from the first image 71D of Figure 6 (a), and the crack origin position SP1 derived from the plurality of crack propagation directions Dr1.
[0068] As Figure 6As shown in (b) of , a plurality of crack propagation directions Dr1 are derived for the first image 71D. In this example, the plurality of crack propagation directions Dr1 are represented by arrows Ar1. For example, if one arrow Ar1 is selected and there is "another arrow Ar2" on the starting side of this arrow Ar1, then this "another arrow Ar2" is set as the new arrow. Regarding the new arrow, if there is "yet another arrow" on the starting side of the new arrow, then this "yet another arrow" is set as the new arrow. By repeating such an operation, for the initially selected arrow Ar1, the starting arrow As1 is determined. Such an operation is performed on the plurality of crack propagation directions Dr1 (a plurality of arrows) arranged two-dimensionally. By repeating the operation, a plurality of starting arrows As1 are derived. The intersection point of the extension lines of each of the plurality of arrows As1 can be inferred as the crack starting position SP1.
[0069] Figure 7 It is a flowchart exemplifying the operation of the fracture surface image analysis device according to the first embodiment.
[0070] Figure 7 Exemplify the second process (step S120). In this example, the plurality of crack propagation directions Dr1 are represented as a plurality of arrows. For example, as one of the plurality of crack propagation directions Dr1, one of the plurality of arrows is selected (for example, arrow Ar1) (step S121). It is determined whether there is another arrow (for example, arrow Ar2) on the starting side of the selected arrow Ar1 (step S122). If there is another arrow (for example, arrow Ar2), move to the other arrow Ar2 (step S123). If there is no other arrow Ar2 in step S122, transfer to step S124.
[0071] In step S124, it is determined whether there are remaining arrows. If there are remaining arrows, return to step S121 to select an arrow. Through step S121 and step S122, the starting arrow As1 for the initially selected one arrow (for example, arrow Ar1) among the plurality of arrows is derived (step S128).
[0072] By repeatedly performing the process including step S121, step S122, step S123, and step S124, for the plurality of crack propagation directions Dr1 arranged two-dimensionally, a plurality of starting arrows As1 are derived.
[0073] If there are no remaining arrows in step S124, transfer to step S125. In step S125, the intersection point of the extension lines of the plurality of starting arrows As1 is calculated. The intersection point is output, for example, as the crack starting position SP1.
[0074] Figure 8This is a schematic diagram illustrating the operation of the fracture surface image analysis device according to the first embodiment.
[0075] Figure 8 This shows an example of a method for deriving the crack start position SP1 based on multiple crack propagation directions Dr1. As Figure 8 shown, the multiple crack propagation directions Dr1 include a first direction D1, a second direction D2, and a third direction D3. The first direction D1 is the direction from the first start point S1 to the first end point E1. The second direction D2 is the direction from the second start point S2 to the second end point E2. The third direction D3 is the direction from the third start point S3 to the third end point E3.
[0076] The position of the second end point E2 on the straight line Ln1 along the first direction D1 is located between the position of the second start point S2 on the straight line Ln1 and the position of the first end point E1 on the straight line Ln1. The position of the first start point S1 on the straight line Ln1 is located between the position of the second end point E2 on the straight line Ln1 and the position of the first end point E1 on the straight line Ln1.
[0077] The first direction D1 corresponds to, for example, the arrow Ar1. The second direction D2 corresponds to another arrow Ar2. The second direction D2 can be derived based on the first direction D1. The second direction D2 (the other arrow Ar2) is, for example, the direction closest to the first direction D1 among other directions (other arrows). The angle between the second direction D2 and the straight line Ln1 is smaller than the angle between other directions (other arrows) and the straight line Ln1. Such a second direction D2 can be determined with respect to the first direction D1.
[0078] For example, the second process includes repeatedly performing the start point derivation process. One process in the start point derivation process includes: determining the second direction D2 based on the first direction D1. The third direction D3 corresponds to the second direction D2 determined by repeatedly performing the start point derivation process. In the second process, the position of the third direction D3 on the extension line Ln2 in the direction from the third end point E3 to the third start point S3 is set as a candidate for the crack start position SP1. By repeatedly performing the start point derivation process, multiple extension lines Ln2 are obtained. The intersection point of the multiple extension lines Ln2 becomes the crack start position SP1.
[0079] Figure 9 of (a) and Figure 9 of (b) are schematic diagrams illustrating the operation of the fracture surface image analysis device according to the first embodiment.
[0080] In Figure 9 of (a) and Figure 9 of (b), the magnifications of the first images 71D are different from each other. Figure 9 The magnification of the first image 71D in (a) of [] is higher than that ofFigure 9 The magnification of the imaging of the first image 71D in (b) is low. Figure 9 The magnification of the plurality of small block regions 71P in (a) is higher than Figure 9 the magnification of the plurality of small block regions 71P in (b). In this way, the first process may also include: changing the range included in at least one of the plurality of small block regions 71P according to the magnification of the imaging of the first image 71D. For example, the input first image 71D is adjusted to a size suitable for the processing model based on deep learning. Appropriate processing can be implemented with high precision.
[0081] Figure 10 of (a) and Figure 10 (b) are schematic diagrams illustrating the operation of the fracture surface image analysis device according to the first embodiment.
[0082] As Figure 10 shown in (a), for example, with respect to the plurality of small block regions 71P, a plurality of crack propagation directions Dr1 are derived. At this time, one of the plurality of crack propagation directions Dr1 may be significantly (exceeding the threshold) different from the other directions of the plurality of crack propagation directions Dr1. The other directions of the plurality of crack propagation directions Dr1 are beside the plurality of crack propagation directions Dr1.
[0083] In such a case, as Figure 10 shown in (b), for one (averaged crack propagation direction Dra1) that is significantly different among the plurality of crack propagation directions Dr1, correction is performed based on the other average direction. For example, the averaged crack propagation direction Dra1 is obtained by averaging at least a part of the plurality of crack propagation directions Dr1. The averaging may also be, for example, averaging in units of the plurality of small block regions 71P. By using the averaged crack propagation direction Dra1, for example, the crack start position SP1 can be derived with higher accuracy.
[0084] Figure 11 are schematic diagrams illustrating the operation of the fracture surface image analysis device according to the first embodiment.
[0085] Figure 11 Shows a display example based on the GUI 79a in the fracture surface image analysis device 110. For example, the derived plurality of crack propagation directions Dr1 may be overlapped and displayed with the input image (the first image 71D). For example, at least a part of the plurality of crack propagation directions Dr1 may be filtered and displayed according to the confidence level CN1. For example, the plurality of crack propagation directions Dr1 may be displayed according to the specified resolution. For example, the crack start position SP1 is displayed. The crack start position SP1 may also be overlapped and displayed with the input image (the first image 71D). For example, based on the image coordinate system for display, a plurality of images and the crack start position SP1 may be displayed.
[0086] In this way, the fracture surface image analysis device 110 may also include a GUI 79a. The GUI 79a is configured to perform at least any one of a first display action, a second display action, a third display action, a fourth display action, a fifth display action, a sixth display action, and a seventh display action.
[0087] In the first display action, the GUI 79a, for example, overlays and displays at least a part of a plurality of crack propagation directions Dr1 on the first image 71D. In the second display action, the GUI 79a, for example, selectively displays a part of the plurality of crack propagation directions Dr1 according to the orientation of the plurality of crack propagation directions Dr1.
[0088] In the third display action, the GUI 79a, for example, displays a part of the plurality of crack propagation directions Dr1 in units of a plurality of small block regions 71P. In the fourth display action, the GUI 79a displays the plurality of crack propagation directions Dr1 related to the whole of the first image 71D.
[0089] In the fifth display action, the GUI 79a, for example, displays an averaged crack propagation direction Dra1. The averaged crack propagation direction Dra1 is a direction obtained by averaging at least a part of the plurality of crack propagation directions Dr1. The averaging method is arbitrary. At least one of the plurality of crack propagation directions Dr1 before averaging may also be displayed.
[0090] In the sixth display action, the GUI 79a displays the plurality of small block regions 71P in the image coordinate system. In the seventh display action, the GUI 79a displays the crack start position SP1.
[0091] The GUI 79a may also be configured to perform at least any one of an eighth display action and a ninth display action. In the eighth display action, the GUI 79a, for example, selectively displays a part of the plurality of crack propagation directions Dr1 based on the confidence level CN1 respectively related to the plurality of crack propagation directions Dr1. In the ninth display action, the GUI 79a, for example, selectively displays a part of the plurality of crack propagation directions Dr1 (such as the arrow As1) used when deriving the crack start position SP1.
[0092] In addition to the above, the GUI 79a may also display arbitrary content. The content includes, for example, conditions related to image processing, etc. The content may also include, for example, commands related to image processing, etc.
[0093] (Second Embodiment)
[0094] The second embodiment relates to a fracture surface image analysis method. In the fracture surface image analysis method, a first image 71D including the fracture surface of the component 81 is acquired. In the fracture surface image analysis method, image analysis of the first image 71D is performed. For example, in the fracture surface image analysis method, the image analysis of the first image 71D is performed by the processing unit 70. For example, the image analysis includes a first process and a second process. The first process includes: deriving a plurality of crack propagation directions Dr1 in the fracture surface. One of the plurality of crack propagation directions Dr1 corresponds to one of the plurality of positions (for example, a plurality of small block regions 71P) included in the fracture surface. The second process includes: deriving a crack starting position SP1 in the fracture surface based on at least a part of the plurality of crack propagation directions Dr1.
[0095] In the fracture surface image analysis method according to the embodiment, at least a part of the operation of the fracture surface image analysis device 110 described in the first embodiment may also be applied.
[0096] According to the embodiment, at least a part of the fracture surface image analysis is automated. The analysis of the fracture surface can be performed quickly and with high accuracy. In the embodiment, based on the image related to the fracture surface, a plurality of crack propagation directions Dr1 are derived. Furthermore, the crack starting position SP1 can be derived based on the plurality of crack propagation directions Dr1. Thereby, an analysis that is difficult to achieve in the image analysis of the reference example based on image classification can be realized. In one example according to the embodiment, models respectively suitable for different crack patterns are adopted. Higher-precision analysis can be performed. In the embodiment, even when a large amount of noise is included in the first image 71D, it is easy to obtain an appropriate analysis result. The fracture surface image analysis method according to the embodiment may also be, for example, a fracture surface image analysis method. The fracture surface image analysis method according to the embodiment may also be, for example, a fracture surface analysis method.
[0097] The embodiment may also include the following technical solutions.
[0098] (Technical solution 1)
[0099] A fracture surface image analysis device, comprising:
[0100] An acquisition unit configured to acquire a first image including the fracture surface of a component; and
[0101] A processing unit configured to perform image analysis of the first image,
[0102] The image analysis includes a first process and a second process,
[0103] The first process includes: deriving a plurality of crack propagation directions in the fracture surface,
[0104] One of the plurality of crack propagation directions corresponds to one of the plurality of positions included in the fracture surface.
[0105] The second process includes: deriving a crack starting position in the fracture surface based on at least a part of the plurality of crack propagation directions.
[0106] (Technical solution 2)
[0107] The fracture surface image analysis device according to Technical solution 1, wherein
[0108] The processing unit includes a plurality of models.
[0109] In the first operation, the processing unit uses one of the plurality of models to perform the first process.
[0110] In the second operation, the processing unit uses another one of the plurality of models to perform the first process.
[0111] Between the one of the plurality of models and the another one of the plurality of models, at least one of the crack pattern of the fracture surface, the material of the component, the shape of the component, and the molding conditions of the component is different.
[0112] (Technical solution 3)
[0113] The fracture surface image analysis device according to Technical solution 1 or 2, wherein
[0114] The first process includes: deriving one of the plurality of crack propagation directions with respect to one of the plurality of small block regions obtained by dividing the first image acquired by the acquisition unit.
[0115] (Technical solution 4)
[0116] The fracture surface image analysis device according to Technical solution 3, wherein
[0117] The first process includes: changing a range included in at least one of the plurality of small block regions according to a magnification of the shooting of the first image.
[0118] (Technical solution 5)
[0119] The fracture surface image analysis device according to any one of Technical solutions 1 to 4, wherein
[0120] The first process further includes: deriving a confidence level respectively related to the derived plurality of crack propagation directions.
[0121] The second processing includes: deriving the crack starting position without using at least one of the plurality of crack propagation directions, and the confidence level corresponding to the at least one of the plurality of crack propagation directions is less than a determined reference value.
[0122] (Technical solution 6)
[0123] The crack surface image analysis device according to any one of Technical solutions 1 to 5, wherein
[0124] The first image is obtained from an imaging device that images the crack surface.
[0125] The second processing includes: deriving the crack starting position by using common coordinates related to the plurality of crack propagation directions.
[0126] The coordinates are the reference coordinates of the imaging device during imaging.
[0127] (Technical solution 7)
[0128] The crack surface image analysis device according to Technical solution 3 or 4, wherein
[0129] The crack surface image analysis device further includes a GUI.
[0130] The GUI is configured to perform at least any one of a first display action, a second display action, a third display action, a fourth display action, a fifth display action, a sixth display action, and a seventh display action.
[0131] In the first display action, the GUI displays at least a part of the plurality of crack propagation directions overlapping on the first image.
[0132] In the second display action, the GUI selectively displays a part of the plurality of crack propagation directions according to the orientations of the plurality of crack propagation directions.
[0133] In the third display action, the GUI displays a part of the plurality of crack propagation directions in units of the plurality of small block regions.
[0134] In the fourth display action, the GUI displays the plurality of crack propagation directions related to the whole of the first image.
[0135] In the fifth display action, the GUI displays an averaged crack propagation direction, and the averaged crack propagation direction is a direction obtained by averaging at least a part of the plurality of crack propagation directions.
[0136] In the sixth display action, the GUI displays the plurality of small block regions in an image coordinate system.
[0137] In the seventh display operation, the GUI displays the crack start position.
[0138] (Technical solution 8)
[0139] The crack surface image analysis device according to any one of Technical solutions 1 to 4, wherein
[0140] The crack surface image analysis device further includes a GUI,
[0141] The GUI is configured to perform at least one of an eighth display operation and a ninth display operation.
[0142] In the eighth display operation, the GUI selectively displays a part of the plurality of crack propagation directions based on the confidence levels respectively associated with the plurality of crack propagation directions.
[0143] In the ninth display operation, the GUI selectively displays the part of the plurality of crack propagation directions that were used when deriving the crack start position.
[0144] (Technical solution 9)
[0145] The crack surface image analysis device according to any one of Technical solutions 1 to 8, wherein
[0146] The plurality of crack propagation directions include a first direction, a second direction, and a third direction.
[0147] The first direction is a direction from a first start point to a first end point.
[0148] The second direction is a direction from a second start point to a second end point.
[0149] The third direction is a direction from a third start point to a third end point.
[0150] The position of the second end point on the straight line along the first direction is located between the position of the second start point on the straight line and the position of the first end point on the straight line.
[0151] The position of the first start point on the straight line is located between the position of the second end point on the straight line and the position of the first end point on the straight line.
[0152] The second process includes repeatedly performing start point derivation processing.
[0153] One process in the start point derivation processing includes determining the second direction based on the first direction.
[0154] The third direction corresponds to the second direction determined by repeatedly performing the starting point derivation process.
[0155] The second process sets the position of the extension line in the direction from the third end point to the third starting point as a candidate for the crack starting point position.
[0156] (Technical solution 10)
[0157] The crack surface image analysis device according to any one of Technical solutions 1 to 9, wherein
[0158] The first process includes a process performed by a regression processing unit configured to perform a process based on machine learning related to a plurality of teacher images including the crack surface and the plurality of crack propagation directions.
[0159] (Technical solution 11)
[0160] A crack surface image analysis method, wherein
[0161] Obtain a first image including the crack surface of a component.
[0162] An image analysis of the first image is performed by a processing unit.
[0163] The image analysis includes a first process and a second process.
[0164] The first process includes: deriving a plurality of crack propagation directions in the crack surface.
[0165] One of the plurality of crack propagation directions corresponds to one of the plurality of positions included in the crack surface.
[0166] The second process includes: deriving a crack starting point position in the crack surface based on at least a part of the plurality of crack propagation directions.
[0167] (Technical solution 12)
[0168] The crack surface image analysis method according to Technical solution 11, wherein
[0169] The processing unit includes a plurality of models.
[0170] In a first operation, the processing unit uses one of the plurality of models to perform the first process.
[0171] In a second operation, the processing unit uses another one of the plurality of models to perform the first process.
[0172] Between one of the plurality of models and another one of the plurality of models, at least one of the crack pattern of the fracture surface, the material of the component, the shape of the component, and the molding conditions of the component is different.
[0173] (Technical solution 13)
[0174] The fracture surface image analysis method according to Technical solution 11 or 12, wherein
[0175] The first process includes: deriving one of the plurality of crack propagation directions with respect to one of the plurality of small block regions obtained by segmenting the first image.
[0176] (Technical solution 14)
[0177] The fracture surface image analysis method according to Technical solution 13, wherein
[0178] The first process includes: changing the range included in at least one of the plurality of small block regions according to the magnification of the imaging of the first image.
[0179] (Technical solution 15)
[0180] The fracture surface image analysis method according to any one of Technical solutions 11 to 14, wherein
[0181] The first process further includes: deriving a confidence level respectively related to the plurality of crack propagation directions derived.
[0182] The second process includes: deriving the crack starting position without using at least one of the plurality of crack propagation directions, and the confidence level corresponding to the at least one of the plurality of crack propagation directions is less than a determined reference value.
[0183] (Technical solution 16)
[0184] The fracture surface image analysis method according to any one of Technical solutions 11 to 15, wherein
[0185] The first image is obtained from an imaging device that images the fracture surface.
[0186] The second process includes: deriving the crack starting position using a common coordinate related to the plurality of crack propagation directions.
[0187] The coordinate is the reference coordinate at the time of imaging of the imaging device.
[0188] (Technical solution 17)
[0189] The fracture surface image analysis method according to Technical solution 13 or 14, wherein
[0190] At least any one of a first display operation, a second display operation, a third display operation, a fourth display operation, a fifth display operation, a sixth display operation, and a seventh display operation is also executed.
[0191] In the first display operation, at least a part of the plurality of crack propagation directions is overlapped and displayed on the first image.
[0192] In the second display operation, a part of the plurality of crack propagation directions is selectively displayed according to the orientations of the plurality of crack propagation directions.
[0193] In the third display operation, a part of the plurality of crack propagation directions is displayed in units of the plurality of small block regions.
[0194] In the fourth display operation, the plurality of crack propagation directions related to the whole of the first image are displayed.
[0195] In the fifth display operation, an averaged crack propagation direction is displayed, and the averaged crack propagation direction is a direction obtained by averaging at least a part of the plurality of crack propagation directions.
[0196] In the sixth display operation, the plurality of small block regions are displayed in an image coordinate system.
[0197] In the seventh display operation, the crack start position is displayed.
[0198] (Technical solution 18)
[0199] The crack surface image analysis method according to any one of Technical solutions 11 to 14, wherein
[0200] At least any one of an eighth display operation and a ninth display operation is also executed.
[0201] In the eighth display operation, a part of the plurality of crack propagation directions is selectively displayed based on the confidence levels respectively related to the plurality of crack propagation directions.
[0202] In the ninth display operation, the part of the plurality of crack propagation directions used when deriving the crack start position is selectively displayed.
[0203] (Technical solution 19)
[0204] The crack surface image analysis method according to any one of Technical solutions 11 to 18, wherein
[0205] The plurality of crack propagation directions include a first direction, a second direction, and a third direction.
[0206] The first direction is the direction from the first starting point to the first ending point,
[0207] The second direction is the direction from the second starting point to the second ending point,
[0208] The third direction is the direction from the third starting point to the third ending point,
[0209] The position of the second ending point on the straight line along the first direction is located between the position of the second starting point on the straight line and the position of the first ending point on the straight line.
[0210] The position of the first starting point on the straight line is located between the position of the second ending point on the straight line and the position of the first ending point on the straight line.
[0211] The second process includes repeatedly performing a starting point derivation process.
[0212] One process in the starting point derivation process includes determining the second direction based on the first direction.
[0213] The third direction corresponds to the second direction determined by repeatedly performing the starting point derivation process.
[0214] The second process sets the position of the extension line in the direction from the third ending point to the third starting point as a candidate for the crack starting point position.
[0215] (Technical solution 20)
[0216] The crack surface image analysis method according to any one of technical solutions 11 to 19, wherein
[0217] The first process includes a process performed by a regression processing unit configured to perform a process based on machine learning related to a plurality of teacher images including the crack surface and the plurality of crack propagation directions.
[0218] According to the embodiment, a crack surface image analysis device and a crack surface image analysis method capable of performing highly accurate analysis can be provided.
[0219] As described above, the embodiments of the present invention have been described with reference to specific examples. However, the present invention is not limited to these specific examples. For example, regarding the specific configurations of each element such as the acquisition unit and the processing unit included in the crack surface image analysis device, as long as those skilled in the art appropriately select from the publicly known range so that the present invention can be similarly implemented and the same effects can be obtained, they are included in the scope of the present invention.
[0220] In addition, within the scope technically possible, any combination of two or more elements of each specific example that encompasses the gist of the present invention is included within the scope of the present invention.
[0221] All other fracture surface image analysis apparatuses and fracture surface image analysis methods that can be appropriately designed and modified by those skilled in the art based on the fracture surface image analysis apparatus and fracture surface image analysis method described above as embodiments of the present invention, as long as they encompass the gist of the present invention, belong to the scope of the present invention.
[0222] In addition, within the scope of the idea of the present invention, various modification examples and correction examples can be conceived by those skilled in the art, and it should be understood that these modification examples and corrections also belong to the scope of the present invention.
[0223] The above describes several embodiments of the present invention, but these embodiments are presented as examples and are not intended to limit the scope of the invention. These new embodiments can be implemented in various other ways, and various omissions, substitutions, and changes can be made without departing from the gist of the invention. These embodiments and their variations are included in the scope or gist of the invention and are included in the invention described in the claims and its equivalent scope.
Claims
1. A fracture surface image analysis device, wherein, Comprising: An acquisition unit configured to acquire a first image including a fracture surface of a component; and A processing unit configured to perform image analysis on the first image, The image analysis includes a first process and a second process, The first process includes: deriving a plurality of crack propagation directions in the fracture surface, One of the plurality of crack propagation directions corresponds to one of the plurality of positions included in the fracture surface, The second process includes: deriving a crack starting position in the fracture surface based on at least a part of the plurality of crack propagation directions.
2. The fracture surface image analysis device according to claim 1, wherein The processing unit includes a plurality of models, In a first operation, the processing unit uses one of the plurality of models to perform the first process, In a second operation, the processing unit uses another one of the plurality of models to perform the first process, Between the one of the plurality of models and the another one of the plurality of models, at least one of the crack pattern of the fracture surface, the material of the component, the shape of the component, and the molding conditions of the component is different.
3. The fracture surface image analysis device according to claim 1, wherein The first process includes: deriving one of the plurality of crack propagation directions for one of the plurality of small block regions obtained by segmenting the first image acquired by the acquisition unit.
4. The fracture surface image analysis device according to claim 3, wherein The first process includes: changing the range included in at least one of the plurality of small block regions according to the magnification of the shooting of the first image.
5. The fracture surface image analysis device according to any one of claims 1 to 4, wherein The first process further includes: deriving a confidence level respectively associated with the derived plurality of crack propagation directions, The second process includes: deriving the crack starting position without using at least one of the plurality of crack propagation directions, and the confidence level corresponding to the at least one of the plurality of crack propagation directions is less than a determined reference value.
6. The fracture surface image analysis device according to claim 1, wherein The first image is obtained from an imaging device that images the fracture surface, The second process includes: using a common coordinate related to the plurality of crack propagation directions to derive the crack starting position, The coordinate is the reference coordinate at the time of imaging of the imaging device.
7. The fracture surface image analysis device according to claim 3 or 4, wherein It further includes a GUI, The GUI is configured to perform at least any one of a first display operation, a second display operation, a third display operation, a fourth display operation, a fifth display operation, a sixth display operation, and a seventh display operation, In the first display operation, the GUI displays at least a part of the plurality of crack propagation directions overlapped on the first image, In the second display operation, the GUI selectively displays a part of the plurality of crack propagation directions according to the orientation of the plurality of crack propagation directions, In the third display operation, the GUI displays a part of the plurality of crack propagation directions in units of the plurality of small block regions. In the fourth display operation, the GUI displays the plurality of crack propagation directions related to the entirety of the first image. In the fifth display operation, the GUI displays an averaged crack propagation direction, which is a direction obtained by averaging at least a part of the plurality of crack propagation directions. In the sixth display operation, the GUI displays the plurality of small block regions in an image coordinate system. In the seventh display operation, the GUI displays the crack start position.
8. The fracture surface image analysis device according to claim 1, wherein it further includes a GUI, the GUI is configured to perform at least one of an eighth display operation and a ninth display operation, in the eighth display operation, the GUI selectively displays a part of the plurality of crack propagation directions based on the confidence levels respectively related to the plurality of crack propagation directions, in the ninth display operation, the GUI selectively displays the part of the plurality of crack propagation directions used when deriving the crack start position.
9. A fracture surface image analysis method, wherein a first image including a fracture surface of a component is acquired, image analysis of the first image is performed by a processing unit, the image analysis includes a first process and a second process, the first process includes: deriving a plurality of crack propagation directions in the fracture surface, one of the plurality of crack propagation directions corresponds to one of the plurality of positions included in the fracture surface, the second process includes: deriving a crack start position in the fracture surface based on at least a part of the plurality of crack propagation directions.
10. The fracture surface image analysis method according to claim 9, wherein the processing unit includes a plurality of models, in a first operation, the processing unit uses one of the plurality of models to perform the first process, in a second operation, the processing unit uses another one of the plurality of models to perform the first process, at least one of the crack pattern of the fracture surface, the material of the component, the shape of the component, and the molding conditions of the component is different between the one of the plurality of models and the another one of the plurality of models.