Fracture automatic identification processing method and system based on deep learning

Abstract

The invention provides an automatic fracture recognition processing method and system based on deep learning, and relates to the technical field of fracture recognition, and the method comprises the steps: obtaining a to-be-recognized image corresponding to a target rock mass; wherein the to-be-recognized image is generated through a cylinder obtained by drilling the interior of a target rock mass; the target rock mass is any rock mass; inputting the to-be-recognized image into a preset crack recognition model to obtain a crack image corresponding to the to-be-recognized image; the fissure image comprises a trunk fissure / or a branch fissure; judging whether a trunk fracture and / or a branch fracture exists in the fracture image or not; if a trunk fracture and a branch fracture exist in the fracture image, removing the branch fracture to obtain a trunk fracture corresponding to the target rock mass; the purposes of removing branch fractures in the fracture image and only retaining a trunk fracture are achieved; therefore, the joint fissures of the rock mass can be automatically and accurately identified.

Description

Technical Field

[0001] The present invention relates to the field of crack identification technology, and in particular to a method and system for automatic crack identification based on deep learning. Background Art

[0002] As a kind of structural surface, rock mass fissures have complex development direction, length, opening, scale and degree of staggering, and their spatial distribution is highly random, anisotropic and hidden. The specific analysis and research of rock mass joint fissure characteristics is one of the important indicators for engineering rock mass classification and rock mass quality evaluation. In the existing technology, the morphology, occurrence, trace length and spatial distribution of joint fissures have been studied by means of field visual identification, trace inference, morphology and distribution model, fractal geometry, probability statistics, etc. However, the basis of the above research is the accurate identification of rock mass joint fissures, and in view of the needs of rock mass structural surface analysis and evaluation and engineering rock mass classification, the identification of rock mass fissures is very critical. How to accurately identify rock mass joint fissures automatically has become a technical problem that needs to be solved urgently. Summary of the invention

[0003] In view of the above technical problems, the technical solution adopted by the present invention is: According to a first aspect of the present application, a method for automatic crack identification and processing based on deep learning is provided, the method comprising the following steps: S100, obtaining an image to be identified corresponding to a target rock mass; wherein the image to be identified is generated by drilling a cylinder inside the target rock mass; the target rock mass is any rock mass; S200, inputting the image to be identified into a preset crack identification model to obtain a crack image corresponding to the image to be identified; the crack image includes a main crack and / or a branch crack; S300, determining whether there are main fissures and / or branch fissures in the fissure image; S400: If there are main fractures and branch fractures in the fracture image, the branch fractures are removed to obtain the main fractures corresponding to the target rock mass.

[0004] According to another aspect of the present application, a deep learning-based automatic crack identification and processing system is also provided, the system comprising: a storage medium and a processor; wherein the storage medium stores at least one instruction or at least one program, and the at least one instruction or the at least one program is loaded and executed by the processor to implement the deep learning-based automatic crack identification and processing method as described in the first aspect.

[0005] The present invention has at least the following beneficial effects: The deep learning-based automatic crack identification and processing method of the present invention obtains an image to be identified corresponding to a target rock mass; wherein the image to be identified is generated by drilling a cylinder obtained inside the target rock mass; the target rock mass is any rock mass; the image to be identified is input into a preset crack identification model to obtain a crack image corresponding to the image to be identified; the crack image includes main cracks and / or branch cracks; it is determined whether there are main cracks and / or branch cracks in the crack image; if there are main cracks and branch cracks in the crack image, the branch cracks are removed to obtain the main cracks corresponding to the target rock mass; the branch cracks in the crack image are removed to achieve the purpose of retaining only the main cracks; thereby the rock mass joints and fissures are automatically and accurately identified. BRIEF DESCRIPTION OF THE DRAWINGS

[0006] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0007] Figure 1 A flowchart of a method for automatically identifying and processing cracks based on deep learning provided by an embodiment of the present invention; Figure 2 A schematic diagram of the rock mass fracture automatic identification algorithm model structure provided by an embodiment of the present invention; Figure 3 A schematic diagram of a crack image provided by an embodiment of the present invention; Figure 4 A flowchart of a rock mass structural surface testing and analysis method based on deep learning provided in an embodiment of the present invention; Figure 5 A schematic diagram of a crack structure surface provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0008] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0009] It should be noted that, based on the present disclosure, those skilled in the art should understand that an aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects described herein can be used to implement a device and / or practice a method. In addition, other structures and / or functionalities other than one or more of the aspects described herein can be used to implement this device and / or practice this method.

[0010] Embodiment 1: The following will refer to Figure 1 The flowchart of the automatic crack identification and processing method based on deep learning is shown, which introduces an automatic crack identification and processing method based on deep learning.

[0011] The deep learning-based automatic crack identification and processing method may include the following steps: S100, obtaining an image to be identified corresponding to a target rock mass; wherein the image to be identified is generated by drilling a cylinder inside the target rock mass; the target rock mass is any rock mass.

[0012] In this embodiment, it should be noted that the fracture surface of the rock mass usually runs through the entire rock mass. Therefore, a cylinder can be drilled at a preset position of the rock mass, and then the circumference of the cylinder can be photographed to obtain an image to be identified corresponding to the target rock mass.

[0013] S200, inputting the image to be identified into a preset crack identification model to obtain a crack image corresponding to the image to be identified; the crack image includes a main crack and / or a branch crack.

[0014] In this embodiment, the preset crack recognition model may be a WAsk R-CNN algorithm model, and the preset crack recognition model is obtained by the following method: like Figure 2 As shown in the figure, a convolutional neural network model based on deep learning is used. Image crack annotations are used as samples to carry out model training and reverse training in the constructed WAsk R-CNN model, so as to construct an intelligent recognition and segmentation model for rock cracks, and realize intelligent recognition and segmentation of observed images; sine wave fitting is used to skeletonize rock joints and cracks, and three-dimensional reconstruction of rock crack morphology is realized.

[0015] The specific contents are as follows: 1) Construction of joint and fissure dataset Crack morphology annotation rules: select points along the edges of joints and cracks to form a closed area; export according to 1m length and 300 pixels; use a training set, validation set, and test set division ratio of 8:1:1, which can balance the size and retain some data to verify the performance of the model; it can better capture the noise and outliers in the data set, making the model more robust.

[0016] 2) Joint and crack image segmentation model Joint and crack image recognition consists of five parts: backbone feature extraction, region proposal network, region alignment, fully connected layer and instance segmentation network.

[0017] On the basis of feature extraction, the RPN region proposal network is used to identify and screen the effective feature layer, and the image of the suspected target is suggested to be intercepted; the ROI region alignment function is used to adjust the size of all the suggested frame intercepted images, and finally adjust them to square images. According to the intercepted suggested frame image, the regression model is used to determine whether there are target joints and cracks in the intercepted area, and the effective feature layer is extracted and decoded through the prediction frame to obtain the final prediction frame.

[0018] The segmentation of rock joints and fissures consists of four parts: fissure marking, scale clipping, binarization, and fissure cutting. The final prediction frame is used to intercept the target image of the effective feature layer, and the intercepted image is passed to the WAsk instance segmentation network for instance segmentation to obtain a segmented image. The skeletonization of joint and fissure features is based on pixel segmentation. The image pixels are peeled off layer by layer, irrelevant pixels are deleted, and target pixels are retained. The morphology of rock joints and fissures is outlined by line entities, important fissure features are strengthened, general morphology is simplified, and complex pixels are simplified into binary images connected by single pixels to distinguish the edge points and middle points of the fissure skeleton. The morphological characteristics of rock joints and fissures are obtained by depicting the central axis of the image.

[0019] The development of rock mass joints and fissures is complex and intertwined, and many micro-cracks are developed around the main joints and fissures. When depicting the joint and fissure skeleton, the target crack cannot be highlighted, so it is necessary to outline and depict the target joint and fissure by branch deletion. The rock mass joint and fissure skeleton extraction uses the image lookup table method to refine the edge points. The binary image is processed using a 3×3 sliding window, and the pixel matrix is ​​used to remove the relevant pixels in 28 pixels.

[0020] S300, determining whether there are main cracks and / or branch cracks in the crack image.

[0021] In this embodiment, it can be understood that the branch cracks in the crack image obtained in the above steps are useless cracks, that is, noise, and it is necessary to first determine whether there are main cracks and / or branch cracks in the crack image.

[0022] Further, step S300 may include the following steps: S310, obtaining the endpoints of the curve corresponding to each crack in the crack image to obtain an endpoint list A=(A 1 , A 2 , …, A i , …, A n ), i=1, 2,...,n; among them, A i is the ith endpoint in the crack image, and n is the number of endpoints in the crack image.

[0023] like Figure 3 As shown, the curve corresponding to each crack in the crack image has an endpoint, and each endpoint can be obtained to obtain A.

[0024] S311, according to A, determine the coordinates of each endpoint in A to obtain the endpoint coordinate list ZA=(ZA 1 , ZA 2 , …, ZA i , …, ZA n ); Among them, ZA i A i Corresponding coordinates; ZA i =(ZA i，x , ZA i，y ); ZA i，x is the X-axis coordinate of the i-th endpoint in the crack image, ZA i，y is the Y-axis coordinate of the i-th endpoint in the crack image.

[0025] In this embodiment, a plane coordinate system corresponding to the crack image may be constructed, and then the coordinates of each end point may be acquired to obtain ZA.

[0026] S312, if n=2, and ZA 1，y =ZA 2，y , then get A 1 With A 2 The horizontal distance between LA=|ZA 1，x -ZA 2，x |.

[0027] In this embodiment, it can be understood that the crack image corresponds to the cylinder, which can be understood as the plane corresponding to the side of the cylinder. Therefore, the two ends of the crack image are connected together corresponding to the cylinder. If n=2, and ZA 1，y =ZA 2，y , then it means A 1 With A 2 It is possible that they correspond to the same point on the cylinder. However, if the main crack is not a through-hole on the cylinder, then even if n = 2 and ZA 1，y=ZA 2，y , A 1 With A 2 It is also possible that they do not correspond to the same point; therefore, it is also necessary to obtain A 1 With A 2 The horizontal distance LA between them is used for further judgment.

[0028] S313, if LA=D, it is determined that there is a main fracture in the fracture image and no branch fracture; wherein D is the circumference of the cylinder obtained by drilling inside the target rock mass.

[0029] In this embodiment, when n=2, and ZA 1，y =ZA 2，y Under the premise that LA=D, we can determine A 1 With A 2 They correspond to the same point on the cylinder; therefore, it is determined that there is a main crack and no branch crack in the crack image.

[0030] Furthermore, after step S313, the method may further include the following steps: S314, A 1 With A 2 The crack corresponding to the curve between the two is determined as the main trunk crack.

[0031] Furthermore, after step S311, the method may further include the following steps: S320: If n>2, obtain a first preset value Q=1 and a second preset value M=2.

[0032] In this embodiment, if n>2, it means that there may be branch images in the fracture image.

[0033] S321, if Q<n-1, then enter S322; otherwise, enter S325.

[0034] S322, if M<n, enter S323; otherwise, enter S324.

[0035] S323, if ZA Q，y =ZA M，y , then ZA Q and ZA M Determine a set of intermediate endpoints; otherwise, obtain M=M+1 and enter S322; S324, obtain Q=Q+1, M=Q+1; enter S322.

[0036] S325, obtain each group of intermediate endpoints to obtain an intermediate endpoint group list B=(B 1 , B 2 , …, B j , …, Bm ), j = 1, 2, ..., m; where B j is the jth group of intermediate endpoints determined, and m is the number of groups of intermediate endpoints determined.

[0037] In this embodiment, through the above-mentioned loop steps, several groups of middle points with the same Y-axis coordinates can be determined, and the curve between two middle points with the same Y-axis coordinates may correspond to the main trunk crack.

[0038] S326, obtaining the horizontal distance between two intermediate endpoints in each group of intermediate endpoints in B, so as to obtain a horizontal distance list LB=(LB 1 , L.B. 2 , …, LB j , …, LB m ); where LB j For B j The horizontal distance between the two middle endpoints.

[0039] S327, traverse LB, if |LB j -D|<ΔL, it is determined that there are main cracks and branch cracks in the crack image; wherein ΔL is a preset horizontal distance difference threshold.

[0040] In this embodiment, there may be errors in image recognition and coordinate construction, so ΔL is set to increase the redundancy of judgment and improve the accuracy of judgment.

[0041] Furthermore, after step S327, the method may further include the following steps: S328, B j The crack corresponding to the curve between the two middle endpoints is determined as the main trunk crack.

[0042] It can be understood that after the main crack is determined, the remaining cracks are branch cracks, thereby achieving the purpose of determining the main cracks and branch cracks.

[0043] S400: If there are main fractures and branch fractures in the fracture image, the branch fractures are removed to obtain the main fractures corresponding to the target rock mass.

[0044] Furthermore, after step S400, the method may further include the following steps: S500: If there are only main cracks in the crack image, then the preset branch crack removal operation is not performed on the crack image.

[0045] In an exemplary embodiment, a deep learning-based automatic crack identification and processing system is provided, the system comprising: a storage medium and a processor; wherein the storage medium stores at least one instruction or at least one program, and the at least one instruction or the at least one program is loaded and executed by the processor to implement the deep learning-based automatic crack identification and processing method as described in the above embodiment.

[0046] The deep learning-based automatic crack identification and processing method of this embodiment obtains an image to be identified corresponding to the target rock mass; wherein the image to be identified is generated by a cylinder obtained by drilling into the inside of the target rock mass; the target rock mass is any rock mass; the image to be identified is input into a preset crack identification model to obtain a crack image corresponding to the image to be identified; the crack image includes main cracks and / or branch cracks; it is determined whether there are main cracks and / or branch cracks in the crack image; if there are main cracks and branch cracks in the crack image, the branch cracks are removed to obtain the main cracks corresponding to the target rock mass; the branch cracks in the crack image are removed to achieve the purpose of retaining only the main cracks; thereby the rock mass joints and fissures are automatically and accurately identified.

[0047] Furthermore, in this embodiment, a deep learning convolutional neural network WAsk R-CNN algorithm model is introduced to realize intelligent identification of rock joints and fissures by capturing joint and fissure features and extracting skeleton features, thereby replacing the manual measurement of joint and fissure occurrence in the original core logging process, improving the efficiency of drilling field operations, and saving human resource costs.

[0048] In addition, this method realizes the modularization and intelligence of professional experience, breaks down the professional knowledge barrier, and makes professional work data-based and digital, which can greatly promote the development of cutting-edge technology and provide valuable reference experience for the development of digital technology in various professions. It can quickly and efficiently complete similar work. In the survey of nuclear engineering projects, it has the advantages of intuitive test results and comprehensive engineering rock mass evaluation, and has won praise from peer experts and owners many times, saving the working time of on-site technicians, so that on-site technicians have sufficient energy for the overall management of the project, which is conducive to the development of the project. It reduces labor costs and the labor intensity of technicians and improves the working environment.

[0049] The method and system in this embodiment can be widely used in the identification of rock joints and fissures in underground surrounding rocks, can analyze rock quality, evaluate rock integrity, and realize the classification of underground surrounding rocks and the grading evaluation of basic rock quality; it can also be applied to rock quality analysis in water conservancy, hydropower, bridges and tunnels and other projects.

[0050] Embodiment 2: In order to analyze the fracture structure surface corresponding to the main fracture, the corresponding fracture structure surface can be generated based on the main fracture obtained in the first embodiment, and then the fracture structure surface is analyzed to provide the following: Figure 4 The method shown may include the following steps: H100, obtaining an initial fracture structure surface WA corresponding to the main fracture of the rock mass; wherein WA is generated according to a fracture image corresponding to the main fracture of the rock mass.

[0051] In this embodiment, Figure 5 As shown, after obtaining the main fracture data, it is input into the existing simulation software to generate the initial fracture structure surface corresponding to the main fracture.

[0052] H200, obtain the tilt angle θ of WA.

[0053] In this embodiment, it can be understood that the main crack is not necessarily horizontal, so there is a certain inclination angle between the generated initial crack structure surface and the horizontal plane.

[0054] H300, obtain the highest point G of the curve corresponding to the main crack in the crack image 1 and the lowest point G 2 Among them, G 1 =(G 1，x , G 1，y ); G 2 =(G 2，x , G 2，y ); G 1，x and G 1，y G 1 The X-axis coordinate and Y-axis coordinate, G 2，x and G 2，y G 2 The X-axis coordinate and Y-axis coordinate of .

[0055] In this embodiment, the line shape corresponding to the main trunk crack is a curve, so there is a highest point and a lowest point, and the highest point G of the curve corresponding to the main trunk crack in the crack image can be obtained. 1 and the lowest point G 2 .

[0056] H400, according to G 1 and G 2 , determine the inclination angle of the main crack γ = arctan (|G 1，y -G 2，y | / |G 1，x -G 2，x |).

[0057] H500, if |θ-γ|<GY, it is determined that WA does not need to be adjusted; otherwise, obtain each preset specified point on the curve corresponding to the main trunk crack to obtain the specified point list E=(E 1 , E 2 , …, E p , …, E q ), p = 1, 2, …, q; where E p = is the pth designated point preset on the curve corresponding to the main trunk crack obtained, q is the number of designated points preset on the curve corresponding to the main trunk crack; GY is the preset tilt angle difference threshold.

[0058] In this embodiment, ideally, θ and γ should be the same, but the main crack is not a standard sine curve. Therefore, there may be a deviation between the inclination angle of the structural surface obtained by the software and the inclination angle of the main crack. If the deviation is within the preset inclination angle difference threshold, it can be determined that WA is normal; otherwise, each preset designated point on the curve corresponding to the main crack is obtained to adjust WA; the designated points are evenly set points.

[0059] H600, according to E and WA, adjusts WA to obtain the target fracture structure surface, and analyzes the target fracture structure surface.

[0060] Further, step H600 may include the following steps: H610, obtain each designated point in E above WA to obtain the first designated point list C = (C 1 , C 2 , …, C r , …, C s ), r = 1, 2, ..., s; where C r is the rth designated point located above WA obtained, and s is the number of designated points located above WA obtained.

[0061] H620, obtain each designated point in E below WA to obtain a second designated point list F = (F 1 , F 2 , …, F u , …, F v ), u = 1, 2, …, v; where F u is the u-th designated point below WA obtained, and v is the number of designated points below WA obtained.

[0062] In this embodiment, it can be understood that WA is a plane, and the main crack is a curve, and a specified point on the main crack may be located above and / or below WA.

[0063] H630, according to C and F, adjust WA to obtain the target fracture structure surface.

[0064] Further, step H630 may include the following steps: H631, if α 1 ≤s / v≤α 2 , then obtain the vertical distance from each specified point in C to WA to obtain the first vertical distance list LC corresponding to C = (LC 1 , L.C. 2 ,…,LC r ,…,LC s ); where LC r C r The vertical distance from WA; and obtain the vertical distance from each specified point in F to WA to obtain the second vertical distance list LF corresponding to F = (LF 1 , LF 2 ,…,LF u ,…,LF v ); LF u F u The vertical distance from WA; α 1 is the first preset weight, α 2 is the second preset weight.

[0065] In this embodiment, α 1 <1,;α 2 >1; if α 1 ≤s / v≤α 2 , indicating that the number of designated points above and below WA is basically the same, but it is not possible to determine whether the vertical distance from WA is uniform. Therefore, the vertical distance from each designated point in C to WA and the vertical distance from each designated point in F to WA are obtained for further judgment.

[0066] H632, based on LC and LF, determine the average distance LC' corresponding to LC and the average distance LF' corresponding to LF.

[0067] H633: If |LC'-LF'|≤LU, WA is not adjusted; where LU is the preset average distance difference threshold.

[0068] In this embodiment, if |LC'-LF'|≤LU, it means that the position of WA is appropriate, and WA is not adjusted.

[0069] Furthermore, after step H633, the method further includes the following steps: H634, if LC'-LF'>LU, move WA downward by LQ along the perpendicular line of WA 1The distance between LQ 1 = (LC'-LF') / 2.

[0070] In this embodiment, if LC'-LF'>LU, it means that the position of WA is too high, so WA is moved down along the vertical line of WA by LQ. 1 The distance between the two fractures makes WA better able to characterize the structural surface corresponding to the main fracture.

[0071] H635, if LF'-LC'>LU, move WA upward by LQ along the perpendicular line of WA 2 The distance between LQ 2 = (LF'-LC') / 2.

[0072] In this embodiment, if LF'-LC'>LU, it means that the position of WA is too low, so WA is moved upward by LQ along the direction of the vertical line of WA. 2 The distance between the two fractures makes WA better able to characterize the structural surface corresponding to the main fracture.

[0073] Furthermore, step H630 also includes the following steps: H61, if s / v<α 1 , then move WA downward several times along the vertical line of WA, and judge whether s / v satisfies α each time. 1 ≤s / v≤α 2 ; The distance moved each time is the same.

[0074] In this embodiment, if s / v<α 1 , indicating that the main trunk crack of WA is located relatively high. Most of the middle points on the main trunk crack are located below WA. WA cannot well represent the actual structural surface corresponding to the main trunk crack. Therefore, WA can be moved downward several times along the direction of the vertical line of WA. Each time it is moved, it is judged whether s / v satisfies α. 1 ≤s / v≤α 2 .

[0075] H62, if s / v satisfies α 1 ≤s / v≤α 2 , then enter H631.

[0076] When s / v satisfies α 1 ≤s / v≤α 2 In this case, WA can be adjusted using the method in steps H631-H635.

[0077] Furthermore, step H630 also includes the following steps: H63, if s / v>α 2, then move WA upward several times along the direction of the vertical line of WA, and judge whether s / v satisfies α each time. 1 ≤s / v≤α 2 ; The distance moved each time is the same.

[0078] In this embodiment, if s / v>α 2 , indicating that the main trunk crack is located at the lower part of WA. Most of the middle points on the main trunk crack are located above WA. WA cannot well represent the actual structural surface corresponding to the main trunk crack. Therefore, WA can be moved upward several times along the vertical line of WA. Each time it is moved, it is judged whether s / v satisfies α. 1 ≤s / v≤α 2 .

[0079] H64, if s / v satisfies α 1 ≤s / v≤α 2 , then enter H631.

[0080] When s / v satisfies α 1 ≤s / v≤α 2 In this case, WA can be adjusted using the method in steps H631-H635.

[0081] In an exemplary embodiment, a rock structure surface testing and analysis system based on deep learning is provided, the system comprising: a storage medium and a processor; wherein the storage medium stores at least one instruction or at least one program, and the at least one instruction or the at least one program is loaded and executed by the processor to implement the rock structure surface testing and analysis method based on deep learning as described in the above embodiment.

[0082] The rock mass structural surface testing and analysis method based on deep learning in this embodiment obtains the initial fracture structural surface WA corresponding to the main fracture of the rock mass; obtains the inclination angle θ of WA; obtains the highest point G of the curve corresponding to the main fracture in the fracture image 1 and the lowest point G 2 According to G 1 and G 2 , determine the inclination angle γ of the main crack; if |θ-γ|<GY, it is determined that WA does not need to be adjusted; otherwise, obtain each preset specified point on the curve corresponding to the main crack to obtain a specified point list E; according to E and WA, adjust WA to obtain the target crack structural surface; the target crack structural surface obtained by the method of the present invention is more consistent with the actual structural surface corresponding to the main crack, so that the generated structural surface of the main crack is more accurate.

[0083] The rock mass structural surface test and analysis method based on deep learning can achieve three-dimensional reconstruction of the drilled rock mass structural surface through image preprocessing, data set construction, model training and parameter tuning, crack mask prediction, crack image segmentation and skeleton characterization, and crack morphology fitting, and further analyze the number of joint groups and average spacing in the development degree of the rock mass structural surface, the degree of combination of the main structural surfaces, and the type of the main structural surfaces. It is mainly used in rock mass quality evaluation and rock mass integrity evaluation. It replaces the original manual identification and analysis, realizes the modularization and intelligence of professional experience, breaks the professional knowledge barrier, makes professional work data-based and digital, saves human resource investment, shortens the project construction period, and reduces the project production cost.

[0084] In addition, although the steps of the method in the present disclosure are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.

[0085] An embodiment of the present invention also provides a non-transitory computer-readable storage medium, which can be set in an electronic device to store at least one instruction or at least one program related to implementing a method in a method embodiment. The at least one instruction or the at least one program is loaded and executed by the processor to implement the method provided in the above embodiment.

[0086] The program product may adopt any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0087] Computer readable signal media may include data signals propagated in baseband or as part of a carrier wave, in which readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Readable signal media may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0088] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.

[0089] Program code for performing the operations of the present application may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).

[0090] An embodiment of the present invention further provides an electronic device, comprising a processor and the aforementioned non-transitory computer-readable storage medium.

[0091] The electronic device is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0092] The electronic device is presented in the form of a general-purpose computing device. The components of the electronic device may include, but are not limited to: the at least one processor mentioned above, the at least one memory mentioned above, and a bus connecting different system components (including the memory and the processor).

[0093] The memory stores program codes, which can be executed by the processor, so that the processor executes the steps in various embodiments described in this specification.

[0094] The memory may include readable media in the form of volatile memory, such as random access memory (RAM) and / or cache memory, and may further include read only memory (ROM).

[0095] The memory may also include a program / utility having a set (at least one) of program modules, such program modules including but not limited to: an operating system, one or more application programs, other program modules and program data, each of which or some combination may include an implementation of a network environment.

[0096] The bus may represent one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures.

[0097] The electronic device may also communicate with one or more external devices (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may communicate with one or more devices that enable a user to interact with the electronic device, and / or may communicate with any device (e.g., routers, modems, etc.) that enables the electronic device to communicate with one or more other computing devices. Such communication may be performed through an input / output (I / O) interface. Furthermore, the electronic device may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) through a network adapter. The network adapter communicates with other modules of the electronic device through a bus. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.

[0098] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0099] An embodiment of the present invention further provides a computer program product, which includes program code. When the program product is run on an electronic device, the program code is used to enable the electronic device to execute the steps of the method according to various exemplary embodiments of the present invention described above in this specification.

[0100] Although some specific embodiments of the present invention have been described in detail by way of example, it should be understood by those skilled in the art that the above examples are only for illustration, not for limiting the scope of the present invention. It should also be understood by those skilled in the art that various modifications may be made to the embodiments without departing from the scope and spirit of the present invention.

Claims

1. A method for automatic crack identification and processing based on deep learning, characterized in that: The method comprises the following steps: S100, obtaining an image to be identified corresponding to a target rock mass; wherein the image to be identified is generated by drilling a cylinder inside the target rock mass; the target rock mass is any rock mass; S200, inputting the image to be identified into a preset crack identification model to obtain a crack image corresponding to the image to be identified; the crack image includes a main crack and / or a branch crack; S300, determining whether there are main fissures and / or branch fissures in the fissure image; S400: If there are main fractures and branch fractures in the fracture image, the branch fractures are removed to obtain the main fractures corresponding to the target rock mass.

2. The deep learning-based automatic crack identification and processing method according to claim 1 is characterized in that: Step S300 includes the following steps: S310, obtaining the endpoints of the curve corresponding to each crack in the crack image to obtain an endpoint list A=(A1, A2, ..., A i , …, A n ), i=1, 2,...,n; among them, A i is the i-th endpoint in the crack image, and n is the number of endpoints in the crack image; S311, according to A, determine the coordinates of each endpoint in A to obtain the endpoint coordinate list ZA=(ZA1, ZA2, ..., ZA i , …, ZA n ); Among them, ZA i A i Corresponding coordinates; ZA i =(ZA i，x , ZA i，y ); ZA i，x is the X-axis coordinate of the i-th endpoint in the crack image, ZA i，y is the Y-axis coordinate of the i-th endpoint in the crack image; S312, if n=2, and ZA 1，y =ZA 2，y , then the horizontal distance between A1 and A2 is obtained as LA=|ZA 1，x -ZA 2，x |; S313, if LA=D, it is determined that there is a main fracture in the fracture image and no branch fracture; wherein D is the circumference of the cylinder obtained by drilling inside the target rock mass.

3. The method for automatic crack identification based on deep learning according to claim 2 is characterized in that: After step S313, the method further includes the following steps: S314, determining the crack corresponding to the curve between A1 and A2 as the main crack.

4. The method for automatic crack identification based on deep learning according to claim 2 is characterized in that: After step S311, the method further includes the following steps: S320, if n>2, obtain a first preset value Q=1 and a second preset value M=2; S321, if Q<n-1, then go to S322; otherwise, go to S325; S322, if M<n, proceed to S323; otherwise, proceed to S324; S323, if ZA Q，y =ZA M，y , then ZA Q and ZA M Determine a set of intermediate endpoints; otherwise, obtain M=M+1 and enter S322; S324, obtain Q=Q+1, M=Q+1; enter S322; S325, obtain each group of intermediate endpoints to obtain an intermediate endpoint group list B = (B1, B2, ..., B j , …, B m ), j = 1, 2, …, m; where B j is the jth group of intermediate endpoints determined, and m is the number of groups of intermediate endpoints determined; S326, obtaining the horizontal distance between two intermediate endpoints in each group of intermediate endpoints in B, so as to obtain a horizontal distance list LB = (LB1, LB2, ..., LB j , …, LB m ); where LB j For B j The horizontal distance between the two middle endpoints; S327, traverse LB, if |LB j -D|<ΔL, it is determined that there are main cracks and branch cracks in the crack image; wherein ΔL is a preset horizontal distance difference threshold.

5. The deep learning-based automatic crack identification and processing method according to claim 4 is characterized in that: After step S327, the method further includes the following steps: S328, B j The crack corresponding to the curve between the two middle endpoints is determined as the main trunk crack.

6. The method for automatic crack identification based on deep learning according to claim 1, characterized in that: The preset crack recognition model includes the WAsk R-CNN algorithm model.

7. The method for automatic crack identification based on deep learning according to claim 1, characterized in that: After step S400, the method further comprises the following steps: S500: If there are only main cracks in the crack image, then the preset branch crack removal operation is not performed on the crack image.

8. A deep learning-based crack automatic identification and processing system, characterized in that: The system includes: a storage medium and a processor; wherein, the storage medium stores at least one instruction or at least one program, and the at least one instruction or the at least one program is loaded and executed by the processor to implement the deep learning-based automatic crack identification and processing method as described in any one of claims 1-7.

Images