Two-step Tunnel Grayscale Image Crack Recognition Method, System, Medium and Device

Abstract

The present invention belongs to the technical field of crack identification of geotechnical structures, and provides a two-step crack identification method, system, medium and device for tunnel grayscale images. Among them, the two-step crack identification method for tunnel grayscale images includes respectively performing edge extraction on a batch of tunnel grayscale images to obtain corresponding tunnel edge images, and then screening out all suspected images that may have cracks based on the comparison between the average gray level of the edge images and the gray level threshold; based on the suspected images and a fine segmentation network model, obtaining crack segmentation images of a batch of tunnel grayscale images, and then obtaining a crack identification result by statistically analyzing the crack segmentation images; the crack identification result includes the number of images with actual cracks, the number of cracks, and the characteristic parameters of each crack.

Description

Technical Field

[0001] The present invention belongs to the technical field of crack identification in rock and soil structures, and in particular relates to a two-step tunnel grayscale image crack identification method, system, medium and equipment. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] Cracks are a common defect in tunnel structures and a key factor affecting their safety. Traditional detection methods rely on manual judgment and auxiliary tools; these methods are inefficient, labor-intensive, and require high levels of professional expertise. Therefore, rapid detection methods are needed, and a commonly used method currently utilizes image processing technology to identify defects.

[0004] Common crack identification methods fall into two main categories: traditional image processing methods and machine learning-based crack identification methods. Traditional image processing methods include image segmentation and skeleton extension, crack width transformation algorithms, multi-scale neighborhood information for automatic pixel-level crack detection, and structural condition assessment using 3D scene reconstruction. Machine learning-based crack identification methods, on the other hand, include spatially tuned robust multi-feature classifiers, the use of random forests for automatic road crack detection, and deep convolutional neural networks for crack classification and region extraction and segmentation. Traditional image processing methods are less effective for crack images with complex backgrounds, fractures, or rough wall surfaces. Machine learning methods also require a high number and variety of crack samples, as well as high training sample requirements.

[0005] The inventors discovered that large amounts of tunnel surface grayscale image data collected through centralized automated processing are large, while cracks are relatively small, resulting in a relatively small amount of data containing cracks. Consequently, high-precision neural network algorithms require high computational complexity, consuming significant time processing invalid data. Therefore, existing machine learning methods are unsuitable for identifying cracks in tunnel structures using large amounts of data. Summary of the Invention

[0006] In order to solve the technical problems existing in the above-mentioned background technology, the present invention provides a two-step tunnel grayscale image crack recognition method, system, medium and equipment, which is suitable for tunnel structure crack recognition in large-scale data scenarios, can greatly reduce data measurement and greatly improve the efficiency of recognition and detection.

[0007] In order to achieve the above object, the present invention adopts the following technical solution:

[0008] A first aspect of the present invention provides a two-step crack identification method for tunnel grayscale images, comprising:

[0009] Edge extraction is performed on a batch of tunnel grayscale images respectively to obtain corresponding tunnel edge images, and then based on the comparison between the average grayscale of the edge images and the grayscale threshold, all suspected images that may have cracks are screened out;

[0010] Based on the suspected images and the fine segmentation network model, crack segmentation images of a batch of tunnel grayscale images are obtained, and then the crack recognition results are obtained by counting the crack segmentation images; the crack recognition results include the number of images with actual cracks, the number of cracks, and the characteristic parameters of each crack.

[0011] The second aspect of the present invention provides a crack recognition system for two-step tunnel grayscale images, which includes:

[0012] A rough screening module, which is used to perform edge extraction on a batch of tunnel grayscale images respectively to obtain corresponding tunnel edge images, and then based on the comparison between the average grayscale of the edge images and the grayscale threshold, all suspected images that may have cracks are screened out;

[0013] A fine recognition module, which is used to obtain crack segmentation images of a batch of tunnel grayscale images based on the suspected images and the fine segmentation network model, and then obtain the crack recognition results by counting the crack segmentation images; the crack recognition results include the number of images with actual cracks, the number of cracks, and the characteristic parameters of each crack.

[0014] The third aspect of the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps in the above-mentioned crack recognition method for two-step tunnel grayscale images are implemented.

[0015] The fourth aspect of the present invention provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps in the above-mentioned crack recognition method for two-step tunnel grayscale images are implemented.

[0016] Compared with the prior art, the beneficial effects of the present invention are:

[0017] Due to the characteristics of the present invention that the ratio of cracks to image size is quite different and the crack size is small, edge extraction is performed on a batch of tunnel grayscale images respectively to obtain corresponding tunnel edge images, and then based on the comparison between the average grayscale of the edge images and the grayscale threshold, all suspected images that may have cracks are screened out; then based on the suspected images and the fine segmentation network model, crack segmentation images of a batch of tunnel grayscale images are obtained, and then the crack recognition results are obtained by counting the crack segmentation images. It is suitable for crack recognition of tunnel structures in large-scale data scenarios, greatly reducing the measurement of data and improving the accuracy and efficiency of recognition and detection.

[0018] Advantages of additional aspects of the present invention will be given in part in the following description, become apparent in part from the following description, or be learned through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings forming a part of this specification are used to provide a further understanding of the present invention. The schematic embodiments and descriptions thereof of the present invention are used to explain the present invention and do not unduly limit the present invention.

[0020] Figure 1 is a schematic diagram of crack recognition principle for a two-step tunnel grayscale image according to an embodiment of the present invention;

[0021] Figure 2 is a flowchart of a crack recognition method for a two-step tunnel grayscale image according to an embodiment of the present invention;

[0022] Figure 3 is a schematic diagram of the structure of a fine segmentation network model according to an embodiment of the present invention;

[0023] Figure 4 is a schematic diagram of the structure of a crack recognition system for a two-step tunnel grayscale image according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0025] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field to which the present invention belongs.

[0026] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular forms are also intended to include the plural forms. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they specify the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0027] Embodiment 1

[0028] As Figure 1 and Figure 2 shown, this embodiment provides a crack recognition method for a two-step tunnel grayscale image, which specifically includes the following steps:

[0029] S101: Edge extraction is performed on the batch of tunnel grayscale images respectively to obtain the corresponding tunnel edge images, and the average grayscale of this data is calculated. Then, based on the comparison between the average grayscale of the edge image and the grayscale threshold, all suspected images that may have cracks are screened out.

[0030] Optionally, the grayscale threshold can usually be determined in two ways. One is a preset grayscale threshold, which can be set to 0.001 through simulation experiments. The other method is to calculate the average grayscale of all edge extraction results, and use this average grayscale as the grayscale threshold to compare with the edge extraction results of each image respectively to determine whether it is a suspected crack image.

[0031] In step S101, the process of obtaining the batch of tunnel grayscale images is as follows:

[0032] Based on the gradient detection method, the upper threshold and the lower threshold of the original tunnel grayscale image are calculated in batches, and the upper threshold and the lower threshold of the original tunnel grayscale image are adjusted to the set grayscale range. This makes the tunnel grayscale images show basically the same performance.

[0033] The Canny edge detection algorithm is used to perform edge extraction on the batch of tunnel grayscale images respectively.

[0034] In a specific implementation, the original tunnel grayscale image is obtained using a tunnel automatic acquisition device.

[0035] S10₂: Based on the suspected images and the fine segmentation network model, the crack segmentation images of the batch of tunnel grayscale images are obtained, and then the crack recognition results are obtained by counting the crack segmentation images; the crack recognition results include the number of images with actual cracks, the number of cracks, and the characteristic parameters of each crack.

[0036] In a specific implementation, before the suspected images are input into the fine segmentation network model, it further includes:

[0037] Normalize all the suspected images.

[0038] The normalization method is as follows:

[0039] The input image size is scaled to 640 * 640 pixels, and the image data is converted into the RGB color mode. Set the image mean mean to [0.485, 0.456, 0.406], set the image variance std to [0.229, 0.224, 0.225], and calculate the normalized image according to the formula P img =(P img -mean) / std, where P img is the color value of each pixel point in the image. Load the image data into the Dataloader for data training.

[0040] As Figure 3 shown, the fine segmentation network model is a U-shaped convolutional neural network including a residual network module. The network model is divided into an encoding module and a decoding module. The encoding module contains 4 downsampling modules, and the decoding module contains 4 upsampling modules. Residual modules are used to connect the corresponding layers in the encoding module and the decoding module.

[0041] For the tunnel crack image samples in the training dataset and validation dataset of the fine segmentation network model, the crack area is marked white, and the non-crack area is uniformly marked black.

[0042] The training process of the fine segmentation network model is as follows:

[0043] Extract and label data from several existing (e.g., 300) tunnel apparent gray-scale images, mark the crack area white, and uniformly mark the non-crack area black;

[0044] Based on the labeled data, construct a training set and a validation set. The ratio of the training set to the validation set is preset, e.g., 0.85:0.15. Use this dataset to train the U-shaped neural network for several rounds, e.g., 30 rounds, and the learning rate can be set to 0.03 to complete the training of the network.

[0045] Among them, the crack recognition results include but are not limited to the number of crack images, the proportion of crack images in the collected images, the number of cracks, the length, width, shape and other characteristic parameters of each crack, and the fine segmentation results of each crack.

[0046] Embodiment 2

[0047] As Figure 4 shown, this embodiment provides a crack recognition system for two-step tunnel gray-scale images, which specifically includes the following modules:

[0048] A rough screening module 201, which is used to perform edge extraction on a batch of tunnel gray-scale images respectively to obtain corresponding tunnel edge images, and then based on the comparison between the average gray scale of the edge images and the gray scale threshold, screen out all suspected images that may have cracks;

[0049] A fine recognition module 202, which is used to obtain the crack segmentation images of a batch of tunnel gray-scale images based on the suspected images and the fine segmentation network model, and then obtain the crack recognition results by counting the crack segmentation images; the crack recognition results include the number of images with actual cracks, the number of cracks and the characteristic parameters of each crack.

[0050] It should be noted here that each module in this embodiment corresponds to each step in Embodiment 1, and the specific implementation process is the same, so it will not be repeated here.

[0051] Example 3

[0052] This embodiment provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the steps in the crack recognition method for the two-step tunnel grayscale image as described above are implemented.

[0053] Example 4

[0054] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps in the crack recognition method for the two-step tunnel grayscale image as described above are implemented.

[0055] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the specified functions in one Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0056] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A crack recognition method for two-step tunnel grayscale images, characterized in that, Including: Performing edge extraction on a batch of tunnel grayscale images respectively to obtain corresponding tunnel edge images, and then screening out all suspected images that may have cracks based on the comparison between the average grayscale of the edge images and the grayscale threshold; the process of obtaining the batch of tunnel grayscale images is as follows: calculating the upper threshold and the lower threshold of the original tunnel grayscale images in batches based on the gradient detection method, and adjusting the upper threshold and the lower threshold of the original tunnel grayscale images to a set grayscale range; using the Canny edge detection algorithm to perform edge extraction on the batch of tunnel grayscale images respectively. Based on the suspected images and the fine segmentation network model, obtaining the crack segmentation images of the batch of tunnel grayscale images, and then obtaining the crack recognition results by statistically analyzing the crack segmentation images; the crack recognition results include the number of images with actual cracks, the number of cracks, and the characteristic parameters of each crack. The fine segmentation network model is a U-shaped convolutional neural network including a residual network module. This network model is divided into two parts: an encoding module and a decoding module. Among them, the encoding module contains 4 downsampling modules, the decoding module contains 4 upsampling modules, and the residual module is used to connect the corresponding layers in the encoding module and the decoding module.

2. The crack recognition method for two-step tunnel grayscale images according to claim 1, characterized in that Before the suspected images are input into the fine segmentation network model, it further includes: Performing normalization processing on all the suspected images.

3. The crack recognition method for two-step tunnel grayscale images according to claim 1, characterized in that For the tunnel crack image samples in the training dataset and the validation dataset of the fine segmentation network model, the crack regions are marked white, and the non-crack regions are uniformly marked black.

4. A crack recognition system for two-step tunnel grayscale images, characterized in that, Including: A rough screening module, which is used to perform edge extraction on a batch of tunnel grayscale images respectively to obtain corresponding tunnel edge images, and then screen out all suspected images that may have cracks based on the comparison between the average grayscale of the edge images and the grayscale threshold; the process of obtaining the batch of tunnel grayscale images is as follows: calculating the upper threshold and the lower threshold of the original tunnel grayscale images in batches based on the gradient detection method, and adjusting the upper threshold and the lower threshold of the original tunnel grayscale images to a set grayscale range; using the Canny edge detection algorithm to perform edge extraction on the batch of tunnel grayscale images respectively. A fine recognition module, which is used to obtain the crack segmentation images of the batch of tunnel grayscale images based on the suspected images and the fine segmentation network model, and then obtain the crack recognition results by statistically analyzing the crack segmentation images; the crack recognition results include the number of images with actual cracks, the number of cracks, and the characteristic parameters of each crack; the fine segmentation network model is a U-shaped convolutional neural network including a residual network module. This network model is divided into two parts: an encoding module and a decoding module. Among them, the encoding module contains 4 downsampling modules, the decoding module contains 4 upsampling modules, and the residual module is used to connect the corresponding layers in the encoding module and the decoding module.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the two-step crack recognition method for tunnel grayscale images described in any one of claims 1-3.

6. A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the two-step crack recognition method for tunnel grayscale images described in any one of claims 1-3.

Images