Image data processing method and device, computer readable storage medium and processor

Abstract

The invention discloses an image data processing method and device, a computer readable storage medium and a processor. The method comprises the following steps: acquiring image data of an underground disease area; performing image enhancement and gray histogram equalization operation on the image data to obtain preprocessed image data; image recognition is carried out on the preprocessed image data, a recognition result is obtained, and the recognition result comprises semantic category labels corresponding to multiple pixels in the image data; determining a disease area and a non-disease area in the image data based on semantic category labels corresponding to the plurality of pixels in the recognition result; the target position of the disease area in the underground disease area is subjected to three-dimensional visualization processing, a detection result of the disease area is obtained, and the detection result is used for representing three-dimensional structure space features of the disease area. The technical problem that underground diseases are difficult to accurately detect is solved.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to an image data processing method, device, computer-readable storage medium and processor. Background Art

[0002] At present, Ground Penetrating Radar (GPR) is the mainstream means of detecting underground diseases, but its image segmentation technology, such as threshold-based methods, region growing methods, edge detection and segmentation based on graph cuts, generally has problems such as sensitivity to environmental conditions, insufficient processing capabilities for complex scenes, high computing resource requirements and reliance on manual experience. Moreover, in terms of three-dimensional visualization, traditional technologies overly rely on the experience of the operating object to set the amplitude threshold, resulting in subjectivity and uncertainty in the results. Therefore, there is a technical problem that it is difficult to accurately detect underground diseases.

[0003] With respect to the above-mentioned technical problem of difficulty in accurately detecting underground diseases, no effective solution has been proposed so far. Summary of the invention

[0004] The embodiments of the present invention provide a method and device for processing image data, a computer-readable storage medium, and a processor, so as to at least solve the technical problem that it is difficult to accurately detect underground diseases.

[0005] According to one aspect of an embodiment of the present invention, a method for processing image data is provided. The method may include: collecting image data of an underground diseased area; performing image enhancement and grayscale histogram equalization operations on the image data to obtain preprocessed image data; performing image recognition on the preprocessed image data to obtain a recognition result, wherein the recognition result includes semantic category labels corresponding to multiple pixels in the image data, and the semantic category labels are used to indicate the diseased area category to which the corresponding pixels belong, and the diseased area category includes at least a disease category and a non-disease category; based on the semantic category labels corresponding to multiple pixels in the recognition result, determining the diseased area and the non-disease area in the image data; performing three-dimensional visualization processing on the target position of the diseased area in the underground diseased area to obtain a detection result of the diseased area, wherein the detection result is used to represent the three-dimensional structural spatial characteristics of the diseased area.

[0006] Optionally, image recognition is performed on the preprocessed image data to obtain a recognition result, including: inputting the preprocessed image data into a target defect detection model for image recognition to obtain a recognition result, wherein the target defect detection model is obtained by training an initial defect detection model with image data samples of the diseased area.

[0007] Optionally, the target defect detection model includes an encoder and a decoder, wherein the encoder consists of a multi-way convolutional feature selection module, a parallel multi-directional attention mechanism and a residual compression module, the multi-way convolutional feature selection module is used to extract features of different scales in the image data, the parallel multi-directional attention mechanism is used to enhance the perception of features of different scales, the residual compression module is used to compress features of different scales, and the decoder is used to reconstruct the image data according to the compressed features of different scales, and generate recognition results based on the reconstructed image data.

[0008] Optionally, the image data processing method also includes: obtaining image data samples of diseased area samples; annotating the image data samples to obtain label data corresponding to the image data samples; using the image data samples and the label data corresponding to the image data samples to train an initial defect detection model to obtain a target defect detection model.

[0009] Optionally, based on the semantic category labels corresponding to multiple pixels in the recognition result, the diseased area and the non-diseased area in the image data are determined, including: determining the area corresponding to the pixels whose semantic category labels in the recognition result are the diseased category as the diseased area; and determining the area corresponding to the pixels whose semantic category labels in the recognition result are the non-diseased category as the non-diseased area.

[0010] Optionally, a three-dimensional visualization processing is performed on the target position of the diseased area in the underground diseased area to obtain a detection result of the diseased area, including: extracting geometric information of the diseased area, wherein the geometric information includes at least the coordinates of the center point of the diseased area and the size information of the diseased area; determining the target position of the diseased area in the underground diseased area based on the coordinates of the center point of the diseased area and the size information of the diseased area; collecting image data at the target position; reconstructing the amplitude information of the image data at the target position at different depths based on an amplitude reconstruction algorithm, wherein the amplitude reconstruction algorithm is constructed based on inverse scattered imaging technology and offset technology; and determining the detection result of the diseased area based on the amplitude information.

[0011] According to another aspect of an embodiment of the present invention, a device for processing image data is also provided. The device may include: an acquisition unit for acquiring image data of an underground diseased area; a preprocessing unit for performing image enhancement and grayscale histogram equalization operations on the image data to obtain preprocessed image data; a recognition unit for performing image recognition on the preprocessed image data to obtain a recognition result, wherein the recognition result includes semantic category labels corresponding to multiple pixels in the image data, the semantic category labels are used to indicate the diseased area category to which the corresponding pixels belong, and the diseased area category includes at least a disease category and a non-disease category; a determination unit for determining the diseased area and non-disease area in the image data based on the semantic category labels corresponding to multiple pixels in the recognition result; and a processing unit for performing three-dimensional visualization processing on the target position of the diseased area in the underground diseased area to obtain a detection result of the diseased area, wherein the detection result is used to represent the three-dimensional structural spatial characteristics of the diseased area.

[0012] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored program, wherein when the program is executed by a processor, the device where the storage medium is located is controlled to execute the method for processing image data in an embodiment of the present invention.

[0013] According to another aspect of an embodiment of the present invention, a processor is further provided. The processor is used to run a program, wherein the image data processing method in the embodiment of the present invention is executed when the program is run.

[0014] According to another aspect of the embodiments of the present invention, a computer program product is provided, which includes computer instructions, and when the computer instructions are executed by a processor, the image data processing method in the embodiments of the present invention is implemented.

[0015] In the embodiment of the present invention, the image data of the underground diseased area is recognized to obtain a recognition result, which includes the recognition result of each pixel in the image data, and then the diseased area in the image data is determined according to the recognition result of each pixel, and then the diseased area is three-dimensionally visualized to determine the three-dimensional spatial distribution characteristics of the diseased area. That is to say, in the embodiment of the present invention, the recognition accuracy of the diseased area can be enhanced through pixel recognition and three-dimensional visualization, thereby solving the technical problem of poor detection accuracy of the underground diseased area. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0017] Figure 1is a flowchart of a method for processing image data according to an embodiment of the present invention;

[0018] Figure 2 is a schematic diagram of a Crack U-Net network structure according to an embodiment of the present invention;

[0019] Figure 3 is a flow chart of a method for image recognition based on a Crack U-Net model according to an embodiment of the present invention;

[0020] Figure 4 is a schematic diagram of a PDMA attention mechanism according to an embodiment of the present invention;

[0021] Figure 5 is a structural diagram of a MC-FS module according to an embodiment of the present invention;

[0022] Figure 6 is a structural diagram of a residual compression module according to an embodiment of the present invention;

[0023] Figure 7 is a flow chart of a three-dimensional visualization imaging method according to an embodiment of the present invention;

[0024] Figure 8 is a schematic diagram of an image data processing device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings 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 ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, functional component or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, functional components or devices.

[0027] According to an embodiment of the present invention, an embodiment of a method for processing image data is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0028] Figure 1 is a flowchart of a method for processing image data according to an embodiment of the present invention. Figure 1 As shown, the method may include the following steps:

[0029] Step S101, collecting image data of underground diseased areas.

[0030] In the technical solution provided in the above step S101 of the present invention, a high dynamic ground penetrating radar (GPR) device is usually used to collect image data of the underground disease area. GPR is a non-destructive detection technology that can emit high-frequency electromagnetic waves and receive reflected signals. By analyzing these signals, a two-dimensional or three-dimensional image of the underground structure is generated, thereby detecting underground diseases.

[0031] In this embodiment, in order to more comprehensively cover the underground disease area and obtain high-resolution data, the survey lines of the GPR ground penetrating radar need to be arranged in a grid shape, such as a "well" shape. This arrangement can ensure that image data is obtained in different directions, thereby improving the accuracy and reliability of disease detection.

[0032] Optionally, a GPR device is used to collect data along a preset survey line. The device continuously emits electromagnetic waves and receives reflected signals to collect image data of the diseased area. After processing, these signals are converted into a two-dimensional image (B-scan), in which each row represents the profile information along the survey line, and each column represents the information at different depths.

[0033] Step S102: performing image enhancement and grayscale histogram equalization operations on the image data to obtain pre-processed image data.

[0034] In the technical solution provided in the above step S102 of the present invention, image enhancement is a process of improving image quality by adjusting image properties such as brightness, contrast, and color. In the processing of underground disease images, the image quality is poor because the disease features may not be obvious enough in the original image, or are affected by the signal strength of the ground penetrating radar equipment and environmental factors (such as soil type, humidity, etc.). Therefore, by adjusting the image contrast and brightness, the difference between the diseased area and the surrounding normal structure is enhanced, which helps to more accurately identify the disease features.

[0035] Optionally, grayscale histogram equalization is an image processing technique used to improve the contrast of an image, especially when the grayscale distribution of the image is uneven. In ground penetrating radar images, grayscale histogram equalization can enhance the details of dark areas in the image and make the information of different grayscale levels in the image more balanced, thereby improving the visibility of disease features.

[0036] Optionally, by performing image enhancement and grayscale histogram equalization operations on the image data, the quality of the defect image can be improved, making its features more obvious and the contrast higher, thereby reducing the processing difficulty of the target defect detection model (Crack U-Net) and improving the segmentation accuracy of the model and the accuracy of three-dimensional visualization processing. Among them, the Crack U-Net model is obtained by improving the U-Net model.

[0037] Step S103, performing image recognition on the preprocessed image data to obtain a recognition result.

[0038] In the technical solution provided in the above step S103 of the present invention, image recognition is performed on the preprocessed image data through the deep learning model target defect detection model, and the obtained recognition result includes the semantic category label of each pixel in the image.

[0039] In this embodiment, the preprocessed image data is input into the target defect detection model, and the image data enters the encoder part of Crack U-Net, and the encoder extracts features from the image through a multi-layer convolutional neural network (CNN). The encoder includes a multi-convolution feature selection module (MC-FS), a parallel multi-directional attention mechanism (PMDA), and a residual compression (RS) module. These modules work together to extract defect features from different scales and angles, and enhance the model's ability to recognize defect details.

[0040] Optionally, features of different scales are extracted through parallel standard convolution and dilated convolution, and the channel attention mechanism performs weighted selection on the extracted features to ensure that the extracted features are highly relevant to the disease.

[0041] Optionally, a parallel multi-directional attention mechanism is used to establish associations between pixels and enhance the network's extraction of crack features, especially when dealing with long and continuous cracks, which can more effectively identify and highlight crack pixels and improve segmentation accuracy.

[0042] Optionally, the residual compression module compresses the shallow features of the encoder before feature concatenation to remove "rough" information, reduce the amount of network parameters, and ensure that the feature information received by the decoder is as effective as possible.

[0043] Optionally, the features extracted by the encoder are passed to the decoder, which uses upsampling and feature fusion techniques to gradually restore the resolution of the image, combine the high-level features of the encoder with the low-level features, and finally generate a recognition result with the same size as the input image. This recognition result contains the semantic category label of each pixel, which indicates whether each pixel belongs to the defect area and which defect category it belongs to (such as cracks, holes, voids, etc.).

[0044] Optionally, the semantic category labels output by the model can be post-processed, such as denoising, boundary refinement, etc., to obtain a clear segmentation map of the diseased area. By comparing the segmentation map with the original image, the location and range of the diseased area can be intuitively identified, providing accurate two-dimensional location information for subsequent three-dimensional visualization and disease analysis.

[0045] In this step, during the entire image recognition process, the Crack U-Net model uses deep learning technology, especially the improved attention mechanism and feature selection module, to automatically learn and extract disease features from the image, thereby achieving high-precision image segmentation. The recognition results obtained can not only distinguish between diseased and non-disease areas, but also provide classification information on disease types, which is of great significance for the health assessment of underground structures and the formulation of maintenance strategies.

[0046] Step S104, determining the diseased area and the non-diseased area in the image data based on the semantic category labels corresponding to the multiple pixels in the recognition result.

[0047] In the technical solution provided in the above step S104 of the present invention, it can be seen from the introduction of the above step S102 that the recognition result includes semantic category labels corresponding to multiple pixels in the image data. Based on this, the diseased area and non-diseased area in the image data can be determined according to the semantic category labels corresponding to the multiple pixels in the recognition result.

[0048] Optionally, according to the recognition result of the Crack U-Net model, each pixel corresponds to a label of a disease category or a non-disease category, and by analyzing these labels, pixels belonging to the diseased area and pixels belonging to the non-disease area can be determined.

[0049] Optionally, the area corresponding to the pixels labeled as the disease category in the recognition result is determined as the disease area, and the area corresponding to the pixels labeled as the non-disease category in the recognition result is determined as the non-disease area.

[0050] For example, pixels with a specific class label are clustered to form continuous damage area boundaries. For example, if the model predicts that a pixel belongs to the crack class, then this pixel and nearby pixels with the same label will be marked as part of the crack damage area.

[0051] In this step, based on the semantic category label of each pixel, the diseased area and non-diseased area in the image data can be accurately determined, providing an accurate data basis for subsequent three-dimensional visualization processing and disease assessment.

[0052] Step S105, performing three-dimensional visualization processing on the target position of the diseased area in the underground diseased area to obtain a detection result of the diseased area.

[0053] In the technical solution provided in the above step S105 of the present invention, three-dimensional visualization processing is performed on the target position of the diseased area in the underground diseased area to obtain the detection result of the diseased area, that is, the semantic category label output by the improved image segmentation algorithm is used, combined with the two-dimensional image data collected by the high-dynamic ground penetrating radar, and a precise three-dimensional model of the diseased area is generated through the true amplitude reconstruction algorithm and three-dimensional visualization technology.

[0054] In this embodiment, after the underground damage image is segmented and the damage area is determined, the true amplitude reconstruction algorithm can be applied to enhance the visualization of the damage features. The true amplitude reconstruction algorithm is based on the inverse scattering imaging technology and the migration technology. By correcting and restoring the amplitude energy of the wave field at the scattering source position, it can more accurately represent the real state of the damage area in the two-dimensional image, improve the image quality and contrast, and make the damage features more prominent.

[0055] Optionally, the conversion of two-dimensional images to three-dimensional space is achieved by fitting and interpolating B-scan data at different time depths, lateral position directions, and longitudinal position directions. This process can construct a complete three-dimensional spatial structure of the diseased area, including its accurate depth, lateral distribution, and longitudinal distribution, as well as the shape and size of the disease, by analyzing and integrating B-scan data in multiple directions.

[0056] Optionally, after the 3D visualization is completed, 3D slice images of the diseased area can be obtained, which provide views of the disease at different angles and depths, helping to more fully understand and analyze the nature and impact of the disease. Through 3D slices, the burial depth of the abnormal body and its distribution in 3D space can be accurately inferred, which is crucial for the subsequent design of repair plans and the formulation of monitoring plans.

[0057] Optionally, the detection result, i.e. the detection information of the diseased area after 3D visualization processing, is used to represent the 3D structure and spatial characteristics of the diseased area. Such information includes but is not limited to the precise geometric shape of the diseased area, its precise position in the underground space, depth, and possible expansion direction. The 3D visualization result not only provides an intuitive image of the disease, but also can represent the parameters of the disease in the form of data, which is convenient for professionals to conduct more in-depth analysis and decision-making.

[0058] In the above steps S101 to S105, the image data of the underground diseased area is identified to obtain a recognition result, which includes the recognition result of each pixel in the image data, and then the diseased area in the image data is determined according to the recognition result of each pixel, and then the diseased area is three-dimensionally visualized to determine the three-dimensional spatial distribution characteristics of the diseased area. That is to say, in the embodiment of the present invention, the recognition accuracy of the diseased area can be enhanced through pixel recognition and three-dimensional visualization, thereby solving the technical problem of poor detection accuracy of the underground diseased area.

[0059] The above method of this embodiment is further introduced below.

[0060] As an optional implementation method, step S103, performing image recognition on the preprocessed image data to obtain a recognition result, includes: inputting the preprocessed image data into a target defect detection model for image recognition to obtain a recognition result, wherein the target defect detection model is obtained by training an initial defect detection model with image data samples of the diseased area.

[0061] In this embodiment, the target defect detection model may be a Crack U-Net model, wherein the Crack U-Net model is improved and optimized based on a U-Net architecture of deep learning.

[0062] Optionally, the standard defect detection model (Crack U-Net) introduces a parallel multi-directional attention mechanism (PMDA), a multi-way convolution-feature selection module (MC-FS) and a residual compression module (RS) based on U-Net. PMDA enhances the model's extraction of crack features by establishing an association relationship between pixel points, especially when dealing with slender and continuous cracks, it can more accurately capture and enhance crack pixels. The MC-FS module increases the diversity of features through parallel convolution and channel attention mechanisms, thereby improving the effectiveness of features and ensuring that the model only extracts key features related to the disease. The RS module compresses the shallow features of the encoder before feature connection, removes "rough" information, reduces the number of model parameters, avoids introducing noise into deep features, and improves operating efficiency and feature effectiveness.

[0063] Optionally, the training of the target defect detection model (Crack U-Net) is based on a large number of labeled diseased area image data samples. These samples include various typical forms and scenarios of underground diseases, such as different types of cracks, cavities, and voids. The model optimizes its parameters by minimizing the difference between the predicted results and the true labels (usually using the cross entropy loss function). During the training process, the model gradually learns how to effectively distinguish between diseased and non-diseased areas, and how to identify different types of diseases. After training, the model can automatically identify and segment the diseased area based on the input image data, and output the semantic category label for each pixel.

[0064] Optionally, after the target defect detection (Crack U-Net) model is trained, the preprocessed image can be input into the trained Crack U-Net model. The encoder part of the model extracts defect features from the image through the MC-FS and PMDA mechanisms, and weights crack pixels and other defect areas to enhance the model's perception and capture of these features. The features extracted by the encoder are fused and upsampled in the decoder to gradually restore the image resolution and ultimately generate a recognition result that is consistent with the input image size. In this process, the RS module compresses the features to ensure that the feature information received by the decoder is the most effective, avoiding unnecessary calculations and information redundancy.

[0065] Optionally, the recognition results output by the Crack U-Net model contain semantic category labels for each pixel, which indicate whether each pixel belongs to a diseased area and which disease category it belongs to. The accuracy of the recognition results depends on the quality of model training and the clarity of the preprocessed images. Through the deep learning capabilities of the Crack U-Net model, high-precision segmentation and classification of underground diseases can be achieved, providing accurate two-dimensional disease area information for subsequent three-dimensional visualization processing.

[0066] In short, image recognition of preprocessed image data is achieved through the application of the Crack U-Net model. This model uses deep learning technology, combined with innovative PMDA, MC-FS and RS modules, to automatically learn and extract disease features from images and output high-precision disease area identification results. This process not only improves the accuracy of image segmentation, but also reduces the reliance on manually set thresholds or seed points, greatly improving work efficiency and the reliability of results.

[0067] As an optional implementation method, the target defect detection model includes an encoder and a decoder, wherein the encoder consists of a multi-way convolutional feature selection module, a parallel multi-directional attention mechanism and a residual compression module, the multi-way convolutional feature selection module is used to extract features of different scales in image data, the parallel multi-directional attention mechanism is used to enhance the perception of features of different scales, the residual compression module is used to compress features of different scales, and the decoder is used to reconstruct the image data according to the compressed features of different scales, and generate recognition results based on the reconstructed image data.

[0068] In this embodiment, the target defect detection model (Crack U-Net) achieves high-precision segmentation of underground disease images through multi-way convolution feature selection, parallel multi-directional attention mechanism and residual compression of the encoder, and feature fusion and image reconstruction of the decoder. The model can not only accurately identify the characteristics of the disease, but also maintain high efficiency during the processing process, reduce unnecessary computing resource consumption, and provide accurate image recognition results for subsequent 3D visualization and other analyses.

[0069] As an optional implementation, the image processing method also includes: obtaining image data samples of diseased area samples; annotating the image data samples to obtain label data corresponding to the image data samples; using the image data samples and the label data corresponding to the image data samples to train an initial defect detection model to obtain a target defect detection model.

[0070] In this embodiment, image data samples of the diseased area samples are obtained and labeled, and then the labeled data is used to train the initial defect detection model, ensuring that the model can learn the characteristics of the diseased area and perform accurate segmentation.

[0071] Optionally, a large number of underground disease image data samples are collected. These samples can be collected at different underground disease detection sites using high-dynamic ground penetrating radar equipment to ensure that the samples cover various types of diseases, such as cracks, cavities, and voids, as well as disease images under different conditions, such as images at different depths, different geological backgrounds, and different lighting and noise levels. Such a sample set can make model training more comprehensive and improve its generalization ability in practical applications.

[0072] Optionally, after acquiring the damaged image data samples, each collected image data sample is labeled, that is, which pixels belong to the damaged area and which belong to the non-damaged area. The labeling result will assign a semantic category label to each pixel, such as crack category, cavity category or non-damaged category. This process is the basis of model training and ensures that the model can learn the correct damage features.

[0073] Optionally, before starting the training process, the Crack U-Net model needs to be pre-trained, for example, using model parameters that perform well on other related tasks as initialization. Pre-training can accelerate the convergence of the model and improve training efficiency. After the initial model is ready, it can be trained with the labeled dataset.

[0074] Optionally, the Crack U-Net model is trained using the labeled image data samples and the corresponding label data. The training process includes two stages: forward propagation and back propagation. In the forward propagation, the input image data passes through the model and the corresponding predicted label is output; in the back propagation, the difference (loss) between the predicted label and the actual label is calculated, and the parameters of the model are adjusted according to this loss to reduce the prediction error. The goal of training is to optimize the model parameters so that the model can achieve the minimum loss on the test data, that is, the model's prediction results are as close as possible to the actual results of manual annotation.

[0075] Optionally, during the training process, the model's hyperparameters need to be adjusted, such as learning rate, batch size, optimizer selection, etc., to optimize the training process and model performance. In addition, the model should be evaluated using an independent validation dataset to ensure that the model not only works well on the training data, but also has the ability to maintain stable performance on unseen data.

[0076] Optionally, after training, the model needs to be evaluated on a test dataset for final performance. The test dataset should contain disease images that are different from the training dataset to verify the generalization ability of the model. If the test results are not satisfactory, it may be necessary to return to the training phase, adjust the model structure or parameters, and repeat the training process until the model achieves satisfactory performance on the test data.

[0077] Through the above steps, a fully trained and optimized Crack U-Net model, i.e., the target defect detection model, can be obtained. This model can automatically identify and segment the defect areas in underground defect images without manual intervention. The generated target model is the core of the entire technical solution and is used for subsequent defect image segmentation and 3D visualization processing.

[0078] As an optional implementation method, based on the semantic category labels corresponding to multiple pixels in the recognition result, the diseased area and the non-diseased area in the image data are determined, including: determining the area corresponding to the pixels whose semantic category label in the recognition result is the diseased category as the diseased area; determining the area corresponding to the pixels whose semantic category label in the recognition result is the non-diseased category as the non-diseased area.

[0079] In this embodiment, the identification results of the diseased area and the non-diseased area in the image data may be determined based on the semantic category label corresponding to each pixel in the identification result.

[0080] Optionally, in the recognition result, the area corresponding to the pixel points marked as the disease category by the model (i.e., its semantic category label is the disease category) will be determined as the disease area. The area corresponding to the pixel points marked as the non-disease category by the model (i.e., its semantic category label is the non-disease category) will be determined as the non-disease area.

[0081] Optionally, after the initial determination of the diseased and non-diseased areas, some post-processing steps are usually required to optimize the recognition results. For example, morphological operations such as dilation and erosion can be used to smooth the boundaries of the diseased areas and eliminate isolated pixels that may be caused by noise. In addition, connected domain analysis can be used to identify and mark complete diseased areas to ensure the consistency and integrity of the detection results.

[0082] In this step, by identifying the semantic category label of each pixel point included in the result, the diseased area and non-disease area in the image can be accurately distinguished, thereby improving the accuracy of disease detection.

[0083] As an optional implementation method, step S105 performs three-dimensional visualization processing on the target position of the diseased area in the underground diseased area to obtain the detection result of the diseased area, including: extracting geometric information of the diseased area, wherein the geometric information at least includes the coordinates of the center point of the diseased area and the size information of the diseased area; determining the target position of the diseased area in the underground diseased area based on the coordinates of the center point of the diseased area and the size information of the diseased area; collecting image data at the target position; reconstructing the amplitude information of the image data at the target position at different depths based on an amplitude reconstruction algorithm, wherein the amplitude reconstruction algorithm is constructed based on inverse scattered imaging technology and offset technology; and determining the detection result of the diseased area based on the amplitude information.

[0084] In this embodiment, performing three-dimensional visualization processing on the diseased area to obtain the detection results is a complex process based on multiple steps, which involves extracting the geometric information of the diseased area from the two-dimensional image data, determining the precise location of the disease in the underground space, and using the amplitude reconstruction algorithm to generate amplitude information of the diseased area at different depths, thereby ultimately determining the three-dimensional structure and spatial characteristics of the diseased area.

[0085] Optionally, in order to perform three-dimensional visualization processing, it is necessary to extract geometric information of the diseased area from the recognition results, including but not limited to the center point coordinates and size information of the diseased area. The center point coordinates are obtained by calculating the average position of the pixels in the diseased area, and the size information is determined by analyzing the boundary of the diseased area, including the length and width of the disease.

[0086] Optionally, after determining the geometric information of the diseased area, the target location of the disease in the underground diseased area can be determined in combination with the two-dimensional image data collected by the ground penetrating radar. This usually involves converting the two-dimensional coordinates into relative positions in the underground three-dimensional space, using the coordinates and size information of the center point of the diseased area, combined with the propagation path and reflection characteristics of the radar signal, to infer the depth and specific location of the disease in the underground space.

[0087] Optionally, after the target locations of the disease in the underground space are determined, these locations can be detected in more detail. This may involve re-routing the survey lines, using a denser detection grid, or adjusting the parameters of the radar equipment to obtain higher quality image data. These image data will be used for subsequent amplitude information reconstruction and 3D visualization.

[0088] Optionally, in order to more accurately represent the characteristics of the disease, it is necessary to reconstruct the amplitude information of the image data collected at the target location. This process is based on inverse scattering imaging technology and offset technology. By restoring the true amplitude energy of the wave field at the scattering source position, the amplitude distribution of the diseased area at different depths can be obtained. Inverse scattering imaging technology can restore the true shape of the object from the scattered wave, while the offset technology can correct the position offset of the scattered wave caused by factors such as wave velocity changes or terrain undulations, ensuring the accuracy and consistency of the amplitude information.

[0089] Optionally, 3D visualization can be performed based on the reconstructed amplitude information. By fitting B-scan data at different time depths, lateral position directions, and longitudinal position directions, the real interpolation of limited B-scan data is completed, and accurate 3D slice imaging of the diseased area is achieved. In this process, the 3D module of the reflexw software is used to process the 2D image data and generate 3D images of the diseased area at different depths and directions, thereby constructing a complete 3D model of the diseased area, including its accurate burial depth, lateral and longitudinal distribution, and the shape and size of the disease.

[0090] Optionally, the result of 3D visualization processing, i.e., the detection result of the diseased area, is used to represent the 3D structure and spatial characteristics of the diseased area. This information not only provides an intuitive image of the disease, but also can represent the disease parameters in the form of data, such as burial depth, volume, location coordinates, etc., to facilitate professionals to conduct more in-depth analysis and decision-making.

[0091] In this step, the three-dimensional visualization of the target position of the diseased area in the underground diseased area is a complex process including data collection, feature extraction, geometric information analysis, amplitude information reconstruction and three-dimensional image generation. Through this series of steps, an accurate three-dimensional model of the diseased area can be obtained, providing a numerical reference for subsequent disease assessment, which not only improves the accuracy of disease detection, but also reduces the dependence on expensive equipment, and improves the efficiency and economy of the entire detection process.

[0092] The technical solution of the embodiment of the present invention is illustrated below in conjunction with preferred implementation modes.

[0093] At present, the existing methods for underground disease image segmentation mainly include threshold-based segmentation, region growing method, edge detection, graph cut-based segmentation, and deep learning-based segmentation. The existing traditional three-dimensional visualization method mainly sets an amplitude threshold for the synthesized three-dimensional data body, only displays the reflection signal with a higher amplitude, and extracts the geometric structure and spatial distribution characteristics of the underground target body. However, since the threshold-based segmentation technology is sensitive to illumination changes and noise, when the image is disturbed by illumination changes or noise, the effect of threshold segmentation may be affected, resulting in inaccurate segmentation. For complex scenes where the grayscale difference between the target and the background is not obvious or there are multiple grayscale levels in the image, threshold segmentation may not be able to effectively distinguish different regions. The region growing method depends on the selection of seed points and the setting of similarity criteria, and may not accurately segment discontinuous or feature-similar regions. Although edge detection can provide information on the shape and contour of the target, it is easily affected by noise and image blur, resulting in inaccurate edge detection or the generation of pseudo edges. Although the segmentation based on graph cut can handle complex scenes, it has a large amount of calculation and may not be suitable for real-time processing or large-scale image data.

[0094] Moreover, the amplitude threshold setting of traditional 3D visualization imaging depends on the experience and judgment of the operating object. Different amplitude thresholds will result in different results. At the same time, the price of ground penetrating radar with visualization function is relatively expensive, often several times the price of ordinary ground penetrating radar, and the visualization effect is relatively general, which also depends on the experience and judgment of the interpreter. There is a technical problem that it is impossible to accurately detect underground diseases.

[0095] However, the embodiment of the present invention proposes a three-dimensional processing method for the segmentation and visualization of underground disease images, which improves the accuracy and robustness of segmentation by introducing a more advanced image segmentation algorithm, so that it can obtain a stable segmentation effect under different lighting conditions and noise environments. By utilizing the multi-layer, nonlinear feature extraction capabilities of the deep learning model, it aims to better handle the complex shapes and textures of underground diseases and achieve more accurate segmentation. Through automated algorithms and intelligent processing technology, human intervention is reduced, and the degree of automation of the entire process is improved, thereby improving work efficiency and accuracy.

[0096] Next, the underground disease image and three-dimensional visualization imaging technology in the embodiment of the present invention are further introduced.

[0097] In this embodiment, the image segmentation algorithm is improved based on the U-Net network structure and is named Crack U-Net. Figure 2 Schematic diagram of a Crack U-Net network structure according to an embodiment of the present invention. Figure 2 As shown in the figure, the Crack U-Net network is divided into two parts, the left half is the encoder and the right half is the decoder. The encoder part is divided into 5 parts, each of which contains a multi-way convolution-feature selection (MC-FS) module, a parallel multi-directional attention mechanism (PMDA) module and a residual compression (RS) module. The image size of the model input is 512×512, which is to facilitate downsampling by a multiple of 2 and to make the size as close to the original image size as possible to prevent feature loss caused by image reduction. The multi-way convolution-feature selection module in the network is placed in the first position of each module to increase the number of channels while extracting multiple features, and then select these features to ensure the validity of the input features in subsequent operations, thereby improving the encoding ability of the encoder. The model chooses to use a 3×3 convolution kernel for further feature extraction because the MC-FS module contains an expanded convolution and already has a large receptive field. At this time, using a small convolution kernel can reduce the overall parameter amount of the model. Then the parallel multi-directional attention mechanism module will establish an association relationship between the pixels of the features, thereby weighting the crack pixels and enhancing the network's ability to extract crack features. The features of the final output are subjected to maximum pooling to reduce the resolution of the features. The above steps are performed five times in a row to complete feature encoding. Starting from the last layer of the encoder, the output features of the last layer of the encoder are upsampled, and the resolution becomes twice the original to form the features of the decoder part, and are connected with the output features of the corresponding encoder. At this time, the features of the encoder will first enter the compression module for channel compression, remove unnecessary shallow features, and reduce the network size before connecting to prevent information loss, and then perform two feature extractions. The above steps are performed four times to complete decoding. The resolution of the decoded feature map is the same as the original. Figure 1The number of channels is the number of categories, and finally all channels of each pixel are activated to complete pixel-level classification.

[0098] Next, the process of image recognition using the Crack U-Net model is further introduced.

[0099] Figure 3 is a flow chart of a method for image recognition based on a Crack U-Net model according to an embodiment of the present invention. Figure 3 As shown, the method comprises the following steps:

[0100] Step S301: data preprocessing.

[0101] In this embodiment, the input underground disease image is preprocessed, including operations such as image enhancement and grayscale histogram equalization, so as to improve the quality and contrast of the image and provide a better basis for subsequent segmentation.

[0102] Step S302, constructing a Crack U-Net model.

[0103] In this embodiment, an improved U-Net model, namely Crack U-Net, is constructed. Based on U-Net, the model introduces a parallel multi-directional attention mechanism (PMDA), a multi-way convolution-feature selection module (MC-FS), and a residual compression (RS) module to enhance the network's perception of features of different scales and segmentation performance.

[0104] Alternatively, cracks usually have the following morphological features: cracks are generally slender and continuous. Therefore, in radar images, when a pixel is a crack pixel, there is a high probability that there is a crack pixel in its adjacent pixels; if this pixel is the background, its adjacent area is also likely to be the background. Therefore, establishing an association relationship between pixels based on this feature is more conducive to crack feature extraction. However, since the channel attention mechanism performs feature weighting from the dimension of the channel, it is not suitable for extracting crack features with an association relationship between pixels, and the current spatial attention mechanism does not include a pixel association mechanism. Therefore, a parallel multi-directional attention mechanism (PMDA) module is proposed to establish a pixel association mechanism to improve the network's ability to extract crack features.

[0105] Optionally, PMDA is a spatial attention mechanism used in the Crack U-Net encoder. Its principle is to decompose a 3×3 convolution kernel into 4 parts, each of which is a pixel point and its neighboring pixel points. Establish an association relationship. Figure 4 is a schematic diagram of a PDMA attention mechanism according to an embodiment of the present invention, such as Figure 4As shown in the figure, this module is divided into four paths. From top to bottom, for the input feature A of shape H×W×C, the first two paths use a 1×2 and a 2×1 convolution kernel to establish an association relationship between each pixel in feature A and its horizontal and vertical pixel respectively. These two convolutions obtain features B and C respectively. The third and fourth paths use a 2×2 convolution kernel with an expansion rate of 2 and a 1×1 convolution kernel. These two paths work together to establish an association relationship between each pixel and its four adjacent pixels on two diagonals. Therefore, the results of these two paths are added to obtain the fused feature D. Then, the connection result F of features B, C, and D is combined with a 2×2 convolution kernel to merge all channels of feature F to obtain a feature channel feature map G, and an activation function is used to convert its value into a weight in the (0, 1) interval. Finally, G is multiplied by the original feature map A to complete the pixel weighting in the feature map to obtain A′. Since only five convolution kernels are used and the convolution kernels are very small, the parallel multi-directional attention module is also a lightweight network module, which only requires adding a few parameters to the network and is easy to implement.

[0106] Alternatively, due to the complexity of road conditions and the noise generated by the ground penetrating radar itself, radar images will more or less have signal interference from other conditions and equipment noise interference, which may cause the network to incorrectly extract features of noise or other conditions, reducing the effectiveness of the extracted features and further reducing the network's segmentation ability. For this reason, a multi-channel convolution-feature selection module (MC-FS) is proposed to improve the effectiveness of the features extracted by the network. The principle of the MC-FS module is to use the diversity of features to ensure effectiveness, that is, first extract multiple features, and then weight the channels through the channel attention mechanism to select more effective features.

[0107] Optionally, the MC-FS module consists of parallel standard convolution, dilated convolution, and channel attention modules, and is used in the encoder of Crack U-Net, just like the PMDA module. Figure 5 is a structural diagram of a MC-FS module according to an embodiment of the present invention, such as Figure 5As shown in the figure, w, h, and c represent the width, height, and number of channels of the input feature, respectively, and C represents the number of output channels. The input feature is first subjected to four parallel convolutions, two of which are 3×3 and 2×2 dilated convolutions with a dilation rate of 2. The purpose is to obtain a larger receptive field with a small number of parameters. The addition of 3×3 and 2×2 standard convolutions avoids the chessboard effect of dilated convolutions and eliminates the discontinuity of features extracted by dilated convolutions and the problem of local feature loss. Therefore, for one input feature, the MC-FS module generates four features, and the number of channels of each feature is 1 / 4 of the output feature. After connecting these four features, the number of output channels is restored. Finally, the channel attention module is added to weight each channel to achieve feature selection and give higher weights to effective channels.

[0108] Optionally, the use of the above modules inevitably increases the number of network parameters, which affects the network's operating efficiency to a certain extent. In addition, the traditional U-Net has the following unreasonable feature connection design, that is, the decoder will connect the features of the corresponding layer encoder as a supplement, but the encoder is a shallow area compared to the decoder and contains a large amount of "rough" information. Connecting all these features will not only greatly increase the number of network parameters, but also make the decoder need to "refine" these features with unnecessary or "rough" information when supplementing information, which is tantamount to adding noise to deep features. Therefore, in order to reduce the number of network parameters and increase the effectiveness of the features connected from the encoder to the decoder in Crack U-Net, a residual compression module (RS) is proposed. The function of the RS module is to reduce the number of parameters of the network model by removing redundant feature channels, and is used for residual connection of shallow features of the encoder to the decoder.

[0109] Optionally, in Crack U-Net, the encoder features only supplement the decoder with appropriate feature information after passing through the RS module, without connecting all features into the decoder. Figure 6 This is a structural diagram of a residual compression module according to an embodiment of the present invention. After the feature enters the compression module, it will be pooled with a step size of 2, retaining the eigenvalue with the largest response, and then using unpooling to restore the resolution. The purpose of this is to retain the most effective features in space for the connection of the encoder. Finally, a 2×2 ordinary convolution is used to change the feature channel. The compression rate here is 0.25, that is, the number of channels is compressed to 1 / 4 of the original. Therefore, the RS module is also a lightweight network module.

[0110] Step S303: model training.

[0111] In this embodiment, a large amount of labeled underground disease image data is used to train the Crack U-Net model, and the model parameters are continuously adjusted by optimizing the loss function until a satisfactory segmentation effect is achieved.

[0112] Step S304: image segmentation.

[0113] In this embodiment, the preprocessed underground disease image is input into the trained Crack U-Net model for image segmentation. The model outputs the semantic category label of each pixel, thereby achieving accurate segmentation of the diseased area.

[0114] Step S305, post-processing.

[0115] In this embodiment, the recognition result is post-processed, such as extracting the circumscribed rectangle of the diseased area, recording the center point coordinates, length and width, etc., to facilitate subsequent analysis and processing.

[0116] In the above steps S301 to S305, by training the Crack U-Net model and then using the trained Crack U-Net model to segment the image, the semantic category label of each pixel in the image data can be obtained, thereby achieving accurate segmentation of the diseased area in the image data.

[0117] Optionally, after obtaining an accurate segmentation result of the diseased area in the image data, the segmentation result may be subjected to three-dimensional visualization imaging.

[0118] Next, the three-dimensional visualization imaging method is further introduced.

[0119] Figure 7 is a flow chart of a three-dimensional visualization imaging method according to an embodiment of the present invention. Figure 7 As shown, the method comprises the following steps:

[0120] Step S701, data acquisition.

[0121] In this embodiment, the abnormal area is re-detected, and the survey lines are arranged in a grid shape (generally arranged in a "well" shape); then a high dynamic ground penetrating radar is used to collect data to obtain two-dimensional radar image data.

[0122] Step S702, data processing.

[0123] In this embodiment, the 2D module of the application software performs simple static correction and gain processing on all survey line data.

[0124] Step S703: true amplitude reconstruction.

[0125] In this embodiment, based on the single-channel signal of the anomaly, the inverse scattering imaging method and the migration method are used to establish a true amplitude reconstruction algorithm to restore the amplitude energy of the wave field at the scattering source position, and then the interface of the anomaly is restored to its true state from point to surface based on the B-scan data.

[0126] Step S704: three-dimensional visualization.

[0127] In this embodiment, the B-scan data of different time depths, lateral position directions, and longitudinal position directions are fitted to complete the true interpolation of limited B-scan data, realize the true three-dimensional slice imaging of the abnormal body, and more accurately infer the burial depth and distribution of the abnormal body.

[0128] Step S705, result verification.

[0129] In this embodiment, the 3D visualization results are compared and analyzed with the drilling verification results to verify the accuracy and reliability of the 3D visualization technology.

[0130] In the above steps S701 to S705, the intuitiveness of the results and the scientific nature of the decision-making are enhanced through three-dimensional visualization, and at the same time, the reliability and application value of the entire detection process are improved through strict result verification.

[0131] According to an embodiment of the present invention, there is also provided an image data processing device. It should be noted that the image data processing device can be used to execute the image data processing method in the embodiment.

[0132] Figure 8 FIG. 1 is a schematic diagram of an image data processing device according to an embodiment of the present invention. Figure 8 As shown, the network configuration policy adjustment device 800 may include: a collection unit 801 , a pre-processing unit 802 , an identification unit 803 , a determination unit 804 and a processing unit 805 .

[0133] The acquisition unit 801 is used to acquire image data of underground disease areas;

[0134] The preprocessing unit 802 is used to perform image enhancement and grayscale histogram equalization operations on the image data to obtain preprocessed image data;

[0135] The recognition unit 803 is used to perform image recognition on the preprocessed image data to obtain a recognition result, wherein the recognition result includes semantic category labels corresponding to a plurality of pixels in the image data, the semantic category labels are used to indicate the diseased region category to which the corresponding pixels belong, and the diseased region category includes at least a diseased category and a non-disease category;

[0136] A determination unit 804, configured to determine a diseased area and a non-diseased area in the image data based on semantic category labels corresponding to a plurality of pixels in the recognition result;

[0137] The processing unit 805 is used to perform three-dimensional visualization processing on the target position of the diseased area in the underground diseased area to obtain a detection result of the diseased area, wherein the detection result is used to represent the three-dimensional structural spatial characteristics of the diseased area.

[0138] Optionally, the recognition unit 803 is also used to: input the preprocessed image data into a target defect detection model for image recognition to obtain a recognition result, wherein the target defect detection model is obtained by training an initial defect detection model with image data samples of the diseased area.

[0139] Optionally, the image data processing device 800 is also used to: obtain image data samples of diseased area samples; annotate the image data samples to obtain label data corresponding to the image data samples; use the image data samples and the label data corresponding to the image data samples to train an initial defect detection model to obtain a target defect detection model.

[0140] Optionally, the determination unit 804 is further used to: determine the area corresponding to the pixels whose semantic category labels are the disease category in the recognition result as the disease area; determine the area corresponding to the pixels whose semantic category labels are the non-disease category in the recognition result as the non-disease area.

[0141] Optionally, the processing unit 805 is also used to: extract geometric information of the diseased area, wherein the geometric information includes at least the coordinates of the center point of the diseased area and the size information of the diseased area; determine the target position of the diseased area in the underground diseased area based on the coordinates of the center point of the diseased area and the size information of the diseased area; collect image data at the target position; reconstruct the amplitude information of the image data at the target position at different depths based on an amplitude reconstruction algorithm, wherein the amplitude reconstruction algorithm is constructed based on inverse scattered imaging technology and offset technology; and determine the detection result of the diseased area based on the amplitude information.

[0142] In this embodiment, the image data of the underground diseased area is recognized to obtain a recognition result, which includes the recognition result of each pixel in the image data, and then the diseased area in the image data is determined according to the recognition result of each pixel, and then the diseased area is three-dimensionally visualized to determine the three-dimensional spatial distribution characteristics of the diseased area. That is to say, in the embodiment of the present invention, the recognition accuracy of the diseased area can be enhanced through pixel recognition and three-dimensional visualization, thereby solving the technical problem of poor detection accuracy of the underground diseased area.

[0143] According to an embodiment of the present invention, a computer-readable storage medium is further provided. The storage medium includes a stored program, wherein the program executes the image data processing method in the embodiment.

[0144] According to an embodiment of the present invention, a processor is further provided. The processor is used to run a program, wherein the method for processing image data in the embodiment is executed when the program is run.

[0145] According to another aspect of the embodiments of the present invention, a computer program product is provided, which includes computer instructions, and when the computer instructions are executed by a processor, the method for processing image data in the embodiments is implemented.

[0146] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0147] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0148] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0149] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed over multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0150] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0151] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent functional component, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software functional component, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, Read-Only Memory (ROM), Random Access Memory (RAM), mobile hard disk, magnetic disk or optical disk, etc., various media that can store program codes.

[0152] The above are only preferred embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for processing image data, characterized in that: include: Collect image data of underground diseased areas; Performing image enhancement and grayscale histogram equalization operations on the image data to obtain the preprocessed image data; Performing image recognition on the preprocessed image data to obtain a recognition result, wherein the recognition result includes semantic category labels corresponding to a plurality of pixels in the image data, the semantic category labels are used to indicate the diseased region category to which the corresponding pixels belong, and the diseased region category includes at least a diseased category and a non-disease category; Determine the diseased area and the non-disease area in the image data based on the semantic category labels corresponding to the multiple pixels in the recognition result; A three-dimensional visualization process is performed on the target position of the diseased area in the underground diseased area to obtain a detection result of the diseased area, wherein the detection result is used to represent the three-dimensional structural spatial characteristics of the diseased area.

2. The method according to claim 1, characterized in that Performing image recognition on the preprocessed image data to obtain a recognition result includes: The preprocessed image data is input into a target defect detection model for image recognition to obtain a recognition result, wherein the target defect detection model is obtained by training an initial defect detection model with image data samples of the defective area.

3. The method according to claim 2, characterized in that The target defect detection model includes an encoder and a decoder, wherein the encoder consists of a multi-way convolutional feature selection module, a parallel multi-directional attention mechanism and a residual compression module, the multi-way convolutional feature selection module is used to extract features of different scales in the image data, the parallel multi-directional attention mechanism is used to enhance the perception of features of different scales, the residual compression module is used to compress the features of different scales, and the decoder is used to reconstruct the image data according to the compressed features of different scales, and generate the recognition result according to the reconstructed image data.

4. The method according to claim 3, characterized in that The method further comprises: Obtain image data samples of diseased area samples; Annotating the image data sample to obtain label data corresponding to the image data sample; The initial defect detection model is trained using the image data samples and label data corresponding to the image data samples to obtain the target defect detection model.

5. The method according to claim 1, characterized in that Determining the diseased area and the non-diseased area in the image data based on the semantic category labels corresponding to the multiple pixels in the recognition result includes: Determine the area corresponding to the pixels whose semantic category label is the disease category in the recognition result as the disease area; The area corresponding to the pixels whose semantic category label is the non-disease category in the recognition result is determined as the non-disease area.

6. The method according to claim 1, characterized in that Performing three-dimensional visualization processing on the target position of the diseased area in the underground diseased area to obtain the detection result of the diseased area includes: Extracting geometric information of the diseased area, wherein the geometric information at least includes the coordinates of the center point of the diseased area and the size information of the diseased area; Determine a target position of the diseased area in the underground diseased area based on the coordinates of the center point of the diseased area and the size information of the diseased area; collecting image data at the target position; Reconstructing the amplitude information of the image data at the target position at different depths based on an amplitude reconstruction algorithm, wherein the amplitude reconstruction algorithm is constructed based on inverse scatter imaging technology and migration technology; Based on the amplitude information, a detection result of the diseased area is determined.

7. An image data processing device, characterized in that: include: A collection unit, used for collecting image data of underground disease areas; A preprocessing unit, used for performing image enhancement and grayscale histogram equalization operations on the image data to obtain the preprocessed image data; A recognition unit, configured to perform image recognition on the preprocessed image data to obtain a recognition result, wherein the recognition result includes semantic category labels corresponding to a plurality of pixels in the image data, the semantic category labels are used to indicate the diseased region category to which the corresponding pixels belong, and the diseased region category includes at least a diseased category and a non-disease category; A determination unit, configured to determine a diseased area and a non-diseased area in the image data based on semantic category labels corresponding to the plurality of pixels in the recognition result; A processing unit is used to perform three-dimensional visualization processing on the target position of the diseased area in the underground diseased area to obtain a detection result of the diseased area, wherein the detection result is used to represent the three-dimensional structural spatial characteristics of the diseased area.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed by a processor, the device where the storage medium is located is controlled to execute the method according to any one of claims 1 to 6.

9. A processor, characterized in that: The processor is used to run a program, wherein the program executes the method according to any one of claims 1 to 6 when running.

10. A computer program product, characterized in that The computer program product comprises computer instructions, and when the computer instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images