Cable intermediate joint defect detection method and system based on multi-modal X-ray

Abstract

The invention relates to the technical field of intelligent detection, and discloses a multi-modal X-ray-based cable intermediate joint defect detection method and system, and the method comprises the steps: carrying out the multi-modal X-ray data collection of a cable intermediate joint through a dual-energy-spectrum X-ray dynamic scanning protocol; preprocessing the acquired image data, wherein the preprocessing comprises contrast enhancement and noise suppression; performing three-dimensional reconstruction based on the preprocessed image data, and extracting defect features; and carrying out classification and identification on the extracted defect features by using an improved convolutional neural network. According to the method, the cable intermediate joint is detected by adopting an X-ray penetrating imaging mode at the cable intermediate joint, and the defects and defect hidden dangers of the cable intermediate joint can be effectively detected through methods of image processing, defect identification and the like.

Description

Technical Field

[0001] This invention relates to the field of intelligent inspection technology, and in particular to a method and system for detecting defects in cable joints based on multimodal X-rays. Background Technology

[0002] With the rapid development of modern society and economy, electricity demand is constantly increasing, and the scale of power grids is becoming increasingly massive. Cables, as a crucial component of power transmission, are widely used in urban power grids, industrial plants, and large buildings. Cable joints are weak points in cable lines, and their number increases rapidly with the expansion of cable lines. For example, in the construction of urban underground cable networks, a large number of joints are needed to extend cable lengths to meet the power supply needs of different areas. If these joints develop internal defects, it can affect the stable operation of the entire power grid.

[0003] Existing methods for detecting defects in cable joints have shortcomings, including low detection accuracy, difficulty in identifying minute defects, limited detection methods, inability to comprehensively cover various defect types, complex and time-consuming detection processes, low efficiency, susceptibility to interference, poor stability, and low detection and positioning accuracy. This invention overcomes these shortcomings by combining multi-energy spectral X-ray imaging, image enhancement algorithms, and convolutional neural networks (CNNs) to form an intelligent detection system for cable joint defects. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for detecting cable joint defects based on multimodal X-rays. It can acquire data through dynamic scanning using a dual-energy X-ray dynamic scanning protocol and a three-dimensional phase contrast imaging algorithm, combined with a convolutional neural network for intelligent identification and learning of cable joint defects, thus forming a set of intelligent cable joint defect detection systems.

[0006] To address the aforementioned technical problems, this invention provides the following technical solution: a method for detecting defects in cable joints based on multimodal X-rays, comprising: acquiring multimodal X-ray data of the cable joint using a dual-energy X-ray dynamic scanning protocol; preprocessing the acquired image data, including contrast enhancement and noise suppression; performing three-dimensional reconstruction based on the preprocessed image data and extracting defect features; and classifying and identifying the extracted defect features using an improved convolutional neural network.

[0007] As a preferred embodiment of the cable joint defect detection method based on multimodal X-rays described in this invention, the multimodal X-ray data acquisition includes using a dual-energy CT scanning mode, alternating between high-energy X-rays and low-energy X-rays for dynamic scanning, and combining a spiral scanning path with vibration compensation control to eliminate motion artifacts.

[0008] As a preferred embodiment of the cable joint defect detection method based on multimodal X-rays described in this invention, the preprocessing includes enhancing the contrast of the image by limiting contrast adaptive histogram equalization and suppressing noise in the image based on wavelet transform threshold denoising method.

[0009] As a preferred embodiment of the cable joint defect detection method based on multimodal X-rays described in this invention, the three-dimensional reconstruction includes enhancing edge information based on a phase contrast imaging algorithm and using an improved filtering back projection algorithm combined with dynamic weighting adjustment to complete the three-dimensional volume data reconstruction.

[0010] As a preferred embodiment of the cable joint defect detection method based on multimodal X-rays described in this invention, the defect feature extraction includes multi-level feature extraction of the reconstructed three-dimensional volume data by introducing a 3D residual network with a channel attention mechanism. The channel attention mechanism dynamically learns the feature channel weights through global average pooling and fully connected layers. The calibration tool corrects the offset between the detector and the rotation center, aligns the projection data, removes dark current and flat field noise, and enhances the clarity of the original projection. Based on the geometric relationship of the X-ray penetration path, the weight of the projection data is dynamically adjusted to reduce cone beam artifacts; high-frequency filtering is applied to the projection data to highlight edge details and suppress noise. The filtered projection data is reverse-mapped to the corresponding pixel positions in three-dimensional space, and bilinear or bicubic interpolation is used to fill in the data at non-integer pixel positions to improve resolution. Ring artifacts and motion blur are eliminated through frequency domain filtering or iterative correction, and nonlocal mean filtering or deep learning models are applied to reduce noise and enhance defect contrast.

[0011] As a preferred embodiment of the cable joint defect detection method based on multimodal X-rays described in this invention, the improved convolutional neural network includes: training using a labeled defect sample dataset; introducing a channel attention mechanism, compressing the global spatial information of each channel into a scalar through global average pooling, and learning the weights of each channel through a fully connected layer, amplifying the response of important channels, weakening secondary channels, and enabling the model to automatically focus on the key defect features in the cable joint; The channel attention mechanism weights are calculated as follows: Where z represents the feature after global pooling, and W1 and W2 are the weights of the fully connected layer. δ σ is a non-linear activation function, σ is the Sigmoid function, and S is the channel attention weight of the final output. The dataset contains five typical defect types, including air gaps, cracks, and impurities.

[0012] As a preferred embodiment of the cable joint defect detection method based on multimodal X-rays described in this invention, the improved convolutional neural network further includes updating the defect feature library periodically via the cloud and dynamically optimizing the classification weights of the convolutional neural network using incremental learning.

[0013] As a preferred embodiment of the cable joint defect detection system based on multimodal X-rays described in this invention, it includes a scanning control module, an image optimization module, a three-dimensional reconstruction module, a defect identification module, and a model update module. The scanning control module controls the dual-energy X-ray alternating scanning of the cable connector, and combines the spiral path and vibration compensation algorithm to eliminate motion artifacts and generate multimodal projection data. The image optimization module performs contrast-limited adaptive histogram equalization and wavelet threshold denoising on the scanned data to improve image contrast and suppress noise. The three-dimensional reconstruction module enhances edge information based on phase contrast imaging, completes three-dimensional volume data reconstruction using an improved filtering back projection algorithm, and improves resolution through dynamic weighted adjustment and bilinear interpolation. The defect identification module uses a 3D residual network with channel attention mechanism to extract multi-level features, and combines the trained convolutional neural network to classify five types of defects: air gaps, cracks, and impurities. The model update module obtains an incremental defect feature library from the cloud and dynamically optimizes the classification model weights, thereby continuously improving detection accuracy.

[0014] A computer device includes a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of a method for detecting defects in cable joints based on multimodal X-rays.

[0015] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a method for detecting defects in cable joints based on multimodal X-rays.

[0016] The beneficial effects of this invention are as follows: Dual-energy spectroscopy fusion effectively solves the problem of uneven penetration caused by differences in atomic number at copper-aluminum joints, improving detection accuracy; phase-contrast imaging increases sensitivity to micron-level cracks by more than three times. The detection speed is fast; the digital imaging system allows for rapid acquisition of detection results, helping to improve detection efficiency, shorten power outage time, and reduce the impact on power grid operation. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of a cable joint defect detection method based on multimodal X-rays provided in one embodiment of the present invention. Detailed Implementation

[0019] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0020] Example 1, referring to Figure 1 This is the first embodiment of the present invention, which provides a method for detecting defects in cable joints based on multimodal X-rays, including: S1: Multimodal X-ray data acquisition of cable joints using dual-energy X-ray dynamic scanning protocol.

[0021] Furthermore, dual-energy spectral CT scanning (160kV / 80kV) was employed, alternating between high-energy (160kV) and low-energy (80kV) X-rays to obtain projection data with different penetration characteristics; A spiral scanning path is set, and dynamic scanning control is performed. Combining spiral scanning with vibration compensation avoids motion artifacts. The spiral scanning path and vibration compensation can be expressed by the following equation: Where A = 0.1 rad (amplitude). (Vibration frequency), θ is the rotation angle, ω is the angular velocity of the helical scan, and t is time; X-ray attenuation model: Where I0 is the incident intensity, I is the transmitted intensity, and μ(E) is the mass decay coefficient of the material at energy E (unit: cm² / g). d is the material density, and d is the penetration thickness.

[0022] In this step, it is necessary to determine the material parameters (including the μ(E) value of the cable connector material, such as copper, aluminum, XLPE) and the scanning parameters (including tube voltage (kV), tube current (mA), and exposure time (ms)).

[0023] By using a spiral scanning path, the image data obtained by the device is processed and combined with material parameters and scanning parameters to improve the comprehensiveness and stability of data acquisition.

[0024] S2: Preprocess the acquired image data, including contrast enhancement and noise suppression.

[0025] Furthermore, contrast enhancement is achieved using Limited Contrast Adaptive Histogram Equalization (CLAHE), which can be expressed by the following formula: In the formula: This is the mapping function corresponding to the gray level; The value is the k-th gray level of the original image; k is the upper limit of the summation (meaning the calculation up to the current gray level). (At this time, it is necessary to accumulate the number of pixels at all gray levels from i=0 to i=k). grayscale The pixel count is MN, the received image is L, the total number of gray levels is i, and the summation index variable is i.

[0026] It should be noted that the threshold denoising based on wavelet transform (general threshold): Where σ is the standard deviation and N is the signal length.

[0027] Required data: Noise assessment (σ calculated using a blank projected image); Gray-level distribution (histogram statistics of the original image). Image data is processed by limiting contrast adaptive histogram equalization and noise suppression.

[0028] In an optional embodiment, noise suppression can be achieved based on local noise estimation. Specifically, the global noise standard deviation is calculated using a blank projection image, and the noise variance of each region is statistically analyzed using a local sliding window (e.g., 16×16 pixels) to generate a noise level distribution map. The image is divided into "high-noise areas" and "low-noise areas" based on the noise variance of each region. The high-noise area uses a 3D block matching (BM3D) framework to jointly filter in a set of non-local similar blocks and use inter-block redundancy to suppress strong noise. The low-noise area uses anisotropic diffusion filtering to smooth noise according to the gradient direction and avoid edge blurring.

[0029] In another alternative embodiment, noise suppression can be transformed into a low-dimensional projection problem. Specifically, the image is segmented into overlapping local blocks, a Hankel matrix is ​​constructed for each block, principal components are extracted through singular value decomposition (SVD) to form a low-rank subspace, the first u singular values ​​(u is dynamically determined by the noise standard deviation) are retained, and the higher-order components representing noise are truncated.

[0030] The truncated low-rank matrix is ​​back-projected into the image space to reconstruct the denoised blocks; the global image is then fused by weighted averaging of overlapping regions to eliminate block artifacts.

[0031] S3: Perform 3D reconstruction based on preprocessed image data and extract defect features.

[0032] Furthermore, phase contrast imaging enhances edge information by utilizing the refractive properties of X-rays. The phase difference calculation formula is as follows: In the formula: δ is the real part of the refractive index, β is the imaginary part (related to attenuation), λ is the X-ray wavelength, and z is the coordinate axis of the X-ray propagation direction.

[0033] The 3D volume data is reconstructed using the improved FDK algorithm combined with filtered back projection (FBP). The 3D ResNet feature extraction (residual block formula) is shown in the following equation: In the formula: x is the input feature, F is the convolutional layer function, y is the output, and W is the input feature. i For weights.

[0034] Image enhancement and defect feature extraction are performed on the imaging images through 3D reconstruction and feature extraction.

[0035] It should be noted that the specific steps are as follows: a. Data preprocessing: The offset between the detector and the rotation center is corrected using calibration tools to ensure the alignment of the projection data, remove dark current and flat field noise, and enhance the clarity of the original projection; b. Dynamic weighted adjustment: Based on the geometric relationship of the X-ray penetration path (such as the distance between the sample and the detector), the weight of the projection data is dynamically adjusted to reduce cone beam artifacts; high-frequency filtering is applied to the projection data to highlight edge details and suppress noise. c. Back projection reconstruction: The filtered projection data is back-mapped to the corresponding voxel positions in three-dimensional space, and bilinear or bicubic interpolation is used to fill in the data at non-integer pixel positions to improve resolution; d. Post-processing optimization: eliminate ring artifacts and motion blur through frequency domain filtering or iterative correction, and reduce noise and enhance defect contrast by applying nonlocal mean filtering or deep learning models.

[0036] In an optional embodiment, back-projection reconstruction is achieved through dynamic compensation based on sparse projection data, specifically, Before backprojection, adaptive interpolation is performed on the projection data of missing angles (due to scanning time / dose limitations). By analyzing the geometric similarity of projections of adjacent angles, a local low-rank matrix model is constructed to predict missing data and reduce fringe artifacts caused by incomplete projection.

[0037] During back-projection, directional constraints are applied to the interpolated projection data—the back-projection weights are dynamically adjusted based on the interpolation confidence level, and the weights are reduced in regions with low confidence levels to avoid introducing spurious structures.

[0038] In another alternative embodiment, back-projection reconstruction can also be achieved based on a physical imaging model. Specifically, back-projection is decomposed into two parallel branches: a projection domain branch and a spatial domain branch. The projection domain branch directly uses the original projection data to generate a rough voxel distribution; the spatial domain branch generates a probability distribution map by using a predefined material prior constraint range of voxel values.

[0039] By fusing the results of the two branches using the Alternating Direction Multiplier Method (ADMM), the projected data is forced to conform to the space physical constraints, and the backprojection error is gradually corrected.

[0040] S4: Use an improved convolutional neural network to classify and identify the extracted defect features.

[0041] Furthermore, a channel attention mechanism (SE module) is introduced, using a convolutional neural network to identify and classify defects in cable joints. Traditional convolutional neural networks perform poorly in filtering all feature channels; only some channels (such as crack edges and air gap textures) are truly useful for defect identification. Therefore, by introducing the channel attention mechanism (SE module), global average pooling is used to compress the global spatial information of each channel into a scalar, and the weights (0~1) of each channel are learned through fully connected layers. This amplifies the response of important channels and weakens secondary channels, allowing the model to automatically focus on key defect features in the cable joint (such as gradient changes in microcracks). The core function of this step is to enhance the expressive power of key defect features by dynamically learning the importance weights of different feature channels, while suppressing irrelevant noise interference. The SE module endows the network with "selective attention" capabilities, making the model closer to the attention mechanism of human vision and providing higher accuracy in defect identification.

[0042] The weights of the SE module are calculated as follows: In the formula: z is the feature after global pooling, and W1 and W2 are the weights of the fully connected layer.

[0043] When using convolutional neural networks to detect defects in cable joints, a defect training dataset is used. The model is trained by identifying at least 1,000 labeled defect samples and five typical defects, including air gaps, cracks, and impurities. At the same time, the data model is collected and updated during detection to achieve more accurate defect identification and classification.

[0044] Example 2 is an embodiment of the present invention, which provides a method for detecting defects in cable intermediate joints based on multimodal X-rays. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.

[0045] When using convolutional neural networks to detect defects in cable joints, a defect training dataset is used. The model is trained by identifying at least 1,000 labeled defect samples and five typical defects, including air gaps, cracks, and impurities. At the same time, the data model is collected and updated during detection to achieve more accurate defect identification and classification.

[0046] The above method enables efficient and accurate identification of defects in cable joints. Compared with traditional methods, this method achieves a defect identification accuracy of 94.2%, as shown in the table below:

[0047] After using CLAHE enhancement, the signal-to-noise ratio (SNR) increased from 22 dB to 35 dB, with an error of <3% compared to the traditional scanning results.

[0048] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0049] Example 3, the third embodiment of the present invention, differs from the previous two embodiments in that: If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0050] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0051] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0052] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0053] Example 4, an embodiment of the present invention, provides a cable joint defect detection system based on multimodal X-rays, including a scanning control module, an image optimization module, a three-dimensional reconstruction module, a defect identification module, and a model update module; The scanning control module controls the alternating scanning of the cable connector with dual-energy X-rays, and combines the spiral path and vibration compensation algorithm to eliminate motion artifacts and generate multimodal projection data. The image optimization module performs contrast-limited adaptive histogram equalization and wavelet threshold denoising on the scanned data to improve image contrast and suppress noise. The 3D reconstruction module enhances edge information based on phase contrast imaging, uses an improved filtered back projection algorithm to reconstruct 3D volume data, and improves resolution through dynamic weighted adjustment and bilinear interpolation. The defect identification module uses a 3D residual network with channel attention mechanism to extract multi-level features, and combines the trained convolutional neural network to classify five types of defects: air gaps, cracks, and impurities. The model update module obtains an incremental defect feature library from the cloud and dynamically optimizes the classification model weights, continuously improving detection accuracy.

[0054] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for detecting defects in cable joints based on multimodal X-rays, characterized in that: include, Multimodal X-ray data acquisition of cable joints was performed using a dual-energy X-ray dynamic scanning protocol. The acquired image data is preprocessed, including contrast enhancement and noise suppression; Three-dimensional reconstruction is performed based on the preprocessed image data, and defect features are extracted; An improved convolutional neural network is used to classify and identify the extracted defect features.

2. The method for detecting defects in cable joints based on multimodal X-rays as described in claim 1, characterized in that: The multimodal X-ray data acquisition includes using a dual-energy CT scanning mode, alternating between high-energy X-rays and low-energy X-rays for dynamic scanning, and combining a spiral scanning path with vibration compensation control to eliminate motion artifacts.

3. The method for detecting defects in cable joints based on multimodal X-rays as described in claim 2, characterized in that: The preprocessing includes enhancing the image contrast by limiting contrast adaptive histogram equalization and suppressing noise in the image.

4. The method for detecting defects in cable joints based on multimodal X-rays as described in claim 3, characterized in that: The three-dimensional reconstruction includes enhancing edge information based on phase contrast imaging and using an improved algorithm combined with dynamic weighted adjustment to complete the reconstruction of three-dimensional volume data.

5. The cable joint defect detection method based on multimodal X-rays as described in claim 4, characterized in that: The extracted defect features include, By introducing a channel attention mechanism, a 3D residual network is used to extract multi-level features from the reconstructed 3D volume data. The channel attention mechanism dynamically learns the feature channel weights through global average pooling and fully connected layers. The calibration tool corrects the offset between the detector and the rotation center, aligns the projection data, removes dark current and flat field noise, and enhances the clarity of the original projection. Based on the geometric relationship of the X-ray penetration path, the weight of the projection data is dynamically adjusted to reduce cone beam artifacts; high-frequency filtering is applied to the projection data to highlight edge details and suppress noise. The filtered projection data is reverse-mapped to the corresponding pixel positions in three-dimensional space, and bilinear or bicubic interpolation is used to fill in the data at non-integer pixel positions to improve resolution. By using frequency domain filtering or iterative correction, ring artifacts and motion blur are eliminated. Nonlocal mean filtering or deep learning models are applied to reduce noise and enhance defect contrast.

6. The cable joint defect detection method based on multimodal X-rays as described in claim 5, characterized in that: The improved convolutional neural network includes training using a labeled defect sample dataset; introducing a channel attention mechanism, which compresses the global spatial information of each channel into a scalar through global average pooling, and learns the weights of each channel through a fully connected layer, amplifying the response of important channels and weakening secondary channels, so that the model automatically focuses on the key defect features in the cable joint. The channel attention mechanism weights are calculated as follows: Where z represents the feature after global pooling, and W1 and W2 are the weights of the fully connected layer. δ σ is a non-linear activation function, σ is the Sigmoid function, and S is the channel attention weight of the final output. The dataset contains five typical defect types, including air gaps, cracks, and impurities.

7. The method for detecting defects in cable joints based on multimodal X-rays as described in claim 6, characterized in that: The improved convolutional neural network also includes regularly updating the defect feature library in the cloud and dynamically optimizing the classification weights of the convolutional neural network using incremental learning.

8. A system employing the multimodal X-ray-based defect detection method for cable joints as described in any one of claims 1 to 7, characterized in that: It includes a scanning control module, an image optimization module, a 3D reconstruction module, a defect identification module, and a model update module; The scanning control module controls the dual-energy X-ray alternating scanning of the cable connector, and combines the spiral path and vibration compensation algorithm to eliminate motion artifacts and generate multimodal projection data. The image optimization module performs contrast-limited adaptive histogram equalization and wavelet threshold denoising on the scanned data to improve image contrast and suppress noise. The three-dimensional reconstruction module enhances edge information based on phase contrast imaging, completes three-dimensional volume data reconstruction using an improved filtering back projection algorithm, and improves resolution through dynamic weighted adjustment and bilinear interpolation. The defect identification module uses a 3D residual network with channel attention mechanism to extract multi-level features, and combines the trained convolutional neural network to classify five types of defects: air gaps, cracks, and impurities. The model update module obtains an incremental defect feature library from the cloud and dynamically optimizes the classification model weights, thereby continuously improving detection accuracy.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images