Contact network part detection method and device, terminal and storage medium

Abstract

The invention provides an overhead line system part detection method and device, a terminal and a storage medium. The method comprises the following steps: acquiring a low-illumination image and a normal-illumination image obtained by shooting the same position of a high-speed railway overhead line system; inputting the low-illumination image and the normal-illumination image into a trained intermediate domain migration model, and respectively migrating the low-illumination image and the normal-illumination image to an intermediate domain through the intermediate domain migration model to obtain a first natural image and a second natural image; and inputting the first natural image and the second natural image into a trained target detection model, and processing to obtain a contact network part detection result corresponding to each natural image. According to the method, the low-illumination image and the normal-illumination image are migrated to the middle domain to obtain the corresponding natural image, so that distribution migration and alignment of data in different image domains can be effectively realized, and the detection precision of the overhead line system parts is further improved.

Description

Technical Field

[0001] The present invention relates to the field of computer vision technology, and in particular to a contact network component detection method, device, terminal and storage medium. Background Art

[0002] The catenary component status detection and monitoring system is an important technical guarantee for the safety of high-speed railway operations. Since the catenary inspection vehicle needs to perform tasks at night to avoid conflicts with trains operating during the day, the image acquisition system will inevitably be affected by uneven illumination, which will lead to a decrease in the quality of the catenary image in low-illumination environments, thereby affecting the recognition accuracy of components.

[0003] At present, the basic solution for low-light catenary image detection is to enhance the image quality by increasing the image brightness. However, this method cannot effectively solve the data distribution difference between normal illumination images and low illumination images, resulting in low recognition accuracy of catenary parts.

[0004] Therefore, the prior art has defects and needs to be improved and developed. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a contact network component detection method, device, terminal and storage medium in view of the above-mentioned defects of the prior art, aiming to solve the problem of low recognition accuracy of contact network components in the prior art.

[0006] The technical solution adopted by the present invention to solve the technical problem is as follows:

[0007] In a first aspect, an embodiment of the present invention provides a method for detecting contact network components, the method comprising:

[0008] Obtain low-illumination images and normal-illumination images obtained by photographing the same position of the high-speed railway contact network;

[0009] Inputting the low illumination image and the normal illumination image into a trained intermediate domain migration model, and migrating the low illumination image and the normal illumination image to an intermediate domain respectively through the intermediate domain migration model to obtain a first natural image and a second natural image;

[0010] The first natural image and the second natural image are input into a trained target detection model, and the contact network component detection results corresponding to each natural image are obtained through processing.

[0011] In one embodiment, the intermediate domain migration model includes an input module, a first image inversion module, an image decomposition module, a medium transmission image refinement module, a second image inversion module and an output module; the low illumination image and the normal illumination image are input into the trained intermediate domain migration model, and the low illumination image and the normal illumination image are respectively migrated to the intermediate domain by the intermediate domain migration model to obtain the first natural image and the second natural image, including:

[0012] Inputting the low illumination image and the normal illumination image into the input module, and transmitting the low illumination image and the normal illumination image to the first image inversion module via the input module;

[0013] The first image inversion module converts the low illumination image and the normal illumination image into corresponding fog images to obtain a first fog image and a second fog image, and transmits the first fog image and the second fog image to an image decomposition module;

[0014] Processing the first fog image and the second fog image by the image decomposition module to obtain a first medium transmission image corresponding to the first fog image and a second medium transmission image corresponding to the second fog image;

[0015] Inputting the first medium transmission image and the second medium transmission image into a medium transmission image refinement module, and performing data correction by the medium transmission image refinement module to obtain a first natural image to be inverted and a second natural image to be inverted;

[0016] The first natural image to be inverted and the second natural image to be inverted are transmitted to the second image inversion module, and the second image inversion module performs inversion processing on the images to output the first natural image and the second natural image;

[0017] The first natural image and the second natural image are outputted via the output module.

[0018] In one implementation, the converting the low illumination image and the normal illumination image into corresponding fog images by the first image inversion module to obtain the first fog image and the second fog image includes:

[0019] The low illumination image and the normal illumination image are respectively normalized by the first image inversion module to obtain a first normalized image and a second normalized image;

[0020] The first normalized image and the second normalized image are respectively inverted to obtain a first fog image and a second fog image.

[0021] In one embodiment, the image decomposition module includes a multi-representation module, a multi-scale module, a multi-scale transmission constraint module and a nonlinear regression module; the first fog image and the second fog image are processed by the image decomposition module to obtain a first medium transmission image corresponding to the first fog image and a second medium transmission image corresponding to the second fog image, including:

[0022] The first fog image and the second fog image are received by the multi-representation module, and the input images are subjected to feature extraction and dimensionality reduction processing by using a plurality of convolutional layers and a maximum pooling layer in the module, and the extracted features are concatenated to obtain a first feature set and a second feature set;

[0023] The first feature set and the second feature set are transmitted to the multiscale module, and the multiscale module uses a preset number of scale kernels to extract multiscale features of the first feature set and the second feature set respectively to obtain first multiscale features and second multiscale features, and the first multiscale features and the second multiscale features are transmitted to the multiscale transmission constraint module;

[0024] The multi-scale transmission constraint module performs multi-scale pooling, convolution feature extraction and upsampling operations on the first multi-scale feature and the second multi-scale feature in sequence to obtain a first feature map and a second feature map, and transmits the first feature map and the second feature map to the nonlinear regression module;

[0025] The first feature map and the second feature map are nonlinearly transformed by the nonlinear regression module to obtain a first medium transmission image corresponding to the first fog image and a second medium transmission image corresponding to the second fog image.

[0026] In one embodiment, the medium transmission image refinement module includes a mean filtering module and a local linear module; the first medium transmission image and the second medium transmission image are input into the medium transmission image refinement module, and the medium transmission image refinement module performs data correction to obtain a first natural image to be inverted and a second natural image to be inverted, including:

[0027] Inputting the first medium transmission image and the second medium transmission image into a mean filter module, and performing a dilated convolution operation through the mean filter module to obtain a third medium transmission image and a fourth medium transmission image;

[0028] The third medium transmission image and the fourth medium transmission image are transmitted to the local linear module, and the third medium transmission image and the fourth medium transmission image are processed by the local linear module to obtain a first natural image to be inverted and a second natural image to be inverted.

[0029] In one embodiment, the training step of the intermediate domain migration model includes:

[0030] Acquire training set data, wherein the training set data includes a plurality of image sets photographing a high-speed railway contact network, each image set including a low-illumination training image and a normal-illumination training image at the same position;

[0031] Initialize the intermediate domain training transfer model to be trained;

[0032] Obtaining a preset first loss function, inputting all of the low-illumination training images and the normal-illumination training images into the intermediate domain training transfer model in batches, and adjusting parameters in the image decomposition module of the intermediate domain training transfer model by using a gradient descent method to minimize the first loss function;

[0033] After completing the first preset round of training for the image decomposition module, a first target medium transmission training image and a second target medium transmission training image corresponding to each of the image sets are obtained;

[0034] Obtaining a preset second loss function, inputting all of the first target media transmission training images and the second target media transmission training images into the media transmission image refinement module of the intermediate domain training transfer model in batches, and adjusting the parameters of the media transmission image refinement module by using a gradient descent method to minimize the second loss function;

[0035] After the second preset round of training of the medium transmission image refinement module is completed, a trained intermediate domain migration model is obtained.

[0036] In one embodiment, the step of inputting the first natural image and the second natural image into a trained target detection model and processing to obtain a contact network component detection result corresponding to each natural image includes:

[0037] Inputting the first natural image and the second natural image into a trained object detection model;

[0038] The multi-layer neural network structure in the target detection model performs feature extraction and analysis on the first natural image and the second natural image to obtain the first contact network component detection result corresponding to the first natural image and the second contact network component detection result corresponding to the second natural image.

[0039] In a second aspect, an embodiment of the present invention further provides a contact network component detection device, comprising:

[0040] An image acquisition module is used to acquire low-illumination images and normal-illumination images obtained by photographing the same position of the high-speed railway contact network;

[0041] A natural image generation module, configured to input the low illumination image and the normal illumination image into a trained intermediate domain migration model, and migrate the low illumination image and the normal illumination image to an intermediate domain respectively through the intermediate domain migration model to obtain a first natural image and a second natural image;

[0042] The result generation module is used to input the first natural image and the second natural image into a trained target detection model, and obtain the contact network component detection results corresponding to each natural image through processing.

[0043] In a third aspect, an embodiment of the present invention further provides a terminal, comprising: a memory, a processor, and a contact network component detection program stored in the memory and executable on the processor, wherein the contact network component detection program implements the steps of the contact network component detection method as described above when executed by the processor.

[0044] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a contact network component detection program, and the contact network component detection program can be executed to implement the steps of the contact network component detection method as described above.

[0045] Beneficial effects of the present invention: The present invention obtains a low-light image and a normal-light image taken at the same position of a high-speed railway contact network; inputs the low-light image and the normal-light image into a trained intermediate domain migration model, and migrates the low-light image and the normal-light image to the intermediate domain respectively through the intermediate domain migration model to obtain a first natural image and a second natural image; inputs the first natural image and the second natural image into a trained target detection model, and obtains contact network component detection results corresponding to each natural image through processing. The present invention can effectively achieve distribution migration and alignment of data in different image domains by migrating low-light images and normal-light images to the intermediate domain to obtain corresponding natural images, thereby improving the detection accuracy of contact network components. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a flow chart of a preferred embodiment of the contact network component detection method in the present invention.

[0047] Figure 2 It is a schematic diagram of imaging of the observation system in a foggy environment in the present invention.

[0048] Figure 3 This is the image alignment diagram in the present invention. Figure 1 .

[0049] Figure 4 This is the image alignment diagram in the present invention. Figure 2 .

[0050] Figure 5 It is a schematic diagram of using the middle domain to perform image alignment in the present invention.

[0051] Figure 6 It is a structural schematic diagram of the intermediate domain migration model of the present invention.

[0052] Figure 7 It is a flow chart for natural image generation in the present invention.

[0053] Figure 8 It is a schematic diagram of a normal illumination image of the contact network in the present invention.

[0054] Fig. 9 It is a schematic diagram of a low-light image of the contact network in the present invention.

[0055] Fig.10 It is a schematic diagram of a normal illumination image of the contact network after migration through the intermediate domain in the present invention.

[0056] Fig.11 It is a schematic diagram of a low-light image of the contact network after migration through the intermediate domain in the present invention.

[0057] Fig.12 It is a schematic diagram of the detection results of a contact network component in the present invention.

[0058] Fig.13 It is a schematic diagram of the process of detecting contact network parts in the present invention.

[0059] Fig.14 It is a structural schematic diagram of a preferred embodiment of the contact network component detection device in the present invention.

[0060] Fig.15 It is a terminal principle block diagram of the present invention. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solution and advantages of the present invention clearer and more specific, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0062] The catenary component status detection and monitoring system is an important technical guarantee for the safety of high-speed railway operations. Since the catenary inspection vehicle needs to perform tasks at night to avoid conflicts with trains operating during the day, the image acquisition system will inevitably be affected by uneven illumination, which will lead to a decrease in the quality of the catenary image in low-illumination environments, thereby affecting the recognition accuracy of components.

[0063] At present, the basic solution for low-light catenary image detection is to enhance the image quality by increasing the image brightness. However, this method cannot effectively solve the data distribution difference between normal illumination images and low illumination images, resulting in low recognition accuracy of catenary parts.

[0064] In view of the above-mentioned defects of the prior art, the present invention provides a contact network component detection method, device, terminal and storage medium, which belongs to the field of computer vision technology. The method includes: obtaining a low-light image and a normal-light image obtained by shooting the same position of the high-speed railway contact network; inputting the low-light image and the normal-light image into a trained intermediate domain migration model, and migrating the low-light image and the normal-light image to the intermediate domain respectively through the intermediate domain migration model to obtain a first natural image and a second natural image; inputting the first natural image and the second natural image into a trained target detection model, and obtaining the contact network component detection result corresponding to each natural image through processing. The present invention can effectively realize the distribution migration and alignment of data in different image domains by migrating the low-light image and the normal-light image to the intermediate domain to obtain the corresponding natural image, thereby improving the detection accuracy of contact network components.

[0065] See also Figure 1 The contact network component detection method described in the embodiment of the present invention comprises the following steps:

[0066] Step S100: acquiring a low-illumination image and a normal-illumination image obtained by photographing the same position of a high-speed railway contact network.

[0067] Specifically, in order to avoid affecting the normal operation of the train, the present invention uses a dedicated comprehensive train inspection vehicle to image the same position of the high-speed railway contact network at night to obtain low-light images and normal-light images. Among them, the low-light image is an image with insufficient fill light, and the normal-light image is an image with sufficient fill light.

[0068] See also Figure 1 The contact network component detection method described in the embodiment of the present invention further includes the following steps:

[0069] Step S200: input the low illumination image and the normal illumination image into a trained intermediate domain migration model, and migrate the low illumination image and the normal illumination image to an intermediate domain respectively through the intermediate domain migration model to obtain a first natural image and a second natural image.

[0070] Specifically, in the prior art, the basic solution for low-illumination contact network image detection is to enhance image quality by increasing image brightness. However, this simple brightness enhancement method cannot effectively solve the data distribution difference between low-illumination images and normal-illumination images. In addition, there is a method based on contrast adjustment to improve the accuracy of damage detection by enhancing railway track images, but this method relies on manual experience and fails to take into account the distribution difference between low-illumination images and normal-illumination images. The intermediate domain migration model of the present invention can effectively migrate low-illumination images and normal-illumination images to the intermediate domain to achieve distribution migration and alignment of data in different image domains. Subsequently, the migrated first natural image and the second natural image are used to detect contact network components. When the data distribution is aligned, the detection model can better focus on the characteristics of the contact network components themselves for analysis and identification, avoiding misjudgment caused by data differences, thereby effectively improving the detection accuracy.

[0071] In one implementation, the intermediate domain migration model includes an input module, a first image inversion module, an image decomposition module, a medium transmission image refinement module, a second image inversion module and an output module; the low illumination image and the normal illumination image are input into the trained intermediate domain migration model, and the low illumination image and the normal illumination image are respectively migrated to the intermediate domain by the intermediate domain migration model to obtain a first natural image and a second natural image, including:

[0072] Inputting the low illumination image and the normal illumination image into the input module, and transmitting the low illumination image and the normal illumination image to the first image inversion module via the input module;

[0073] The first image inversion module converts the low illumination image and the normal illumination image into corresponding fog images to obtain a first fog image and a second fog image, and transmits the first fog image and the second fog image to an image decomposition module;

[0074] Processing the first fog image and the second fog image by the image decomposition module to obtain a first medium transmission image corresponding to the first fog image and a second medium transmission image corresponding to the second fog image;

[0075] Inputting the first medium transmission image and the second medium transmission image into a medium transmission image refinement module, and performing data correction by the medium transmission image refinement module to obtain a first natural image to be inverted and a second natural image to be inverted;

[0076] The first natural image to be inverted and the second natural image to be inverted are transmitted to the second image inversion module, and the second image inversion module performs inversion processing on the images to output the first natural image and the second natural image;

[0077] The first natural image and the second natural image are outputted via the output module.

[0078] Specifically, the intermediate domain migration model of the present invention is constructed based on the atmospheric scattering model, dark channel theory and guided filtering model.

[0079] Since the photos taken at night are highly similar to fog images after being reversed, the present invention performs defogging on fog images based on the atmospheric scattering model to enhance low-illumination images and normal-illumination images, which is beneficial to the detection of contact network components.

[0080] The mathematical formulas corresponding to the imaging process of foggy weather images and their key elements are shown in formula (1).

[0081] I(x)=R(x)t(x)+A(x)(1-t(x)) (1)

[0082] Among them, I(x) is the fog image, R(x) is the natural image, t(x) is the medium transfer function, and A(x) is the natural light coefficient.

[0083] like Figure 2 As shown in the figure, the quality of the observation system imaging deteriorates in foggy environment, which is mainly attributed to the following two aspects: first, in foggy environment, the reflected light of the object is absorbed and scattered by the suspended particles in the atmosphere, which weakens the reflected light of the object and leads to low illumination of the observation system imaging result. This process corresponds to the R(x)t(x) part of formula (1); second, in foggy environment, the ambient illumination such as sunlight is scattered by the scattering medium in the atmosphere to form background illumination, which leads to blurred imaging result of the observation system. This process corresponds to the A(x)(1-t(x)) part of formula (1).

[0084] The dark channel theory uses statistics to collect statistics on a large number of outdoor haze-free images and finds that among the three RGB color channels of each pixel, there is always a channel with a very low brightness value, even close to zero. As shown in formula (2). The medium transfer function t(x) can be approximated by the dark channel as t(x)∝1-R dark (x). Therefore, when decomposing an image, the dark channel theory can be used to estimate and solve the medium transfer function.

[0085]

[0086] Where c represents the RGB channels of the image, Ω r (x) represents the image region at position x in the image with radius r, R c (y) represents the image Ω r (x) The image in the region, R dark(x) in the image Ω r (x) An image composed of the lowest values ​​of the RGB three-pass pixels in the region.

[0087] The construction of the image decomposition module of the present invention is based on the atmospheric scattering model and dark channel theory, and is mainly used to decompose the media transmission image in the fog image. However, the media transmission image initially decomposed often has the problem of not being smooth enough, resulting in the natural image not being smooth enough. In view of this, a media transmission image refinement module is introduced to fine-tune the media transmission image, so that the media transmission image is smoother and more natural, and then a smoother natural image is decomposed. In the present invention, the media transmission image refinement module is constructed based on the guided filtering model.

[0088] Guided filtering assumes that a point on a function is linearly related to the points in its neighborhood. Similarly, for an image, it can be regarded as a two-dimensional function. Based on this, it is assumed that the output and input of the guided filtering function satisfy a linear relationship in a two-dimensional window, as shown below:

[0089]

[0090] q i =p i -n i (4)

[0091] in, and is the parameter obtained by linear regression in the kth local window of the image, I i is the guide image (the low-light image itself and the normal-light image itself used here), q i is the filtered reconstructed image to be output, ω k is the kth local window in the image, p i is the input image to be filtered, n i is the noise, i is the subscript.

[0092] The next core task is to find the coefficient of the linear function, which is essentially a linear regression problem. Since we hope to minimize the difference between the output value q of the fitting function and the true value p, we transform it into an optimization problem, that is, to minimize the following formula (5).

[0093]

[0094] Among them, ε is the regularization coefficient, which prevents the model parameter a from being too large, thereby ensuring the stability and effectiveness of the model.

[0095] In the prior art, there are the following methods for low-light images: First, do not use empirical alignment of the source domain, as shown in the schematic diagram Figure 3As shown; second, use the source domain for reverse migration alignment, the schematic diagram is as follows Figure 4 As shown. In the figure, the source domain is a normal illumination image, and the target domain is a low illumination image. Both cannot solve the problem of data distribution migration and alignment in different image domains. However, the present invention uses the atmospheric scattering model and the intermediate image domain as a bridge to migrate low illumination images and normal illumination images to the same area, which can quickly and efficiently achieve the migration and alignment of data in different image domains, thereby effectively solving the problem of low detection accuracy of contact network components caused by inconsistent data distribution of low illumination images and normal illumination images. The schematic diagram is shown in FIG. Figure 5 shown.

[0096] The intermediate domain migration model of the present invention is constructed based on the actual physical model (atmospheric scattering model and guided filtering model), and the intermediate migration network model of the present invention has higher interpretability and generalization. In addition, since the construction of the intermediate domain is based on a physical model with image enhancement function, it can not only migrate the data distribution of low-illumination images and normal-illumination images, but also improve the detection accuracy of normal-illumination images while improving the detection accuracy of low-illumination contact network parts, thereby achieving a comprehensive improvement in the overall data detection accuracy. It fills the gap in contact network detection, especially under low-illumination conditions, and significantly improves the detection accuracy.

[0097] In one implementation, converting the low illumination image and the normal illumination image into corresponding fog images by the first image inversion module to obtain a first fog image and a second fog image includes:

[0098] The low illumination image and the normal illumination image are respectively normalized by the first image inversion module to obtain a first normalized image and a second normalized image;

[0099] The first normalized image and the second normalized image are respectively inverted to obtain a first fog image and a second fog image.

[0100] Specifically, considering that the images taken at night show a high similarity with foggy images after the inversion operation, the low-light image and the normal-light image are first normalized to obtain a first normalized image and a second normalized image. Normalization can standardize the pixel values ​​and other data of the image to a specific range. After this step, the first normalized image and the second normalized image are inverted, respectively. Through such an inversion operation, they can be converted into images with similar features to the foggy images, that is, the first fog image and the second fog image required by the present invention. Subsequently, based on these fog images, the present invention can extract natural images from them through a specific processing method, which can then be used in subsequent related processes such as contact network component detection.

[0101] In one implementation, the image decomposition module includes a multi-representation module, a multi-scale module, a multi-scale transmission constraint module and a nonlinear regression module; the image decomposition module processes the first fog image and the second fog image to obtain a first medium transmission image corresponding to the first fog image and a second medium transmission image corresponding to the second fog image, including:

[0102] The first fog image and the second fog image are received by the multi-representation module, and the input images are subjected to feature extraction and dimensionality reduction processing by using a plurality of convolutional layers and a maximum pooling layer in the module, and the extracted features are concatenated to obtain a first feature set and a second feature set;

[0103] The first feature set and the second feature set are transmitted to the multiscale module, and the multiscale module uses a preset number of scale kernels to extract multiscale features of the first feature set and the second feature set respectively to obtain first multiscale features and second multiscale features, and the first multiscale features and the second multiscale features are transmitted to the multiscale transmission constraint module;

[0104] The multi-scale transmission constraint module performs multi-scale pooling, convolution feature extraction and upsampling operations on the first multi-scale feature and the second multi-scale feature in sequence to obtain a first feature map and a second feature map, and transmits the first feature map and the second feature map to the nonlinear regression module;

[0105] The first feature map and the second feature map are nonlinearly transformed by the nonlinear regression module to obtain a first medium transmission image corresponding to the first fog image and a second medium transmission image corresponding to the second fog image.

[0106] Specifically, the structure of the intermediate domain migration network of the present invention is as follows: Figure 6 As shown. After receiving the first fog image and the second fog image, they are transmitted to the multi-representation module, and the multi-representation module performs a convolution operation on the fog image. In order to better extract features, the multi-representation module of the present invention adopts a 4-channel convolutional feature extraction network to obtain a first feature set and a second feature set. In the multi-scale module, in order to solve the problem of image target scale invariance, 3 scale kernels are used to extract multi-scale features to obtain first multi-scale features and second multi-scale features. In the multi-scale transmission constraint module, in order to solve the problem of image target spatial invariance, the present invention introduces multi-scale pooling, convolution feature extraction and upsampling operations to ensure the spatial invariance and scale invariance of the target. In the nonlinear regression module, its output value is limited to between 0 and 1. This limitation helps to ensure that the gradient converges more easily, thereby improving the training efficiency and stability of the entire model. Based on this, the present invention uses a double boundary constraint function to constrain the output, as shown in formula (6).

[0107] F4=min(t max ,max(t min ,W4*F3+B4)) (6)

[0108] Among them, W4 and B4 are the model parameters of the nonlinear regression module, F3 is the feature output from the multi-scale transmission constraint module, and t min =0,t max = 1. This setting enables the nonlinear regression module to run accurately within the established framework, providing a high-quality data basis for subsequent processing procedures. After processing by the nonlinear regression module, the first medium transmission image corresponding to the first fog image and the second medium transmission image corresponding to the second fog image can be output.

[0109] In one implementation, the medium transmission image refinement module includes a mean filtering module and a local linear module; the first medium transmission image and the second medium transmission image are input into the medium transmission image refinement module, and the medium transmission image refinement module performs data correction to obtain a first natural image to be inverted and a second natural image to be inverted, including:

[0110] Inputting the first medium transmission image and the second medium transmission image into a mean filter module, and performing a dilated convolution operation through the mean filter module to obtain a third medium transmission image and a fourth medium transmission image;

[0111] The third medium transmission image and the fourth medium transmission image are transmitted to the local linear module, and the third medium transmission image and the fourth medium transmission image are processed by the local linear module to obtain a first natural image to be inverted and a second natural image to be inverted.

[0112] Specifically, after the medium transmission image is decomposed, the decomposed natural image R low (natural image decomposed from low-light image) and R nor (Natural image decomposed from normal illumination image) is not smooth enough and needs to be further optimized. Therefore, the medium transmission image refinement module constructed by guided filtering theory is used to filter the medium transmission function image t(x) to obtain a more refined image. So that the decomposed natural image R low and R nor Smoother and more natural.

[0113] In the optimization process, first, the mean filter module is used to perform grouped dilated convolution, which can effectively denoise the image while maintaining the edge information of the image, making the image texture clearer and the features more prominent. Through this step, the third medium transmission image and the fourth medium transmission image can be obtained. Then, the local linear module is used to process it. Specifically, the local linear module includes three local linearizers. The first local linearizer is used to calculate various average values, correlation calculations, and variance and covariance. Its calculation formula (7) is shown below.

[0114]

[0115] Among them, f mean Represents the average operation performed on the data, mean I is the mean of the image transmitted by the third medium, mean p is the mean value of the image transmitted by the fourth medium, corr I is the correlation calculation of the image transmitted by the third medium, corr Ip is the correlation calculation between the third medium transmission image and the fourth medium transmission image, var I is the variance calculation of the image transmitted by the third medium, cov Ip is the covariance calculation of the third medium image and the fourth medium image.

[0116] The second local linearizer is used to calculate the correlation coefficient in formula (3) to obtain the filtered image. The formula (8) used is as follows:

[0117]

[0118] According to the relationship between the medium transfer function and the dark channel Usually, the value of 0.1% of the brightest value of the dark channel is selected to estimate A. Due to the particularity of the overhead line image: the image is taken at night, and nearly half of each image is the sky, so the A value can be set to 1. and A, formula (1) can be transformed into formula (9):

[0119]

[0120] The third linearizer is based on equation (9) and The natural image R(x) to be inverted can be obtained.

[0121] The flowchart for natural image generation in the present invention is as follows: Figure 7As shown. It can be seen that after the image is inverted, the medium transmission image can be obtained by processing with the multi-representation module, the multi-scale module, the multi-scale transmission constraint module and the nonlinear regression module, and then the image is optimized by the mean filter module and the three local linearizers, and the image is inverted to obtain the final natural image (i.e., the intermediate domain migration image). It can be understood that since the first image inversion module inverts the original image into a foggy image, the second image inversion module is set to restore the natural image to be inverted decomposed from the foggy image to an image in a natural state.

[0122] In one embodiment, the training step of the intermediate domain migration model includes:

[0123] Acquire training set data, wherein the training set data includes a plurality of image sets photographing a high-speed railway contact network, each image set including a low-illumination training image and a normal-illumination training image at the same position;

[0124] Initialize the intermediate domain training transfer model to be trained;

[0125] Obtaining a preset first loss function, inputting all of the low-illumination training images and the normal-illumination training images into the intermediate domain training transfer model in batches, and adjusting parameters in the image decomposition module of the intermediate domain training transfer model by using a gradient descent method to minimize the first loss function;

[0126] After completing the first preset round of training for the image decomposition module, a first target medium transmission training image and a second target medium transmission training image corresponding to each of the image sets are obtained;

[0127] Obtaining a preset second loss function, inputting all of the first target media transmission training images and the second target media transmission training images into the media transmission image refinement module of the intermediate domain training transfer model in batches, and adjusting the parameters of the media transmission image refinement module by using a gradient descent method to minimize the second loss function;

[0128] After the second preset round of training of the medium transmission image refinement module is completed, a trained intermediate domain migration model is obtained.

[0129] Specifically, in order to ensure the fineness of the final output image, the present invention trains the intermediate domain migration model. Before training, a training data set is first constructed. The training set data contains multiple image sets of high-speed railway contact networks, each image set contains low-light training images and normal-light training images at the same position. During the training process, the image decomposition module is first trained, and then the medium transmission image refinement module is trained.

[0130] The first loss function corresponding to the image decomposition module is constructed based on the medium transmission function image t(x) estimated based on the dark channel hypothesis, and is expressed as:

[0131]

[0132] Among them, t i (x) = imerode (1-I i (x)), imerode is the image erosion operation. Θ is the sum of the image decomposition module parameters. i is the index, I i (x) is the i-th input sample image, N is the number of samples for each input training, and L is the calculated loss. λ(Ii(x); Θ) represents the output of the image decomposition module. During training, the model parameters are optimized by minimizing the first loss function, that is, minimizing the gap between the output of the image decomposition module and t(x) estimated by dark channel theory.

[0133] The second loss function corresponding to the media transmission image refinement module is expressed as:

[0134] L G =argmin(E(a k ,b k )+MMD(F,X,Y)) (11)

[0135] in,

[0136] Specifically, argmin(E(a k ,b k )) The corresponding formula is formula (5), x∈X is the low-light image of the contact network, y∈Y is the normal-light image of the contact network, and f∈F is the Gaussian kernel function operation. n and m are the number of input samples of each corresponding low-light image and normal-light image. The natural image R decomposed in the previous step low (natural image decomposed from low-light image) and R nor (Natural images decomposed from normal illumination images) cannot be guaranteed to have the same distribution. Therefore, formula (5) is further introduced into MMD loss to obtain formula (11). By minimizing the distribution distance between low illumination and normal illumination images of the contact network during training, the distribution of low illumination and normal illumination images of the contact network can be further reduced in practical applications. This process is to migrate low illumination images and normal illumination images to the intermediate domain. The second objective loss function is calculated by the gradient descent algorithm, and the parameters of the medium transmission image refinement module can be optimized to obtain the correlation function a→A map Substituting b→B into formula (3), we get the following formula (13):

[0137]

[0138] When it is used formally later, the medium transmission image can be obtained using formula (13):

[0139] When both modules are trained, a trained intermediate domain migration network is obtained. The present invention introduces MMD loss in the second loss function to make the illumination distribution of the low-illumination image closer to that of the normal illumination image, thereby reducing the illumination difference in the low-illumination image and making the distribution of the two images more consistent. In this way, the accuracy of subsequent contact network component detection can be effectively improved.

[0140] In the present invention, the normal illumination images of several overhead lines are as follows: Figure 8 As shown in Figure 2, low-light images of several contact networks are shown in Figure 2. Fig. 9 As shown in FIG. 1 , after being processed by the intermediate domain migration model, it can be seen that the normal illumination image (i.e., the second natural image) after the intermediate domain migration is as follows: Fig.10 As shown in FIG. 1 , the low-light image (i.e., the first natural image) after intermediate domain migration is as follows: Fig.11 As shown in Figure 2, the distribution of the two images can be made more consistent through the processing of the intermediate domain transfer model.

[0141] See also Figure 1 The contact network component detection method described in the embodiment of the present invention further includes the following steps:

[0142] Step S300: input the first natural image and the second natural image into a trained target detection model, and obtain the contact network component detection results corresponding to each natural image through processing.

[0143] Specifically, the target detection model of the present invention is a Faster-RCNN network model. The gradient descent method is used to train it, so that the deep learning network can locate and classify the contact network components more accurately. Fig.12 , is a schematic diagram of the detection results of a contact network component in the present invention. Fig.12 In the figure, 1 is the insulator base, 2 is the insulator, 3 is the support sleeve, 4 is the support sleeve tightening bolt, 5 is the rotating double ear, 6 is the positioner base, 7 is the inclined wire positioning hook, 8 is the double sleeve connector, 9 is the load-bearing cable base, and 10 is the windproof wire ring. Fig.12 This is only a schematic diagram of component detection results, and does not mean that the present invention can only support the component detection as described above.

[0144] In one implementation, the step of inputting the first natural image and the second natural image into a trained target detection model and obtaining a contact network component detection result corresponding to each natural image through processing includes:

[0145] Inputting the first natural image and the second natural image into a trained object detection model;

[0146] The multi-layer neural network structure in the target detection model performs feature extraction and analysis on the first natural image and the second natural image to obtain the first contact network component detection result corresponding to the first natural image and the second contact network component detection result corresponding to the second natural image.

[0147] Specifically, the multi-layer neural network structure in the Faster-RCNN network model is used to process natural images, and the catenary component detection results corresponding to each natural image can be effectively obtained. The training data set of the Faster-RCNN network model is obtained by annotating the component samples of the collected images using the image annotation tool LabelMe.

[0148] In summary, the present invention obtains a low-light image and a normal-light image taken at the same position of the high-speed railway contact network; inputs the low-light image and the normal-light image into a trained intermediate domain migration model, and migrates the low-light image and the normal-light image to the intermediate domain respectively through the intermediate domain migration model to obtain a first natural image and a second natural image; inputs the first natural image and the second natural image into a trained target detection model, and obtains the contact network component detection results corresponding to each natural image through processing. The process can be as follows: Fig.13 The present invention can effectively achieve the distribution migration and alignment of data in different image domains by migrating low-illumination images and normal-illumination images to the middle domain to obtain corresponding natural images, thereby improving the detection accuracy of contact network components.

[0149] In one embodiment, if Fig.14 As shown, based on the above-mentioned contact network component detection method, the present invention also provides a contact network component detection device, including:

[0150] An image acquisition module 100 is used to acquire a low-illumination image and a normal-illumination image obtained by photographing the same position of the high-speed railway contact network;

[0151] A natural image generation module 200, configured to input the low illumination image and the normal illumination image into a trained intermediate domain migration model, and migrate the low illumination image and the normal illumination image to an intermediate domain respectively through the intermediate domain migration model to obtain a first natural image and a second natural image;

[0152] The result generation module 300 is used to input the first natural image and the second natural image into a trained target detection model, and obtain the contact network component detection result corresponding to each natural image through processing.

[0153] In one embodiment, the intermediate domain migration model includes an input module, a first image inversion module, an image decomposition module, a medium transmission image refinement module, a second image inversion module and an output module; the natural image generation module includes:

[0154] a first image input unit, configured to input the low illumination image and the normal illumination image into the input module, and transmit the low illumination image and the normal illumination image to the first image inversion module via the input module;

[0155] A first image inversion unit, configured to convert the low illumination image and the normal illumination image into corresponding fog images via the first image inversion module to obtain a first fog image and a second fog image, and transmit the first fog image and the second fog image to an image decomposition module;

[0156] an image decomposition unit, configured to process the first fog image and the second fog image through the image decomposition module to obtain a first medium transmission image corresponding to the first fog image and a second medium transmission image corresponding to the second fog image;

[0157] An image optimization unit, configured to input the first medium transmission image and the second medium transmission image into a medium transmission image refinement module, and perform data correction via the medium transmission image refinement module to obtain a first natural image to be inverted and a second natural image to be inverted;

[0158] A second image inversion unit, configured to transmit the first natural image to be inverted and the second natural image to be inverted to the second image inversion module, and the second image inversion module inverts the images to output the first natural image and the second natural image;

[0159] The first image output unit is used to output the first natural image and the second natural image via the output module.

[0160] In one embodiment, the device further comprises:

[0161] a normalization unit, configured to perform normalization processing on the low illumination image and the normal illumination image respectively through the first image inversion module to obtain a first normalized image and a second normalized image;

[0162] The inversion subunit is used to perform inversion processing on the first normalized image and the second normalized image respectively to obtain a first fog image and a second fog image.

[0163] In one embodiment, the image decomposition module includes a multi-representation module, a multi-scale module, a multi-scale transmission constraint module and a nonlinear regression module; the device also includes:

[0164] A feature extraction unit, configured to receive the first fog image and the second fog image through the multi-representation module, perform feature extraction and dimensionality reduction processing on the input images using a plurality of convolutional layers and a maximum pooling layer in the module, and concatenate the extracted features to obtain a first feature set and a second feature set;

[0165] a multi-scale feature extraction unit, configured to transmit the first feature set and the second feature set to the multi-scale module, wherein the multi-scale module uses a preset number of scale kernels to respectively extract multi-scale features of the first feature set and the second feature set to obtain first multi-scale features and second multi-scale features, and transmit the first multi-scale features and the second multi-scale features to the multi-scale transmission constraint module;

[0166] a feature map generating unit, configured to perform multi-scale pooling, convolution feature extraction and upsampling operations on the first multi-scale feature and the second multi-scale feature in sequence through the multi-scale transmission constraint module to obtain a first feature map and a second feature map, and transmit the first feature map and the second feature map to the nonlinear regression module;

[0167] A nonlinear transformation unit is used to perform nonlinear transformation on the first feature map and the second feature map through the nonlinear regression module to obtain a first medium transmission image corresponding to the first fog image and a second medium transmission image corresponding to the second fog image.

[0168] In one embodiment, the medium transmission image refinement module includes a mean filtering module and a local linear module, and the device further includes:

[0169] a dilated convolution unit, configured to input the first medium transmission image and the second medium transmission image into a mean filter module, and perform a dilated convolution operation on the mean filter module to obtain a third medium transmission image and a fourth medium transmission image;

[0170] The natural image generation unit is used to transmit the third medium transmission image and the fourth medium transmission image to the local linear module, and the local linear module processes the third medium transmission image and the fourth medium transmission image to obtain a first natural image to be inverted and a second natural image to be inverted.

[0171] In one embodiment, the device further comprises:

[0172] A training set acquisition unit, used to acquire training set data, wherein the training set data includes a plurality of image sets photographing a high-speed railway contact network, each image set including a low-illumination training image and a normal-illumination training image at the same position;

[0173] An initialization unit, used to initialize the intermediate domain training transfer model to be trained;

[0174] A first training unit is configured to obtain a preset first loss function, input all of the low-illumination training images and the normal-illumination training images into the intermediate domain training transfer model in batches, and adjust parameters in an image decomposition module of the intermediate domain training transfer model by using a gradient descent method to minimize the first loss function;

[0175] A training image generating unit, configured to obtain a first target medium transmission training image and a second target medium transmission training image corresponding to each of the image sets after completing a first preset round of training on the image decomposition module;

[0176] a second training unit, configured to obtain a preset second loss function, input all of the first target media transmission training images and the second target media transmission training images into a media transmission image refinement module of the intermediate domain training transfer model in batches, and adjust parameters of the media transmission image refinement module by using a gradient descent method to minimize the second loss function;

[0177] The training completion unit is used to obtain a trained intermediate domain migration model after completing a second preset round of training for the media transmission image refinement module.

[0178] In one embodiment, the result generation module includes:

[0179] A second image input unit, configured to input the first natural image and the second natural image into a trained object detection model;

[0180] The result output unit is used to extract and analyze the features of the first natural image and the second natural image through the multi-layer neural network structure in the target detection model to obtain the first contact network component detection result corresponding to the first natural image and the second contact network component detection result corresponding to the second natural image.

[0181] Based on the above embodiment, the present invention further provides a terminal, whose principle block diagram can be shown as follows: Fig.15As shown. The terminal includes a processor, a memory, a network interface and a display screen connected via a device bus. The processor of the terminal is used to provide computing and control capabilities. The memory of the terminal includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating device and a contact network component detection program. The internal memory provides an environment for the operation of the operating device and the contact network component detection program in the non-volatile storage medium. The network interface of the terminal is used to communicate with an external terminal via a network connection. When the contact network component detection program is executed by the processor, the steps of any one of the above-mentioned contact network component detection methods are implemented. The display screen of the terminal can be a liquid crystal display screen or an electronic ink display screen.

[0182] Those skilled in the art will understand that Fig.15 The principle block diagram shown in the figure is only a block diagram of a partial structure related to the scheme of the present invention, and does not constitute a limitation on the terminal to which the scheme of the present invention is applied. The specific terminal may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0183] In one embodiment, a terminal is provided, which includes a memory, a processor, and a contact network component detection program stored in the memory and executable on the processor. When the contact network component detection program is executed by the processor, the steps of any one of the contact network component detection methods provided in the embodiments of the present invention are implemented.

[0184] An embodiment of the present invention further provides a computer-readable storage medium, on which a contact network component detection program is stored. When the contact network component detection program is executed by a processor, the steps of any one of the contact network component detection methods provided in the embodiments of the present invention are implemented.

[0185] It should be understood that the sequence numbers of the steps in the above embodiments do not imply a sequence of execution, and the execution sequence of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0186] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the above-mentioned device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention. The specific working process of the units and modules in the above-mentioned device can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0187] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0188] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0189] In the embodiments provided by the present invention, it should be understood that the disclosed apparatus / terminal equipment and method can be implemented in other ways. For example, the apparatus / terminal equipment embodiments described above are only illustrative, for example, the division of the above modules or units is only a logical function division, and in actual implementation, other division methods can be used, for example, multiple units or components can be combined or integrated into another apparatus, or some features can be ignored or not executed.

[0190] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, it should be understood by those skilled in the art that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features therein may be replaced by equivalents. However, these modifications or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A method for detecting contact network components, characterized in that: The method comprises: Obtain low-illumination images and normal-illumination images obtained by photographing the same position of the high-speed railway contact network; Inputting the low illumination image and the normal illumination image into a trained intermediate domain migration model, and migrating the low illumination image and the normal illumination image to an intermediate domain respectively through the intermediate domain migration model to obtain a first natural image and a second natural image; The first natural image and the second natural image are input into a trained target detection model, and the contact network component detection results corresponding to each natural image are obtained through processing.

2. The method for detecting contact network components according to claim 1, characterized in that: The intermediate domain migration model includes an input module, a first image inversion module, an image decomposition module, a medium transmission image refinement module, a second image inversion module and an output module; the low illumination image and the normal illumination image are input into the trained intermediate domain migration model, and the low illumination image and the normal illumination image are respectively migrated to the intermediate domain by the intermediate domain migration model to obtain a first natural image and a second natural image, including: Inputting the low illumination image and the normal illumination image into the input module, and transmitting the low illumination image and the normal illumination image to the first image inversion module via the input module; The first image inversion module converts the low illumination image and the normal illumination image into corresponding fog images to obtain a first fog image and a second fog image, and transmits the first fog image and the second fog image to an image decomposition module; Processing the first fog image and the second fog image by the image decomposition module to obtain a first medium transmission image corresponding to the first fog image and a second medium transmission image corresponding to the second fog image; Inputting the first medium transmission image and the second medium transmission image into a medium transmission image refinement module, and performing data correction by the medium transmission image refinement module to obtain a first natural image to be inverted and a second natural image to be inverted; The first natural image to be inverted and the second natural image to be inverted are transmitted to the second image inversion module, and the second image inversion module performs inversion processing on the images to output the first natural image and the second natural image; The first natural image and the second natural image are outputted via the output module.

3. The contact network component detection method according to claim 2, characterized in that: The step of converting the low illumination image and the normal illumination image into corresponding fog images by the first image inversion module to obtain a first fog image and a second fog image includes: The low illumination image and the normal illumination image are respectively normalized by the first image inversion module to obtain a first normalized image and a second normalized image; The first normalized image and the second normalized image are respectively inverted to obtain a first fog image and a second fog image.

4. The method for detecting contact network components according to claim 2, characterized in that: The image decomposition module includes a multi-representation module, a multi-scale module, a multi-scale transmission constraint module and a nonlinear regression module; the first fog image and the second fog image are processed by the image decomposition module to obtain a first medium transmission image corresponding to the first fog image and a second medium transmission image corresponding to the second fog image, including: The first fog image and the second fog image are received by the multi-representation module, and the input images are subjected to feature extraction and dimensionality reduction processing by using a plurality of convolutional layers and a maximum pooling layer in the module, and the extracted features are concatenated to obtain a first feature set and a second feature set; The first feature set and the second feature set are transmitted to the multiscale module, and the multiscale module uses a preset number of scale kernels to extract multiscale features of the first feature set and the second feature set respectively to obtain first multiscale features and second multiscale features, and the first multiscale features and the second multiscale features are transmitted to the multiscale transmission constraint module; The multi-scale transmission constraint module performs multi-scale pooling, convolution feature extraction and upsampling operations on the first multi-scale feature and the second multi-scale feature in sequence to obtain a first feature map and a second feature map, and transmits the first feature map and the second feature map to the nonlinear regression module; The first feature map and the second feature map are nonlinearly transformed by the nonlinear regression module to obtain a first medium transmission image corresponding to the first fog image and a second medium transmission image corresponding to the second fog image.

5. The method for detecting contact network components according to claim 2, characterized in that: The medium transmission image refinement module includes a mean filter module and a local linear module; the first medium transmission image and the second medium transmission image are input into the medium transmission image refinement module, and the medium transmission image refinement module performs data correction to obtain a first natural image to be inverted and a second natural image to be inverted, including: Inputting the first medium transmission image and the second medium transmission image into a mean filter module, and performing a dilated convolution operation through the mean filter module to obtain a third medium transmission image and a fourth medium transmission image; The third medium transmission image and the fourth medium transmission image are transmitted to the local linear module, and the third medium transmission image and the fourth medium transmission image are processed by the local linear module to obtain a first natural image to be inverted and a second natural image to be inverted.

6. The method for detecting contact network components according to claim 1, characterized in that: The training steps of the intermediate domain migration model include: Acquire training set data, wherein the training set data includes a plurality of image sets photographing a high-speed railway contact network, each image set including a low-illumination training image and a normal-illumination training image at the same position; Initialize the intermediate domain training transfer model to be trained; Obtaining a preset first loss function, inputting all of the low-illumination training images and the normal-illumination training images into the intermediate domain training transfer model in batches, and adjusting parameters in the image decomposition module of the intermediate domain training transfer model by using a gradient descent method to minimize the first loss function; After completing the first preset round of training for the image decomposition module, a first target medium transmission training image and a second target medium transmission training image corresponding to each of the image sets are obtained; Obtaining a preset second loss function, inputting all of the first target media transmission training images and the second target media transmission training images into the media transmission image refinement module of the intermediate domain training transfer model in batches, and adjusting the parameters of the media transmission image refinement module by using a gradient descent method to minimize the second loss function; After the second preset round of training of the medium transmission image refinement module is completed, a trained intermediate domain migration model is obtained.

7. The method for detecting contact network components according to claim 1, characterized in that: The step of inputting the first natural image and the second natural image into a trained target detection model and obtaining a contact network component detection result corresponding to each natural image through processing includes: Inputting the first natural image and the second natural image into a trained object detection model; The multi-layer neural network structure in the target detection model performs feature extraction and analysis on the first natural image and the second natural image to obtain the first contact network component detection result corresponding to the first natural image and the second contact network component detection result corresponding to the second natural image.

8. A contact network component detection device, characterized in that: include: An image acquisition module is used to acquire low-illumination images and normal-illumination images obtained by photographing the same position of the high-speed railway contact network; A natural image generation module, configured to input the low illumination image and the normal illumination image into a trained intermediate domain migration model, and migrate the low illumination image and the normal illumination image to an intermediate domain respectively through the intermediate domain migration model to obtain a first natural image and a second natural image; The result generation module is used to input the first natural image and the second natural image into a trained target detection model, and obtain the contact network component detection results corresponding to each natural image through processing.

9. A terminal, characterized in that: The terminal includes a memory, a processor, and a contact network component detection program stored in the memory and executable on the processor. When the contact network component detection program is executed by the processor, the steps of the contact network component detection method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a contact network component detection program, and when the contact network component detection program is executed by the processor, the steps of the contact network component detection method according to any one of claims 1 to 7 are implemented.

Images