Road segmentation method for remote sensing images based on topology-aware neural network

Abstract

The present invention belongs to the technical field of remote sensing image semantic segmentation methods, specifically a remote sensing image road segmentation method based on topology-aware neural network. The method comprises the following steps: 1) dividing a high-resolution remote sensing data set into a training set and a test set, and preprocessing the remote sensing images in the training set and the test set respectively; 2) building an image semantic segmentation network model; 3) inputting the training set into the semantic segmentation network, first initializing the semantic segmentation network, then updating the parameters in the model, and optimizing the loss function until convergence; 4) inputting the test set data into the trained generator module to obtain a high-precision semantic segmentation result. The present invention solves the problems existing in the existing remote sensing image road automatic extraction technology, such as insufficient feature expression capability, inability to accurately perceive the topological structure of the road, context information being limited to a single training sample, large loss of edge and detail information in the semantic segmentation process, and a large number of samples required for training.

Description

Technical Field

[0001] The invention belongs to the technical field of remote sensing image semantic segmentation methods, and specifically is a remote sensing image road segmentation method based on a topology-aware neural network. Background Art

[0002] With the continuous launch of high-resolution remote sensing satellites, the increasing application of drone technology, and the continuous improvement of the spatial resolution of remote sensing images, earth observation technology has entered the era of high-resolution remote sensing big data. In high-resolution remote sensing big data, road network information not only directly reflects the economic and social development level of a country or region, but also has important application value in traffic management, urban planning, automatic navigation and other fields.

[0003] Traditional high-resolution remote sensing image road segmentation mainly relies on manual annotation, which is not only time-consuming and laborious, but also highly subjective. In order to improve detection efficiency, various object-based semantic segmentation methods have been continuously proposed and have achieved certain application results. However, these technologies rely heavily on low-level semantic features designed manually, such as texture, geometry, and edges, and cannot be well applied to complex scenes such as high-resolution remote sensing images. Therefore, they have poor generalization performance and are easily affected by various noise interferences.

[0004] The fully convolutional neural network based on deep learning can automatically obtain high-level semantic features from remote sensing images through a multi-layer network structure and nonlinear transformation. In addition, the end-to-end, pixel-to-pixel implementation of the fully convolutional neural network can provide pixel-level road recognition and positioning, and is currently the most promising method for extracting roads from remote sensing images. However, the fully convolutional neural network currently used for road extraction from remote sensing images mainly focuses on pixel classification accuracy, cannot accurately perceive the topological structure of the road, and the edge and detail information is lost in the process of semantic segmentation, and is easily affected by occlusion or shadows, which seriously affects the accuracy and integrity of the results. In addition, the fully convolutional model often relies on a large number of high-precision training samples or pre-trained models trained from a large number of samples of daily scenes. However, pixel-level annotation of remote sensing images is often time-consuming and laborious. At the same time, due to the large difference between daily scene images and remote sensing images, the segmentation accuracy of the pre-trained model trained from daily scene images is often unsatisfactory. In addition, the huge number of parameters in these complex models places high demands on storage and computing devices, and training and applying models are also very time-consuming. These defects greatly limit the practical application of these methods. Summary of the invention

[0005] The present invention aims to comprehensively solve the problems that the existing high-resolution remote sensing image road segmentation methods cannot accurately perceive the topological structure, the edge and detail information are lost greatly during the semantic segmentation process, they are easily disturbed by occlusion and shadows, the model is inefficient and difficult to train. A remote sensing image road segmentation method based on a topology-aware neural network is provided.

[0006] The present invention adopts the following technical scheme: a high-resolution remote sensing image semantic segmentation network, including a downsampling path, a spatial context module and an upsampling path, the downsampling path includes a deformable convolution layer and 5 continuous downsampling units, the input data first passes through the deformable convolution layer to obtain a semantic feature map, then passes through 5 continuous downsampling units for feature extraction and downsampling, and finally outputs a feature map; the upsampling path is composed of 5 continuous upsampling units, an A aggregation operation, a convolution layer and a Softmax layer in sequence, and finally generates a segmentation prediction map; the 5 downsampling units and 5 upsampling units correspond one to one, and are connected by a jump connection adjusted by the attention of a channel attention module; the spatial context module performs segmentation and fusion on the feature map output by the downsampling path and outputs it to the upsampling path.

[0007] In some embodiments, the deformable convolution layer is a deformable convolution with a convolution kernel size of 3×3 and a stride of 1, which adds a learnable offset variable and weight coefficient to the position of each sampling point in the 3×3 convolution kernel.

[0008] In some embodiments, the downsampling unit includes a primary aggregation module, an aggregation operation, and a down-conversion module, wherein the primary aggregation module is responsible for extracting features, and then the aggregation operation aggregates the input and output of the primary aggregation module, and then transmits the resulting feature map to the down-conversion module and the channel attention module respectively, wherein the output of the channel attention module is multiplied by the input by corresponding elements to obtain the jump connection features adjusted by the channel attention and input to the corresponding part of the upsampling path.

[0009] In some embodiments, the upsampling unit consists of a group of up-conversion modules, B aggregation operations and one-time aggregation modules, wherein the up-conversion module is responsible for restoring the spatial resolution of the feature map through upsampling, the B aggregation operation is responsible for aggregating the jump connection feature map adjusted by the channel attention module and the feature map obtained by the up-conversion module, and the one-time aggregation module is responsible for extracting features from the results of the B aggregation operation.

[0010] In some embodiments, the structure of the primary aggregation module is as follows: After the feature map I0 is input into the primary aggregation module, it first passes through n convolution modules to obtain n new feature maps I1, I2, ... n , where the first two convolution modules are deformable convolution modules; then the obtained n feature maps I1, I2…I nA channel stacking operation is performed to obtain a feature map J0 after channel stacking; the convolution module is composed of a batch normalization layer, a ReLU activation function layer, a 3×3 convolution layer and a random dropout layer in sequence; the deformable convolution module is composed of a batch normalization layer, a ReLU activation function layer, a 3×3 deformable convolution layer and a random dropout layer in sequence.

[0011] In some embodiments, the structure of the channel attention module is: the input features of size C′×H′×W′ are respectively subjected to global maximum pooling and global average pooling to obtain two C′×1×1 feature maps; then they are respectively sent to a shared multilayer perceptron: the number of neurons in the first layer is C′ / r, r is the reduction rate, r=16, the activation function is Relu, and the number of neurons in the second layer is C′; the two output features are added, and then subjected to Sigmoid activation operation to generate the final channel attention; the channel attention and the input features perform corresponding element multiplication operations.

[0012] In some embodiments, the spatial context module includes two paths: "height direction" and "width direction";

[0013] The “height direction” path comes from the feature map of the downsampling path, with a size of H×WC, where C represents the number of channels, H represents the feature map height, and W represents the feature map width. It is divided into H slices in the vertical direction: P1=P 11 , P 12 …, P 1H , for each slice P 1i , 1≤i≤H, use a 1×1 convolution operation to perform linear projection to obtain F i , then T i Input the first linear layer M k , get attention A i , and then focus on A i Softmax and L1 regularization are used in the feature map size and channel dimension respectively, and then the regularized attention is input into the second linear layer M v , and get a new slice P′ 1i , and finally aggregate all new slices along the “height direction” into a feature map of size C×HW;

[0014] In the “width direction” path, the feature map from the downsampling path is divided into W slices along the “width” direction, and the slices are updated and aggregated into feature maps in the same way;

[0015] The feature maps from both the “height direction” and “width direction” paths, both with a size of H×W×C, are fused through addition operations.

[0016] A remote sensing image road segmentation method based on topology-aware neural network includes the following steps:

[0017] S100: Divide the high-resolution remote sensing dataset into a training set and a test set, and preprocess the images in the training set and the test set respectively;

[0018] S200: Building a high-resolution remote sensing image semantic segmentation network;

[0019] S300: input the preprocessed training set images in S100 into the high-resolution remote sensing image semantic segmentation network in S200 for training. First, the high-resolution remote sensing image segmentation network is initialized using the He Uniform method, and then the parameters in the model are updated to optimize the loss function until convergence.

[0020] S400: Input the preprocessed test set remote sensing images into the segmentation network trained in S200, and output the semantic segmentation results of the high-resolution remote sensing images.

[0021] In step S100, the preprocessing includes manual image annotation, image cropping and data enhancement;

[0022] The manual image annotation specifically includes: manually annotating the roads in the high-resolution image at the pixel level in ArcGIS software to obtain a labeled surface crack image;

[0023] The image cropping is specifically as follows: randomly cropping the labeled high-resolution remote sensing image into sub-images of 512 pixels×512 pixels;

[0024] The data enhancement includes: performing random scale transformation on sub-images, random angle image rotation, and vertical and horizontal flipping of images to obtain a high-resolution remote sensing image.

[0025] In step S300, the loss function adopts Dice Loss, which is specifically:

[0026]

[0027] Where X represents the set of predicted values ​​corresponding to all pixels in the prediction image, Y represents the set of values ​​corresponding to all pixels in the label image, |XIY| represents the intersection of X and Y, and |X| and |Y| represent the number of elements in X and Y.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] 1) The present invention adopts a spatial context module, which converts the traditional layer-by-layer convolution connection form into a slice-by-slice convolution form in the feature map, so that information can be transmitted between pixel rows and columns in the map. Therefore, even when the road is blocked or other appearance clues are poor, the spatial context module is still suitable for detecting crack targets with long-distance continuous shapes due to its powerful topological morphology perception ability.

[0030] 2) The spatial context module in the present invention uses two linear layers with memory function to replace the Key vector and Value vector in the traditional self-attention mechanism, which can not only obtain global context information from all samples in the entire training set, but also reduce the computational complexity.

[0031] 3) The present invention uses wavelet transform and inverse wavelet transform in the down-conversion module and up-conversion module in the network to implement down-sampling and up-sampling to further avoid the loss of details and edge information in the model, thereby improving the model performance.

[0032] 4) In the downsampling and upsampling paths of the present invention, a one-time aggregation module is used to extract road features. This design makes the model more reusable and reduces redundant calculations, so the model is easy to train, does not require a pre-trained model, and reduces the number of samples required for model training; at the same time, the one-time aggregation module is also combined with deformable convolution, which is conducive to obtaining accurate road geometry features and enhancing topology perception capabilities.

[0033] 5) The jump connection of the present invention adopts a channel attention module, which is conducive to selecting appropriate channel information for fusion and avoiding interference, so as to better improve the semantic segmentation accuracy of the model.

[0034] Through the above beneficial effects, the present invention comprehensively solves the problems existing in the existing high-resolution remote sensing image road extraction methods, such as the inability to accurately perceive the topological structure of the road, the acquisition of global context information is limited to a single sample, the loss of edge and detail information in the semantic segmentation process is large, and the number of samples required for training is large. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A schematic diagram of the overall structure of a high-resolution remote sensing image semantic segmentation network constructed in the method of the present invention;

[0036] Figure 2 A schematic diagram of the structural composition of a primary aggregation module combined with deformable convolution in a high-resolution remote sensing image semantic segmentation network constructed in the method of the present invention;

[0037] Figure 3A schematic diagram of the composition structure of a channel attention module in a high-resolution remote sensing image semantic segmentation network constructed in the method of the present invention;

[0038] Figure 4 The figure is a schematic diagram of the composition structure of the spatial context module in the high-resolution remote sensing image semantic segmentation network constructed in the method of the present invention. DETAILED DESCRIPTION

[0039] The specific implementation modes of the present invention are described below in conjunction with the accompanying drawings so that those skilled in the art can better understand the present invention.

[0040] The present invention relates to a remote sensing image road segmentation method based on a multi-scale topology perception neural network, comprising the following steps:

[0041] Step 1: Divide the high-resolution remote sensing dataset into a training set and a test set, and preprocess the images in the training set and the test set respectively.

[0042] The preprocessing includes manual image annotation, image cropping and data enhancement; the manual image annotation specifically includes: manually performing pixel-level semantic annotation on roads in high-resolution images in ArcGIS software to obtain labeled surface crack images.

[0043] The image cropping is specifically as follows: randomly cropping the labeled high-resolution remote sensing image into sub-images of 512 pixels×512 pixels.

[0044] The data enhancement includes: performing random scale transformation on sub-images, random angle image rotation, and vertical and horizontal flipping of images to obtain a high-resolution remote sensing image.

[0045] Step 2: Build a high-resolution remote sensing image semantic segmentation network, such as Figure 1 As shown, the high-resolution remote sensing image semantic segmentation network is composed of a downsampling path, a spatial context module and an upsampling path in sequence, and the downsampling path and the upsampling path are connected by a jump connection adjusted by channel attention.

[0046] Downsampling path: Starting from the input data, it first passes through a deformable convolution layer with a convolution kernel size of 3×3 and a stride of 1 to obtain a semantic feature map with a channel number m of 48, and then passes through 5 consecutive downsampling units for feature extraction and downsampling.

[0047] Deformable convolution adds a learnable offset variable and weight coefficient to the position of each sampling point in the conventional convolution kernel; through these variables, the convolution kernel can sample arbitrarily near the current position and eliminate the interference of useless context information, and is no longer limited to the regular grid of conventional convolution; deformable convolution is conducive to the accurate extraction of complex geometric features of the road.

[0048] The downsampling unit is composed of a single aggregation module, an aggregation operation and a down-conversion module in sequence. The single aggregation module is responsible for extracting features. The aggregation operation then aggregates the input and output of the single aggregation module, and then transmits the resulting feature map to the down-conversion module and the channel attention module respectively. The output of the channel attention module (channel attention) is multiplied by the input by corresponding elements to obtain the jump connection features adjusted by the channel attention and input them into the corresponding part of the upsampling path.

[0049] like Figure 2 As shown in the figure, the structure of the primary aggregation module is as follows: After the feature map I0 is input into the primary aggregation module, it first passes through n convolution modules to obtain n new feature maps I1, I2...I n (The first two convolution modules are deformable convolution modules); then the obtained n feature maps I1, I2…I n Perform channel stacking operation to obtain the feature map J0 after channel stacking; the one-time aggregation module converts the feature Figure 1 Secondary aggregation reuses the features extracted by the network and avoids redundant calculations, thus effectively reducing the difficulty of model training and the number of samples required.

[0050] The total number n of conventional convolution and deformable convolution modules contained in each one-time aggregation module in the downsampling path is 4, 5, 7, 10, and 12, respectively, and the corresponding number of output feature map channels m is 112, 192, 304, 464, and 656, respectively; the total number n of conventional convolution and deformable convolution modules contained in the one-time aggregation module in each upsampling unit is 12, 10, 7, 5, and 4, respectively, and the number of feature map channels m output to the next up-conversion module is 192, 160, 112, 80, and 64.

[0051] The convolution module or deformable convolution module in the primary aggregation module is composed of a batch normalization layer, a ReLU activation function layer, a 3×3 convolution layer or a deformable convolution layer (channel number m=16) and a random inactivation layer (inactivation probability p=0.2) in sequence.

[0052] like Figure 3 As shown in the figure, the structure of the channel attention module is as follows: the input features of size C′×H′×W′ are respectively subjected to global maximum pooling and global average pooling to obtain two C′×1×1 feature maps, and then they are respectively sent to a shared multi-layer perceptron: the number of neurons in the first layer is C′ / r (r is the reduction rate, r=16), the activation function is Relu, and the number of neurons in the second layer is C′; the two output features are added, and then subjected to Sigmoid activation operation to generate the final channel attention; the channel attention and the input feature map perform corresponding element multiplication operations.

[0053] The down-conversion module uses Daubechies wavelet as the wavelet basis. After a wavelet transform, each channel of the image is decomposed into four sub-images with half the original height and width: I GG ,I GD ,I DG and I DD , where the order of the letters in the subscripts represents the horizontal and vertical directions respectively, G represents high-frequency information, and D represents low-frequency information. For example, GG represents: both the horizontal and vertical directions are high-frequency information subscripts; I DD The sub-image is output, while the remaining sub-images are transmitted to the modules corresponding to the upsampling path.

[0054] like Figure 4 As shown, the spatial context module consists of two paths, namely, the “height direction” and the “width direction”. In the “height direction” path, the feature map from the downsampling path (assuming a size of H×W×C, where C represents the number of channels, H represents the feature map height, and W represents the feature map width) is divided into H slices in the vertical direction: P1=P 11 , P 12 …, P 1H , first for each slice P 1i (1≤i≤H) uses a 1×1 convolution operation to perform linear projection to obtain F i , then T i Input the first linear layer M k (S×C, where S=64, C=656), get attention A i , and then focus on A i Softmax and L1 regularization are used in the feature map size and channel dimension respectively, and then the regularized attention is input into the second linear layer M v (C×S, where C=656, S=64), and get the new slice P′ 1i, and finally all new slices are aggregated into a feature map of size C×H×W along the "height direction"; in the "width direction" path, the feature map from the downsampling path is divided into W slices along the "width" direction, and the slices are updated and aggregated in the same way as the first two stages; the feature maps of size H×W×C from the two paths of "height direction" and "width direction" are fused by addition operation; the spatial context module converts the traditional convolutional layer-by-layer connection form into the slice-by-slice linear complexity self-attention form in the feature map, so that information can be transmitted between pixel rows and columns in the map, and a linear layer with memory function is used to replace the Key vector and Value vector in the traditional self-attention. Therefore, even if the road in the remote sensing image is blocked or covered by shadows, the spatial context module is still suitable for detecting road targets with long-distance continuous shapes, and can extract global context information distributed in the entire training set while reducing computational complexity.

[0055] The upsampling path is composed of 5 consecutive upsampling units, a convolution layer with a convolution kernel size of 1×1 and a step size of 1, and a Softmax layer, and finally generates a segmentation prediction map.

[0056] The upsampling path is composed of 5 consecutive upsampling units, an aggregation operation (denoted as A aggregation operation), a convolution layer with a convolution kernel size of 1×1 and a step size of 1, and a Softmax layer, and finally generates a segmentation prediction map.

[0057] The upsampling unit consists of a group of up-conversion modules, aggregation operations (denoted as B aggregation operations) and a one-time aggregation module, wherein the up-conversion module is responsible for restoring the spatial resolution of the feature map through upsampling, the B aggregation operation is responsible for aggregating the skip connection feature map adjusted by the channel attention module and the feature map obtained by the up-conversion module, and the one-time aggregation module is responsible for extracting features from the results of the B aggregation operation.

[0058] The up-conversion module converts the high-frequency information feature map (I GG ,I GD and I DG ) After the number of channels is increased through 1×1 convolution, it is combined with the new features output by the spatial context module or the previous one-time aggregation module, and an inverse wavelet transform with Daubechies wavelet as the wavelet basis is performed to achieve upsampling.

[0059] The A aggregation operation in the upsampling path aggregates the result of the B aggregation operation in the last upsampling unit with the output of the primary aggregation module again.

[0060] The convolution layer with a convolution kernel size of 1×1 and a stride of 1 converts the number of channels to 2.

[0061] Step 3: Input the preprocessed training set images in step 1 into the high-resolution remote sensing image semantic segmentation network in step 2 for training. First, use the He Uniform method to initialize the high-resolution remote sensing image segmentation network, then update the parameters in the model, and optimize the loss function until convergence.

[0062] Where X represents the set of predicted values ​​corresponding to all pixels in the prediction image, Y represents the set of values ​​corresponding to all pixels in the label image, |XIY| represents the intersection of X and Y, and |X| and |Y| represent the number of elements in X and Y.

[0063] Step 4: Input the preprocessed test set remote sensing images into the segmentation network trained in step 3, and output the semantic segmentation results of the high-resolution remote sensing images.

[0064] What needs to be explained about the specific structure of the present invention is that the connection relationship between the various component modules adopted in the present invention is definite and feasible. Except for the special instructions in the embodiments, the specific connection relationship can bring about the corresponding technical effects, and solve the technical problems raised by the present invention without relying on the execution of the corresponding software program. Except for the specific instructions, the models and connection methods of the components, modules and specific components appearing in the present invention belong to the existing technologies such as published patents, published journal articles, or common knowledge that can be obtained by technical personnel in this field before the application date, and there is no need to elaborate, so that the technical solution provided in this case is clear, complete and feasible, and the corresponding physical products can be reproduced or obtained according to the technical means.

[0065] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A high-resolution remote sensing image semantic segmentation network system, characterized by: Including downsampling path, spatial context module and upsampling path, The downsampling path includes a deformable convolution layer and 5 consecutive downsampling units. The input data first passes through the deformable convolution layer to obtain a semantic feature map, and then passes through 5 consecutive downsampling units for feature extraction and downsampling, and finally outputs a feature map; The upsampling path is composed of 5 consecutive upsampling units, an aggregation operation A, a convolution layer and a Softmax layer, and finally generates a segmentation prediction map; The upsampling unit is composed of a group of up-conversion modules, an aggregation operation B and a primary aggregation module, wherein the up-conversion module is responsible for restoring the spatial resolution of the feature map by up-sampling, the aggregation operation B is responsible for aggregating the skip connection feature map adjusted by the channel attention module and the feature map obtained by the up-conversion module, and the primary aggregation module is responsible for extracting features from the result of the aggregation operation B; the 5 down-sampling units correspond to the 5 up-sampling units one by one, and are connected by skip connections adjusted by the attention of the channel attention module; The downsampling unit includes a primary aggregation module, an aggregation operation and a down-conversion module, wherein the primary aggregation module is responsible for extracting features, and then the aggregation operation aggregates the input and output of the primary aggregation module, and then transmits the resulting feature map to the down-conversion module and the channel attention module respectively, wherein the output of the channel attention module is multiplied by the corresponding element of the input to obtain the jump connection feature adjusted by the channel attention and input to the corresponding part of the upsampling path; The spatial context module segments and fuses the feature maps output by the downsampling path and outputs them to the upsampling path.

2. The high-resolution remote sensing image semantic segmentation network system according to claim 1, characterized in that: The deformable convolution layer is a deformable convolution with a convolution kernel size of 3×3 and a stride of 1. This operation adds a learnable offset variable and weight coefficient to the position of each sampling point in the 3×3 convolution kernel.

3. The high-resolution remote sensing image semantic segmentation network system according to claim 1, characterized in that: The structure of the primary aggregation module is as follows: After the feature map I0 is input into the primary aggregation module, it first passes through n convolution modules to obtain n new feature maps I1, I2...I n , where the first two convolution modules are deformable convolution modules; then the obtained n feature maps I1, I2…I n Perform channel stacking operation to obtain the feature map J0 after channel stacking; The convolution module is composed of a batch normalization layer, a ReLU activation function layer, a 3×3 convolution layer and a random dropout layer in sequence; the deformable convolution module is composed of a batch normalization layer, a ReLU activation function layer, a 3×3 deformable convolution layer and a random dropout layer in sequence.

4. The high-resolution remote sensing image semantic segmentation network system according to claim 3, characterized in that: The structure of the channel attention module is as follows: the input features of size C′×H′×W′ are subjected to global maximum pooling and global average pooling respectively to obtain two feature maps of size C′×1×1; Then they are sent to a shared multilayer perceptron: the number of neurons in the first layer is C′ / r, r is the reduction rate, r=16, the activation function is Relu, and the number of neurons in the second layer is C′; The two output features are added together, and then the Sigmoid activation operation is performed to generate the final channel attention; the channel attention and the input features are multiplied by the corresponding elements.

5. The high-resolution remote sensing image semantic segmentation network system according to claim 4, characterized in that: The spatial context module includes two paths: "height direction" and "width direction"; The "height direction" path comes from the feature map of the downsampling path, with a size of H×W×C, where C represents the number of channels, H represents the feature map height, and W represents the feature map width. It is divided into H slices in the vertical direction: P1=P 11 , P 12 …, P 1H , for each slice P 1i , 1≤i≤H, use a 1×1 convolution operation to perform linear projection to obtain F i , then T i Input the first linear layer M k , get attention A i , and then focus on A i Softmax and L1 regularization are used in the feature map size and channel dimension respectively, and then the regularized attention is input into the second linear layer M v , and get a new slice P′ 1i , and finally aggregate all new slices along the "height direction" into a feature map of size C×H×W; In the "width direction" path, the feature map from the downsampling path is divided into W slices along the "width" direction, and the slices are updated and aggregated into feature maps in the same way; The feature maps from both the "height direction" and "width direction" paths, both of size H×W×C, are fused through addition operations.

6. A remote sensing image road segmentation method, characterized in that: The following steps are included: S100: Divide the high-resolution remote sensing dataset into a training set and a test set, and preprocess the images in the training set and the test set respectively; S200: constructing the high-resolution remote sensing image semantic segmentation network system as claimed in claim 1; S300: inputting the preprocessed training set images in S100 into the high-resolution remote sensing image semantic segmentation network system in S200 for training, firstly initializing the high-resolution remote sensing image semantic segmentation network system using the He Uniform method, then updating the parameters in the model, and optimizing the loss function until convergence; S400: Input the preprocessed test set remote sensing images into the high-resolution remote sensing image semantic segmentation network system trained in S200, and output the high-resolution remote sensing image semantic segmentation results.

7. The method for remote sensing image road segmentation according to claim 6, characterized in that: In step S100, the preprocessing includes manual image annotation, image cropping and data enhancement; the manual image annotation specifically includes: manually annotating the roads in the high-resolution image at the pixel level in ArcGIS software to obtain a labeled road image; The image cropping is specifically as follows: randomly cropping the labeled high-resolution remote sensing image into sub-images of 512 pixels×512 pixels; The data enhancement includes: performing random scale transformation on sub-images, random angle image rotation, and vertical and horizontal flipping of images to obtain a high-resolution remote sensing image.

8. The remote sensing image road segmentation method according to claim 6, characterized in that: In step S300, the loss function adopts Dice Loss, which is specifically: Where X represents the set of predicted values ​​corresponding to all pixels in the prediction image, Y represents the set of values ​​corresponding to all pixels in the label image, |XIY| represents the intersection of X and Y, and |X| and |Y| represent the number of elements in X and Y.

Images