A hyperspectral image segmentation method based on combined spectral and spatial features

Abstract

This invention relates to the field of image target detection technology and discloses a hyperspectral image segmentation method based on a combination of spectral and spatial features. The method comprises the following steps: 1) acquiring a hyperspectral image dataset; 2) performing hyperspectral image segmentation using a deep learning-based multi-scale spatial and spectral feature combination; and 3) evaluating the effectiveness of the network segmentation. This invention utilizes deep learning methods to segment hyperspectral images. By proposing a segmentation method that combines spectral and spatial features, the method effectively extracts and fuses the spectral and spatial features of the hyperspectral image, while also improving the model's prediction accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of image target detection, and in particular to a hyperspectral image segmentation method based on combined spectral and spatial features. Background Art

[0002] As an important type of aircraft, drones offer advantages over satellites or larger aircraft for remote sensing, such as low cost, high flexibility, rapid image acquisition, and high spatial resolution. Drone platforms can carry a range of remote sensing instruments, including visible light, multispectral, and thermal infrared sensors. The rapid development of drone remote sensing has greatly simplified research and practical applications in various fields. Leveraging rapid advances in artificial intelligence (AI), drones can now achieve autonomous target recognition, enabling more intelligent flight and mission execution. Consequently, they are finding widespread application in agriculture, environmental monitoring, and security. The confrontation between reconnaissance and camouflage in modern warfare has become increasingly intense with the rapid development of aerospace, optoelectronic, and material technologies. Camouflage, as a crucial means of survival for military targets, has become a key concern for militaries worldwide. Currently, camouflage methods include camouflage uniforms, paint, and camouflage nets. Traditional reconnaissance methods, such as visible light and infrared, have various limitations in identifying camouflaged targets and are unable to effectively identify them. However, with the development of hyperspectral remote sensing technology, spectral resolution continues to increase, making it possible to reconnaissance targets across hundreds of wavelengths. Furthermore, the continuous spectral information of each pixel in the image is well preserved, and contains the target's spatial characteristics, embodying a wealth of deep-dimensional information. These characteristics significantly enhance the information richness of hyperspectral data compared to ordinary color images, giving it advantages over other detection methods in detecting, identifying, and classifying camouflaged targets.

[0003] Therefore, in order to meet the highly complex battlefield conditions in the future, cope with the needs of high real-time response and strong secrecy in wars, and meet the needs of agile detection of camouflaged targets, we should take advantage of the miniaturization and intelligence of drones in modern warfare, and combine spectral imaging with rich spectral information and high spatial resolution to capture more detailed surface information and improve the detection ability of camouflaged targets, which is of great significance to the current intelligent detection methods of camouflaged targets.

[0004] A large number of deep learning models, represented by convolutional neural networks, have been applied to intelligent information extraction tasks in remote sensing images. However, compared with visible light remote sensing images, hyperspectral remote sensing images contain richer spectral and spatial information. When these methods are directly applied to hyperspectral images, the spectral and spatial information in the hyperspectral images cannot be fully and effectively utilized, resulting in suboptimal segmentation accuracy. Secondly, the scales of different types of targets in hyperspectral images vary greatly. How to effectively extract target features at different scales is also one of the difficulties in improving segmentation accuracy. Finally, because camouflaged targets have similar characteristics to the background environment and their outline boundaries are more blurred, the recognition of camouflaged targets is also more difficult. Summary of the Invention

[0005] To solve these problems, this paper proposes a hyperspectral image segmentation method based on the combination of spectral and spatial features.

[0006] The technical solutions of the present invention are as follows:

[0007] A hyperspectral image segmentation method based on spectral-spatial feature combination includes the following steps:

[0008] 1) Obtain hyperspectral image dataset:

[0009] The dataset is collected by a hyperspectral video camera, and the experimental scene consists of green grass and green painted camouflaged vehicles.

[0010] 2) Hyperspectral image segmentation based on multi-scale spatial-spectral feature combination:

[0011] Construct a multi-scale spatial-spectral feature fusion network model:

[0012] Including the following aspects:

[0013] a) Propose a multi-scale convolution window and spectral attention module to fully integrate the spectral and spatial information contained in the hyperspectral image during the downsampling stage, thereby improving the semantic segmentation accuracy of the hyperspectral image;

[0014] b) UAV remote sensing images have different object scales. By adding a new multi-scale feature extraction module, the segmentation performance of ground objects at different scales is improved;

[0015] c) The semantic information between different levels is integrated, and the deep semantic information is used as a guide for the shallow semantic information to strengthen the connection between categories, realize the complementarity of spatial and geometric information, and refine the segmentation edges, thereby enhancing the spatial detail information and the recognition ability of camouflaged targets.

[0016] 3) Network segmentation effect evaluation

[0017] The dataset was divided into training and test sets. Comparative experiments were conducted using SegNet, PspNet, DeepLabV3+, and a multi-scale spatial-spectral feature fusion network model. Aside from the network model, all other experimental conditions, including the dataset and data augmentation methods, remained the same. The effectiveness of this method was verified by comparing it with other segmentation methods using pixel accuracy (PA), mean intersection-over-union (MIOU), and recognition results.

[0018] Further, the process of step 2) is:

[0019] Step 2.1: Multi-scale convolution window:

[0020] In order to effectively extract spectral features from hyperspectral images in the downsampling stage of the backbone network and combine them with spatial features, a multi-scale convolutional window is proposed as the first convolutional layer of the hyperspectral image input.

[0021] The convolution kernel sizes are 1×1, 3×3, and 5×5. 1×1 convolution is used to extract spectral features from hyperspectral images, while 3×3 and 5×5 convolutions are used to extract local spatial features. Feature maps extracted by convolution kernels of different sizes are combined to form a joint spatial-spectral feature map, which is then subjected to 3×3 max pooling as input to subsequent convolutional layers.

[0022] Step 2.2: Spectral Attention Module:

[0023] The channel attention mechanism can generate additional weight information on each channel of the feature. Therefore, the channel attention is introduced after the residual structure of Res2Net to extract key band features and suppress non-key band data.

[0024] Design spectral channel attention module: where X∈R C×H×W is the input feature map of the channel attention module, R is a set of real numbers; H, W, and C represent the height, width, and number of channels of the feature map respectively. sq Represents a dimension compression operation, which uses global average pooling to compress the spatial dimension of the hyperspectral image to obtain a feature map of size 1×1×C. This method can better capture the correlation between different bands. ex For the excitation operation, it consists of two 1×1 convolutional layers, a ReLU activation function, and a Sigmoid activation function. First, the dimension is reduced to the original 1 / r, and then the dimension is restored to C channels through the next fully connected layer, where r represents the compression factor. Through these two convolutional layers, the network can learn the band characteristics, so that the output weights can reflect the importance differences of different bands. Finally, F scale operate, The symbol represents channel-by-channel multiplication. This step is similar to the gating mechanism, which adaptively assigns corresponding weights to each channel of the input feature map.

[0025] Step 2.3: Multi-scale dense feature fusion module:

[0026] Multi-scale dense feature fusion module: Dilated convolution can obtain a larger receptive field than ordinary convolution kernels. For a dilated convolution with a dilation rate of d and a convolution kernel size of K, the equivalent receptive field size can be expressed as:

[0027] R=(d-1)×(K-1)+K (1)

[0028] We use dilated convolutions with different dilation rates to capture the different scale features of the input features. We combine dilated convolution layers with different dilation rates in a cascaded manner. The input and output features of each layer are concatenated with all outputs from the lower layers. Each dilation layer can be represented as follows:

[0029]

[0030] where d l represents the expansion rate of a layer; [...] represents the splicing operation between feature layers; [X l-1 ,X l-2 ,...,X0] represents the output of all layers before concatenation. This dense connection can form a denser feature pyramid and a larger receptive field. At the same time, adaptive average pooling can integrate a wide range of spatial information and prevent overfitting. Therefore, global average pooling is added to this module, and these six branches are densely connected backward.

[0031] Step 2.4: Feature fusion module:

[0032] Feature fusion module, which realizes the effective fusion of low-level semantic features and deep-level semantic features;

[0033] First, in order to accurately utilize spatial detail information and semantic information, the two features are weighted and fused as the input of the feature fusion module; the weighted fusion operation adaptively adjusts the size of the weight according to the contribution of the two features, which can be specifically expressed as:

[0034] F=α·F Low +(1-α)·F High (3)

[0035] Among them, F Low Represents the shallow features of the output of ResBlock1, F High represents the deep features after the multi-scale fusion module, and F represents the fused features.

[0036] In order to further enhance the representation of spectral, multi-scale and spatial features and improve the accuracy of edge detail segmentation of camouflaged targets, three branches are used to process the input features X∈R C×H×W be applied. Specifically, the spectral channel attention is the same as that in the encoder, and global average pooling is first used to generate a 1×1×C channel attention map. The channel size is reduced by 1×1 convolution and restored to the original channel size after the ReLu activation function and 1×1 convolution. Then, the Sigmoid activation function is used to obtain the channel weight distribution, which is multiplied with the original input element to obtain a feature map containing channel attention. The multi-scale branch first uses 1×1 convolution to reduce the channel dimension, and then extracts multi-scale information through 3 parallel separable dilated convolutions, where the convolution kernel size is 3×3 and the dilation rate is R. i (i=1,2,3) are 7, 9, and 11, respectively. After being restored to the original channel dimension through 1×1 convolution, the three parallel results are subsequently fused through a summation operation. The output of the multi-scale branch is summed through 1×1 convolution and residual connection. The spatial branch first generates a feature map using maximum pooling and average pooling, which is then concatenated along the channel dimension to generate a spatial attention map of size 2×H×W. The spatial weight distribution is obtained through 7×7 convolution and a sigmoid activation function, which is multiplied with the original input element to obtain a feature map containing spatial attention. Finally, the results of the three branches are fused through a summation operation to produce the final feature map.

[0037] Further, the process of step 3) is:

[0038] Step 3.1: During training, all images are fixed to a pixel size of 360×480, the training batch size is set to 8, and the training is repeated for 150 cycles. The training set of 2310 images is trained in batches of 200. The network is also verified on the validation set. After training, the network weights are saved. The Adam optimizer is used as the optimizer, the initial learning rate is 0.001, and the learning rate is decayed by 10% every ten batches. The cross entropy function is used as the loss function.

[0039] To evaluate the accuracy of the model segmentation, we use mean intersection over union (Miou), pixel accuracy (Pixelaccuracy), and F1 score (F1 Score) to measure segmentation performance, as shown in formula (4-6). TP stands for true positive, FP stands for false positive, TN stands for true negative, and FN stands for false negative.

[0040]

[0041] In the formula, Precision represents the precision rate, and Recall represents the recall rate. The specific calculation is shown in formula (7-8).

[0042]

[0043] In order to illustrate the effectiveness of the proposed algorithm, the test images were input into SegNet, PspNet, DeepLabV3+ and the multi-scale spatial-spectral feature fusion network model for comparative experiments, and several predicted images were randomly selected for visualization.

[0044] The beneficial effects of the present invention are mainly manifested in: the present invention uses deep learning methods to segment hyperspectral images, and by proposing a segmentation method that combines spectral and spatial features, it effectively extracts and fuses the spectral features and spatial features of hyperspectral images at the encoder stage, extracts multi-scale spatial features through a multi-scale dense feature fusion module, and at the decoder stage, uses deep semantic features to guide shallow spatial details to improve detail segmentation capabilities and the prediction accuracy of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is the overall network structure diagram proposed by the present invention;

[0046] Figure 2 This is a multi-scale convolution window structure diagram proposed by the present invention;

[0047] Figure 3 This is the spectral channel attention structure diagram proposed by the present invention;

[0048] Figure 4 This is the structural diagram of the multi-scale dense feature fusion module proposed in the present invention;

[0049] Figure 5 This is a structural diagram of the feature fusion module proposed in the present invention;

[0050] Figure 6 This is a comparison chart of the recognition results proposed by the present invention. DETAILED DESCRIPTION

[0051] The present invention will be further described below with reference to the accompanying drawings.

[0052] Reference Figures 1 to 6 , a hyperspectral image segmentation method based on spectral-spatial feature combination, comprising the following steps:

[0053] Step 1) Obtain a hyperspectral image dataset of camouflaged vehicles:

[0054] The dataset was acquired using a Lingyunguang HV-125 series hyperspectral video camera. The experimental scene consists of green grass and a green-painted camouflaged vehicle. The image size is 1387 pixels × 1829 pixels, with 128 spectral response bands in the wavelength range of 450-950 nanometers and a spectral resolution of 4 nanometers.

[0055] Step 2) Hyperspectral image segmentation using multi-scale spatial-spectral feature combination

[0056] The spatial-spectral feature fusion network of the proposed scale includes the following three aspects:

[0057] 1) A multi-scale convolution window and spectral attention module are proposed to fully fuse the spectral and spatial information contained in hyperspectral images in the downsampling stage, thereby improving the semantic segmentation accuracy of hyperspectral images; 2) UAV remote sensing images have different target scales. By adding a new multi-scale feature extraction module, the segmentation performance of ground objects at different scales is improved; 3) Semantic information between different levels is fused, and deep semantic information is used as a guide for shallow semantic information to strengthen the connection between categories, realize the complementarity of spatial and geometric information, and refine the segmentation edges, thereby enhancing spatial detail information and the ability to recognize camouflaged targets.

[0058] Step 2.1: In order to effectively extract the spectral features in the hyperspectral image and combine them with the spatial features in the downsampling stage of the backbone network, a multi-scale convolution window is proposed as the first convolution layer of the hyperspectral image input;

[0059] The convolution kernel sizes are 1×1, 3×3, and 5×5. 1×1 convolution is used to extract spectral features from hyperspectral images, while 3×3 and 5×5 convolutions are used to extract local spatial features. Feature maps extracted by convolution kernels of different sizes are combined to form a joint spatial-spectral feature map, which is then subjected to 3×3 max pooling as input to subsequent convolutional layers.

[0060] Step 2.2: The channel attention mechanism can generate additional weight information on each channel of the feature. Therefore, the channel attention is introduced after the residual structure of Res2Net to extract key band features and suppress non-key band data.

[0061] Inspired by SENet, this paper designs a spectral channel attention. C×H×W is the input feature map of the channel attention module, R is a set of real numbers; H, W, and C represent the height, width, and number of channels of the feature map respectively. sq Represents a dimension compression operation, which uses global average pooling to compress the spatial dimension of the hyperspectral image to obtain a feature map of size 1×1×C. This method can better capture the correlation between different bands. exFor the excitation operation, it consists of two 1×1 convolutional layers, a ReLU activation function, and a Sigmoid activation function. First, the dimension is reduced to the original 1 / r, and then the dimension is restored to C channels through the next fully connected layer, where r represents the compression factor. Through these two convolutional layers, the network can learn the band characteristics, so that the output weights can reflect the importance differences of different bands. Finally, F scale operate, The symbol represents channel-by-channel multiplication. This step is similar to the gating mechanism, which adaptively assigns corresponding weights to each channel of the input feature map.

[0062] Step 2.3: Multi-scale dense feature fusion module

[0063] Multi-scale dense feature fusion module, the internal structure of this module is as follows Figure 4 As shown, dilated convolution can achieve a larger receptive field than ordinary convolution kernels. Dilated convolutions with different dilation rates are used to capture the different scale features of input features. Dilated convolution layers with dilation rates of (3, 5, 7, and 9) are combined in a cascade manner. The input and output features of each layer are concatenated with all outputs from lower layers. This dense connection can form a denser feature pyramid and a larger receptive field, and can capture more sampling points for fewer categories and smaller target objects. At the same time, adaptive average pooling is added to this module to integrate a wide range of spatial information and prevent overfitting. These six branches are tightly connected backward.

[0064] Step 2.4: Feature Fusion Module

[0065] Feature fusion module, see the attached for its specific structure Figure 5 , to achieve the effective fusion of low-level semantic features and deep semantic features.

[0066] First, in order to accurately utilize spatial detail information and semantic information, the two features are weighted and fused as the input of the feature fusion module; the weighted fusion operation adaptively adjusts the size of the weight according to the contribution of the two features, which can be specifically expressed as:

[0067] F=α·F Low +(1-α)·F High (3)

[0068] Among them, F Low Represents the shallow features of the output of ResBlock1, F High represents the deep features after the multi-scale fusion module, and F represents the fused features.

[0069] In order to further enhance the representation of spectral, multi-scale and spatial features and improve the accuracy of edge detail segmentation of camouflaged targets, three branches are used to process the input features X∈R C×H×W be applied. Specifically, the spectral channel attention is the same as that in the encoder, and global average pooling is first used to generate a 1×1×C channel attention map. The channel size is reduced by 1×1 convolution and restored to the original channel size after the ReLu activation function and 1×1 convolution. Then, the Sigmoid activation function is used to obtain the channel weight distribution, which is multiplied with the original input element to obtain a feature map containing channel attention. The multi-scale branch first uses 1×1 convolution to reduce the channel dimension, and then extracts multi-scale information through 3 parallel separable dilated convolutions, where the convolution kernel size is 3×3 and the dilation rate is R. i (i=1,2,3) are 7, 9, and 11, respectively. After being restored to the original channel dimension through 1×1 convolution, the three parallel results are subsequently fused through a summation operation. The output of the multi-scale branch is summed through 1×1 convolution and residual connection. The spatial branch first generates a feature map using maximum pooling and average pooling, which is then concatenated along the channel dimension to generate a spatial attention map of size 2×H×W. The spatial weight distribution is obtained through 7×7 convolution and a sigmoid activation function, which is multiplied with the original input element to obtain a feature map containing spatial attention. Finally, the results of the three branches are fused through a summation operation to produce the final feature map.

[0070] Step 3) Network segmentation effect evaluation

[0071] The dataset was divided into training and test sets. Comparative experiments were conducted using SegNet, PspNet, DeepLabV3+, and the proposed model. Aside from the network model, all other experimental conditions, including the dataset and data augmentation methods, remained the same. The effectiveness of our proposed method was verified by comparing it with other segmentation methods using pixel accuracy (PA), mean intersection-over-union (MIOU), and recognition results.

[0072] Step 3.1: During training, all images are fixed to a pixel size of 360×480, the training batch size is set to 8, and the training set of 2310 images is trained in batches of 200. The network is also validated on the validation set, and the network weights are saved after training. The Adam optimizer is used as the optimizer, with an initial learning rate of 5e-6 and a learning rate decay of 10% every ten batches. The cross entropy function is used as the loss function.

[0073] To evaluate the accuracy of the model segmentation, we use mean intersection over union (Miou), pixel accuracy (Pixelaccuracy), and F1 score (F1 Score) to measure segmentation performance, as shown in formula (4-6). TP stands for true positive, FP stands for false positive, TN stands for true negative, and FN stands for false negative.

[0074]

[0075] In the formula, Precision represents the precision rate, and Recall represents the recall rate. The specific calculation is shown in formula (7-8).

[0076]

[0077] In order to illustrate the effectiveness of the proposed algorithm, this paper inputs the test images into SegNet, PspNet, DeepLabV3+ and the model proposed in this paper for comparative experiments, and then selects several predicted images for visualization. Figure 6 Shown is the comparison between the current network and the network prediction results; Figure 6 Figure (a) is the original image, (b) is the true value image of the original image, (c) is the recognition result image of DeepLabV3+, (d) is the recognition result image of PspNet, (e) is the recognition result image of SegNet, and (f) is the recognition result image of the present invention.

[0078] In the model comparison experiment, except for changing the network model, the rest of the experimental conditions are the same, including the dataset, data enhancement method, etc. The specific experimental results are shown in Table 1

[0079] Table 1 Comparative experiment

[0080]

[0081] The contents described in the embodiments of this specification are merely an enumeration of the implementation forms of the inventive concept. The scope of protection of the present invention should not be regarded as limited to the specific forms described in the embodiments. The scope of protection of the present invention also extends to equivalent technical means that can be conceived by those skilled in the art based on the inventive concept.

Claims

1. A hyperspectral image segmentation method based on spectral-spatial feature combination, characterized in that: The following steps are involved: 1) Obtain hyperspectral image dataset: Set up the experimental scene and use a hyperspectral video camera to acquire a hyperspectral image dataset; 2) Hyperspectral image segmentation based on multi-scale spatial-spectral feature combination: Construct a multi-scale spatial-spectral feature fusion network, including: Multi-scale convolution window: It can simultaneously extract channel and spatial features in the image, and fully integrate the spectral and spatial information contained in the hyperspectral image during the downsampling stage; Spectral Attention Module: The spectral attention module adaptively recalibrates the weight of each channel by generating attention masks on all feature channels, thereby improving the semantic segmentation accuracy of hyperspectral images. The multi-scale dense feature fusion module improves the segmentation performance of ground objects at different scales by fusing and extracting multi-scale context features at the encoder stage; The feature fusion module fuses semantic features from different layers and uses deep semantic features as a guide for shallow semantic features, strengthening the connection between categories, achieving the complementarity of spatial and geometric information, and refining segmentation edges, thereby enhancing spatial detail information and the ability to recognize camouflaged targets. 3) Network segmentation effect evaluation: The dataset is divided into training and test sets, and the test set images are input into SegNet, PspNet, DeepLabV3+ and multi-scale spatial-spectral feature fusion network models for comparative experiments.

2. The hyperspectral image segmentation method based on spectral-spatial feature combination according to claim 1, characterized in that: The process of step 2) is: Step 2.1: A multi-scale convolution window is used as the first convolution layer for the hyperspectral image input, with kernel sizes of 1×1, 3×3, and 5×5. The 1×1 convolution is used to extract the spectral features of the hyperspectral image, while the 3×3 and 5×5 convolutions are used to extract the local spatial features of the image. The feature maps extracted by convolution kernels of different scales are combined to form a joint spatial-spectral feature map, which is then subjected to 3×3 maximum pooling as the input of the subsequent convolutional layer; Step 2.2: The backbone network uses ResNet50. After the residual structure of the backbone network Res2Net, a channel attention mechanism is introduced to extract feature bands and suppress non-feature band data. Spectral channel attention module: X∈R C×H×W is the input feature map of the channel attention mechanism, R is a set of real numbers; H, W, and C represent the height, width, and number of channels of the feature map respectively; Dimensionality compression operation F sq , this operation uses global average pooling to compress the spatial dimension of the hyperspectral image and obtain a channel attention map, which is a feature map of size 1×1×C; Stimulus operation F ex , which consists of two 1×1 convolutional layers, a ReLU activation function, and a Sigmoid activation function; first, the channel dimension is reduced to the original 1 / r through the 1×1 convolutional layer, where r represents the compression factor; then, a ReLU activation function and a 1×1 convolutional layer are used to restore the dimension to C channels; and then the Sigmoid activation function is used to obtain the weight distribution of different channels; Multiplication operation F scale Operation, multiply the weight distribution by the original input element by element, adaptively assign corresponding weights to each channel of the input feature map, and obtain features containing channel attention; Step 2.3: Multi-scale dense feature fusion module: Capture input features X∈R using dilated convolutions with different dilation rates C×H×W For different scale features of , for a dilation rate d and a convolution kernel size K of the void convolution, the equivalent receptive field size is expressed as: R=(d-1)×(K-1)+K (1) The dilated convolutional layers with different dilation rates are combined in a cascade manner, and the input and output features of each layer are concatenated with all outputs from the previous layer. Each dilation layer is represented as follows: where d l represents the expansion rate of a layer; [...] represents the splicing operation between feature layers; [X l-1 ,X l-2 ,...,X0] represents the output of all layers before splicing; Step 2.4: Feature fusion module: The shallow semantic features and deep semantic features are weighted and fused as the input of the feature fusion module. The weighted fusion operation adaptively adjusts the weights according to the contributions of the two features, which can be expressed as: F=α·F Low +(1-α)·F High (3) Among them, F Low Represents the shallow features of the output of ResBlock1, F High represents the deep features after the multi-scale fusion module, and F represents the fused features; Global average pooling is used to generate a 1×1×C channel attention map; the channel size is reduced by 1×1 convolution and restored to the original channel size after the ReLu activation function and 1×1 convolution; then, the Sigmoid activation function is used to obtain the channel weight distribution, which is multiplied with the original input element to obtain a feature map containing channel attention; The multi-scale branch first uses 1×1 convolution to reduce the channel dimension, and then extracts multi-scale information through three parallel separable dilated convolutions, where the convolution kernel size is 3×3 and the dilation rate is R. i (i=1,2,3) are 7, 9, and 11 respectively; after being restored to the original channel dimension through 1×1 convolution, the three parallel results are then fused through summation; the output of the multi-scale branch is obtained by summing through 1×1 convolution and residual connection; The spatial branch first generates a feature map using maximum pooling and average pooling, and then concatenates it in the channel dimension to generate a spatial attention map of size 2×H×W; the spatial weight distribution is obtained through a 7×7 convolution and a Sigmoid activation function, which is multiplied with the original input element to obtain a feature map containing spatial attention; finally, the results of the three branches are fused through a sum operation to produce the final feature map.

3. The hyperspectral image segmentation method based on spectral-spatial feature combination according to claim 1, characterized in that: The specific process of step 3) is as follows: Step 3.1: During training, the training set is augmented using random rotations, random flips, and random spatial affine transformations. The pixel size of all images is fixed, the batch size and period of training are set, and the training set is trained in batches. The network is validated on the validation set, and the network weights are saved after training. The Adam optimizer is used, with an initial learning rate of 0.001 and a cosine decay strategy for adjusting the learning rate. The cross-entropy function is used as the loss function. The segmentation performance is measured using Miou, pixel accuracy, and F1 score, as shown in formulas (3)-(5). Where TP stands for true positive, FP stands for false positive, TN stands for true negative, and FN stands for false negative; In the formula, Precision represents the precision rate, Recall represents the recall rate, and the specific calculation is shown in formulas (6)-(7); The test set images are input into SegNet, PspNet, DeepLabV3+ and multi-scale spatial-spectral feature fusion network models for comparative experiments, and several predicted images are randomly selected for visualization.

Images