SAR sea ice segmentation method, device and equipment based on multi-scale feature extraction fusion

Abstract

The application provides a SAR sea ice segmentation method, device and equipment based on multi-scale feature extraction fusion. The SAR sea ice segmentation method comprises: in the imaging process of SAR sea ice image data, performing incident angle correction on the SAR sea ice image data to obtain a SAR image product of sea ice; performing multi-scale feature decomposition and reconstruction on the SAR image product by using a deep learning network embedded with a wavelet transform module, and performing channel enhancement by using a residual channel attention mechanism in the deep learning network to obtain a SAR image product after spot noise suppression; and performing segmentation on the SAR image product after spot noise suppression by using a sea ice segmentation model integrated with a lightweight adaptive extraction module and a multi-scale feature extraction module to obtain a segmentation result of sea ice.

Description

Technical Field

[0001] This invention relates to the fields of artificial intelligence, SAR image processing, and SAR image segmentation, specifically to a SAR sea ice segmentation method, apparatus, and electronic device based on multi-scale feature extraction and fusion. Background Technology

[0002] Synthetic Aperture Radar (SAR) is a key remote sensing tool for sea ice monitoring, capable of penetrating cloud layers and acquiring high-resolution images under various meteorological conditions. Changes in sea ice distribution and types have a profound impact on Arctic shipping safety, global climate change, and the ecological environment. Accurate identification and classification of sea ice types (such as new ice, old ice, and open water) are crucial not only for marine resource development, shipping safety, and disaster prevention, but also for providing important data support for climate change research. Sea ice classification requires sea ice segmentation, and the classification result can be obtained from the segmentation results. Traditional sea ice segmentation methods mainly rely on the echo intensity of SAR images or machine learning algorithms based on manual feature extraction, but these methods have limited performance in complex environments. In recent years, the introduction of deep learning technology has significantly improved the accuracy of sea ice segmentation, especially under complex meteorological and oceanic conditions, demonstrating strong adaptability and reliability. Summary of the Invention

[0003] In view of this, the present invention provides a SAR sea ice segmentation method, apparatus and electronic device based on multi-scale feature extraction and fusion.

[0004] According to a first aspect of the present invention, a SAR sea ice segmentation method based on multi-scale feature extraction and fusion is provided, comprising: during the imaging process of SAR sea ice image data, performing incident angle correction on the SAR sea ice image data to obtain a SAR image product of sea ice; using a deep learning network embedded with a wavelet transform module to perform multi-scale feature decomposition and reconstruction on the SAR image product, and using a residual channel attention mechanism in the deep learning network to perform channel enhancement, thereby obtaining a SAR image product with speckle noise suppression; and segmenting the SAR image product with speckle noise suppression based on a sea ice segmentation model integrating a lightweight adaptive extraction module and a multi-scale feature extraction module to obtain a segmentation result of sea ice.

[0005] According to an embodiment of the present invention, the step of correcting the incident angle of SAR sea ice image data includes: parsing radar parameters from the scene element file, determining correction coefficients based on the radar parameters and the geometric relationship between slant range and incident angle; and correcting the backscattering coefficients of SAR sea ice image data at different incident angles to the theoretical value at a reference incident angle based on the correction coefficients.

[0006] According to an embodiment of the present invention, the step of determining the correction coefficient based on radar parameters and the geometric relationship between slant range and incident angle includes: calculating the center slant range and range resolution based on radar parameters; calculating the incident angle corresponding to each column based on the center slant range and range resolution; using the beam center incident angle as a reference incident angle, determining the correction coefficient based on the beam center incident angle and the incident angle corresponding to each column.

[0007] According to an embodiment of the present invention, the steps of performing multi-scale feature decomposition and reconstruction of SAR image products using a deep learning network embedded with a wavelet transform module include: inputting the SAR image product into the deep learning network, extracting deep features of the SAR image product by alternately using a first convolutional layer and a discrete wavelet transform module; and alternately using an inverse wavelet transform module and a second convolutional layer to fuse and reconstruct the decomposed deep features to obtain the original structural information that retains the deep features.

[0008] According to an embodiment of the present invention, the step of using a residual channel attention mechanism for channel enhancement in a deep learning network includes: using the residual channel attention mechanism to perform channel enhancement on the feature map extracted by each first convolutional layer, and transmitting the channel-enhanced feature map to the corresponding second convolutional layer.

[0009] According to an embodiment of the present invention, the residual channel attention mechanism includes a 1×1 convolutional layer and a deep convolutional layer, the 1×1 convolutional layer and the deep convolutional layer are connected, and the deep convolutional layer is used to perform cross-channel local interaction on the features after dimensionality reduction of the 1×1 convolutional layer.

[0010] According to an embodiment of the present invention, the step of segmenting a SAR image product after speckle noise suppression to obtain a sea ice segmentation result includes: sequentially performing feature extraction, feature fusion, multi-scale information processing, and high-resolution reconstruction on the SAR image product after speckle noise suppression to determine the sea ice segmentation result.

[0011] According to an embodiment of the present invention, the steps of sequentially performing feature extraction, feature fusion, multi-scale information processing, and high-resolution reconstruction on the SAR image product after speckle noise suppression include: extracting initial multi-scale feature maps through the convolutional layer of the sea ice segmentation model; downsampling and feature enhancement of each initial multi-scale feature map using a lightweight adaptive extraction module, and fusing feature maps of different levels through skip connections; capturing detailed information from the fused feature maps using a multi-scale feature extraction module, and fusing it with shallow features to complete the multi-scale information processing of the features; upsampling the multi-scale information processed features to the original resolution to obtain the segmentation result.

[0012] A second aspect of the present invention provides a SAR sea ice segmentation device based on multi-scale feature extraction and fusion, comprising: a correction module for correcting the incident angle of SAR sea ice image data during the imaging process to obtain a SAR image product of sea ice; a denoising module for performing multi-scale feature decomposition and reconstruction of the SAR image product using a deep learning network embedded with a wavelet transform module, and performing channel enhancement using a residual channel attention mechanism in the deep learning network to obtain a SAR image product with speckle noise suppression; and a segmentation module for segmenting the SAR image product with speckle noise suppression based on a sea ice segmentation model integrating a lightweight adaptive extraction module and a multi-scale feature extraction module to obtain a segmentation result of sea ice.

[0013] A third aspect of the present invention provides an electronic device comprising: one or more processors; and a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors perform the method in any of the above embodiments. Attached Figure Description

[0014] The above-described features, other objects, and advantages of the present invention will become clearer from the following description of embodiments of the invention with reference to the accompanying drawings, in which:

[0015] Figure 1 A flowchart illustrating a SAR sea ice segmentation method based on multi-scale feature extraction and fusion according to an embodiment of the present invention is shown.

[0016] Figure 2 A flowchart illustrating a SAR sea ice segmentation method based on multi-scale feature extraction and fusion according to another embodiment of the present invention is shown.

[0017] Figure 3 A flowchart illustrating incident angle correction according to an embodiment of the present invention is shown schematically;

[0018] Figure 4 This diagram schematically illustrates the overall structure of a deep learning network according to an embodiment of the present invention.

[0019] Figure 5 This schematically illustrates the structure of a residual channel attention mechanism module in a deep learning network according to an embodiment of the present invention.

[0020] Figure 6 This schematic diagram illustrates the overall structure of a sea ice segmentation model according to an embodiment of the present invention;

[0021] Figure 7 The schematic diagram illustrates the structure of a lightweight adaptive extraction module for a sea ice segmentation model according to an embodiment of the present invention.

[0022] Figure 8 The diagram schematically illustrates the structure of a multi-scale feature extraction module in a sea ice segmentation model according to an embodiment of the present invention.

[0023] Figure 9 The schematic diagram illustrates the structural block diagram of a SAR sea ice segmentation device based on multi-scale feature extraction and fusion according to an embodiment of the present invention;

[0024] Figure 10 A block diagram of an electronic device suitable for implementing a SAR sea ice segmentation method based on multi-scale feature extraction fusion according to an embodiment of the present invention is shown schematically. Detailed Implementation

[0025] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the invention. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the invention for ease of explanation. However, it will be apparent that one or more embodiments may be practiced without these specific details. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0026] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention. The terms “comprising,” “including,” etc., as used herein indicate the presence of features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0027] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0028] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0029] The applicant found in his research that existing SAR sea ice segmentation methods still face some challenges.

[0030] (1) The change in incident angle leads to different scattering intensities in different regions of the image, which in turn affects the distinguishability of different sea ice types. For example, multi-year ice exhibits higher scattering due to its rough surface at low incident angles, while volume scattering caused by its internal bubble layer is more significant at high incident angles. The incident angle correction method based on visual interpretation and refitting a linear model is difficult to apply directly to an unlabeled SAR image.

[0031] (2) Speckle noise in SAR images can significantly affect the accuracy of sea ice segmentation. This granular noise caused by radar signal interference can lead to blurred sea ice edges and distorted texture features, resulting in abnormal fluctuations in radar backscattering values ​​for the same type of sea ice. Traditional denoising methods are prone to losing edge and texture details; adding a multi-view step during product processing will reduce resolution while suppressing speckle noise.

[0032] (3) The segmentation accuracy remains limited in complex scenarios, especially in distinguishing sea ice types with similar echo features. Due to the influence of noise, imaging conditions, and the physical properties of sea ice on SAR images, some sea ice types are very similar in echo features, making it difficult for the model to distinguish them accurately. Multi-scale targets in complex environments and the boundary regions between sea ice and open water further increase the segmentation difficulty.

[0033] (4) The generalization ability of the model still needs to be improved, and its performance varies significantly in SAR sea ice images of different regions or different categories. Due to differences in geographical environment, climate conditions and imaging parameters, the performance of the model on cross-regional data is unstable. In addition, the feature distribution of different categories of sea ice in SAR images also varies greatly, resulting in poor segmentation effect of the model for some categories.

[0034] In view of this, the present invention proposes a SAR sea ice segmentation method based on multi-scale feature extraction and fusion. The following description, in conjunction with the accompanying drawings, further illustrates the SAR sea ice segmentation method based on multi-scale feature extraction and fusion.

[0035] Figure 1 The flowchart illustrates a SAR sea ice segmentation method based on multi-scale feature extraction and fusion according to an embodiment of the present invention.

[0036] like Figure 1 As shown, the SAR sea ice segmentation method based on multi-scale feature extraction and fusion may include steps S110 to S130.

[0037] In step S110, during the imaging process of SAR sea ice image data, the incident angle of the SAR sea ice image data is corrected to obtain the SAR image product of sea ice.

[0038] In step S120, a deep learning network with an embedded wavelet transform module is used to perform multi-scale feature decomposition and reconstruction on the SAR image product, and a residual channel attention mechanism is used in the deep learning network to enhance the channel, thereby obtaining a SAR image product with speckle noise suppression.

[0039] In step S130, based on the sea ice segmentation model that integrates a lightweight adaptive extraction module and a multi-scale feature extraction module, the SAR image product after speckle noise suppression is segmented to obtain the sea ice segmentation result.

[0040] The SAR sea ice segmentation method based on multi-scale feature extraction and fusion provided in this embodiment can be divided into an incident angle correction stage (i.e., step S110), a noise suppression stage (i.e., step S120), and a sea ice segmentation stage (i.e., step S130). In the incident angle correction stage, the incident angle is corrected during the SAR product imaging process, meaning imaging and correction are completed simultaneously. This reduces error propagation and meets the requirements of automated processing, eliminating the need for manual intervention and improving processing efficiency and adaptability. In the noise suppression stage, wavelet transform, through multi-scale decomposition and adaptive thresholding, effectively suppresses SAR speckle noise while preserving details. The introduction of Residual Channel Attention Aggregation (RCAA) enables cross-scale information aggregation at multiple levels, making features at different depths more discriminative during fusion. The deep learning network combined with wavelet transform can automatically learn the optimal wavelet coefficient processing method, avoiding the limitations of traditional methods that rely on manually setting thresholds or basis functions, making the network more accurate in extracting and reconstructing features. In the sea ice segmentation stage, to address the issue of insufficient segmentation accuracy in complex scenarios, this embodiment introduces a Lightweight Adaptive Extraction (LAE) module, which is integrated into the feature extraction network in the form of a residual network. By preserving more contextual information and high-resolution details on multi-scale feature maps, it reduces the impact of noise while retaining rich semantic information, thereby improving the ability to distinguish sea ice categories in complex scenarios. To address the issue of insufficient model generalization ability, a Multi-Scale Feature Extraction (MSFE) module is introduced. Through a dual aggregation mechanism of spatial and channel features, it enhances feature representation capabilities, replaces traditional skip connections, and better supports multi-level information fusion in the decoder, thereby enhancing the model's generalization ability in SAR sea ice images of different regions and categories.

[0041] In some embodiments, the incident angle correction stage (i.e., step S110) may include: parsing radar parameters from the scene file, determining correction coefficients based on the radar parameters and the geometric relationship between slant range and incident angle; and, based on the correction coefficients, uniformly correcting the backscattering coefficients of SAR sea ice image data at different incident angles to the theoretical values ​​at the reference incident angle. Figure 2 A flowchart illustrating a SAR sea ice segmentation method based on multi-scale feature extraction and fusion according to another embodiment of the present invention is shown. Figure 3 A flowchart illustrating incident angle correction according to an embodiment of the present invention is shown schematically.

[0042] like Figure 2 As shown in this embodiment, the SAR sea ice segmentation method based on multi-scale feature extraction and fusion can start from level zero data (or L0 raw data) to perform product processing steps. Before generating a standard product, imaging algorithm processing is performed, and the incident angle of the SAR sea ice image data is corrected to obtain the SAR image product. Then, noise suppression and sea ice segmentation steps are performed. The noise suppression and sea ice segmentation steps will be further explained later.

[0043] like Figure 3 As shown, the steps for determining the correction coefficients based on radar parameters and the geometric relationship between slant range and incident angle may include: calculating the center slant range and range resolution based on radar parameters; calculating the incident angle corresponding to each column based on the center slant range and range resolution; using the beam center incident angle as a reference incident angle, determining the correction coefficients based on the beam center incident angle and the incident angle corresponding to each column.

[0044] Specifically, the incident angle correction steps may include the following:

[0045] (1) Read the scene file and parse the radar parameters. For example, data such as beam center incident angle, bandwidth, sampling rate, pulse delay number, pulse repetition frequency, and pulse width can be obtained.

[0046] (2) Calculate the center slope distance Where dpnum is the pulse delay number, PRF is the pulse repetition frequency, d is the pulse delay time, τ is the pulse width, and c is the speed of light.

[0047] (3) Calculate the distance resolution B w This refers to the signal bandwidth.

[0048] (4) Calculate the incident angle for each column.

[0049] ,

[0050] Where R iLet N be the slope distance corresponding to the i-th point, and N be the number of distance vector points. H represents the platform height, which can be calculated from the GPS location information in the auxiliary data.

[0051] (5) Incident angle correction (input is the image data at the end of imaging). The attenuation of radar echo intensity in the range direction can be obtained from geometric relationships:

[0052]

[0053] Where σ is the backscattering coefficient, characterizing the ability of ground objects to scatter radar signals, and θ is the incident angle, i.e., the angle between the radar beam and the normal to the target surface. P rec This indicates the received power, referring to the power of the target-scattered signal received by the radar system. A eff The effective receiving area represents the radar's ability to collect echo signals. P0 represents the transmit power, i.e., the initial power of the electromagnetic wave transmitted by the radar. G(θ) represents the antenna gain. σ target This represents the target's scattering cross section, characterizing the target's scattering efficiency of radar signals. According to the above formula, as the incident angle increases, the backscattering intensity decreases. Therefore, in the SAR product manufacturing process, the incident angle at the beam center can be selected as a reference angle for correction. The correction formula is as follows:

[0054]

[0055] Where θ ref For reference angle, σ is the correction factor. 0 (θ i ) represents the original backscattering coefficient (corresponding to the incident angle θ) i ).

[0056] In the incident angle correction stage, this embodiment adopts a SAR imaging process incident angle correction method based on physical model embedding. By directly integrating radiation and geometric correction into the imaging algorithm, the correction coefficient is dynamically calculated using radar system parameters and real-time geometric relationships (such as slant range, viewing angle, and digital elevation model). This eliminates the reliance on manual visual interpretation and statistical fitting in traditional methods, and realizes universal modeling of the relationship between sea ice backscattering intensity and incident angle, significantly improving processing efficiency and adaptability.

[0057] Please continue reading. Figure 1In some embodiments, the steps of using a deep learning network with an embedded wavelet transform module to perform multi-scale feature decomposition and reconstruction of SAR image products may include: inputting the SAR image product into the deep learning network, extracting deep features of the SAR image product by alternately using a first convolutional layer and a discrete wavelet transform module; and alternately using an inverse wavelet transform module and a second convolutional layer to fuse and reconstruct the decomposed deep features to obtain the original structural information that retains the deep features. Figure 4 The diagram illustrates the overall structure of a deep learning network according to an embodiment of the present invention. Figure 5 The diagram illustrates the structure of the residual channel attention mechanism module of a deep learning network according to an embodiment of the present invention.

[0058] like Figure 4 As shown, the deep learning network in this embodiment can be composed of a backbone network and a residual channel attention mechanism module. By combining wavelet transform and channel attention mechanism, the deep learning network can reduce speckle noise in SAR images, enhance feature extraction capabilities, and improve the processing effect of remote sensing images.

[0059] First, the SAR image product after incident angle correction can be input into the CR module (Conv2d convolution + ReLU activation function) for initial feature extraction. Subsequently, Discrete Wavelet Transform (DWT) can be used to decompose the features at multiple scales to obtain information at different frequencies. The deep learning network can alternate between the CR module and DWT at different stages to continuously extract deeper features while enhancing its ability to express features at different scales. The CR module, used alternately with DWT, can be referred to as the first convolutional layer.

[0060] During feature fusion and reconstruction, deep learning networks can reconstruct the features obtained from DWT decomposition using inverse wavelet transform (IWT), restoring the original structural information while preserving the enhancement effect of deep features. The alternating use of CR and IWT multiple times allows the network to maintain the advantages of decomposition features without losing important information from the original image. The CR module, used alternately with IWT, can be referred to as the second convolutional layer.

[0061] Continue reading Figure 4 Residual Channel Attention (RCAA) can be used to enhance the channels of the feature maps extracted by each first convolutional layer, and then the enhanced feature maps are passed to the corresponding second convolutional layer. RCAA can include 1×1 convolutional layers and depthwise convolutional layers (DWConv). The 1×1 convolutional layers and depthwise convolutional layers are connected, and the depthwise convolutional layers are used to perform cross-channel local interactions on the features reduced in dimensionality by the 1×1 convolutional layers. Figure 5As shown, in some embodiments, each RCAA module can consist of global average pooling, 1×1 convolution, depthwise convolution, and a sigmoid activation function. RCAA modules enhance key features by calculating attention weights between channels while suppressing irrelevant information, thus improving the network's expressive power and robustness. Multiple RCAA modules operate in parallel on feature maps at different stages, further optimizing information flow and ensuring the network focuses on important feature regions.

[0062] Continue reading Figure 4 The image is then further processed by a multi-layer perceptron (MLP) to obtain the final output image. Overall, this deep learning network, by combining wavelet transform, depthwise convolution, and attention mechanisms, improves the ability to restore details of remote sensing images while ensuring feature extraction capabilities. It can be applied to tasks such as super-resolution, denoising, or target detection of remote sensing images.

[0063] Compared to traditional denoising methods, this embodiment combines wavelet transform (including DWT and IWT) for multi-scale feature modeling. DWT can decompose the input image into sub-bands of different frequencies, thereby preserving more high-frequency detail information, while IWT can gradually recover the global structure during reconstruction. This approach automatically learns the optimal wavelet coefficient processing method, avoiding the limitations of traditional methods that rely on manually setting thresholds or basis functions, making the network more accurate in extracting and reconstructing features.

[0064] This embodiment also enhances channel-level feature representation by introducing RCAA. The deep learning network integrates RCAA modules at multiple stages, calculating attention weights through global average pooling, 1×1 convolutions, depthwise convolutions, and the sigmoid function to dynamically adjust the channel importance of the feature maps. Unlike traditional channel attention mechanisms, RCAA can aggregate information across multiple levels, making features at different depths more discriminative during fusion. This mechanism enhances the expressive power of key features while reducing interference from irrelevant information, enabling the network to extract important regions from remote sensing images more accurately.

[0065] Figure 6 The diagram schematically illustrates the overall structure of a sea ice segmentation model according to an embodiment of the present invention.

[0066] like Figure 6 As shown, the sea ice segmentation model can employ a deep learning architecture based on multi-scale feature extraction and progressive reconstruction, and can be used for tasks such as super-resolution, denoising, and remote sensing image processing. The processing of the sea ice segmentation model can include four stages: feature extraction, feature fusion, multi-scale information processing, and high-resolution reconstruction.

[0067] In the feature extraction stage, for example, if the input is an a×a image, it can be processed through multiple CRP (Conv2d convolution + ReLU activation function + pooling) modules to progressively extract deeper features while reducing the resolution. For example, it can be progressively downsampled to obtain a / 2×a / 2, a / 4×a / 4, and a / 8×a / 8 images. This progressive downsampling method not only extracts higher-level semantic information but also reduces computation and improves the network's processing efficiency.

[0068] In the feature fusion and multi-scale information processing stages, to address the insufficient ability to distinguish sea ice categories in complex scenarios, a Lightweight Adaptive Extraction (LAE) module can be integrated into the sea ice segmentation model network of this embodiment in the form of a residual network. By preserving more contextual information and high-resolution details on multi-scale feature maps, the impact of noise is reduced, and rich semantic information is retained. To improve the model's generalization ability, this embodiment also introduces a Multi-Scale Feature Extraction (MSFE) module, which improves feature representation ability through dual aggregation of spatial and channel features, replacing traditional skip connections and better supporting multi-level information fusion in the decoder.

[0069] After completing multi-scale feature extraction, the sea ice segmentation model network enters the progressive reconstruction stage, which utilizes the CRU (Conv2d convolution + ReLU activation function + Upsampling) module. The CRU module combines convolutional layers for feature extraction, the ReLU activation function to provide non-linear mapping, and the upsampling module to progressively restore resolution, enabling the network to recover high-resolution images from low-dimensional features. Furthermore, the network employs multiple Feature Fusion Modules (FFMs) during reconstruction for feature fusion at different stages. These modules can include channel attention or feature weighting mechanisms, allowing information from different stages to complement each other more effectively, improving the final reconstruction quality. The CRU and FFM modules can be used alternately to ensure effective information flow while features are progressively amplified.

[0070] In conclusion, such as Figure 6As shown, the sea ice segmentation model can take an a×a image as input and extract multi-scale feature maps, such as a / 2×a / 2, a / 4×a / 4, and a / 8×a / 8 resolutions. Each feature map is downsampled and enhanced by the LAE module, and features at different levels are fused through skip connections. The fused features are first optimized by the MSFE module to capture finer details, and then fused with shallow features by the FFM module. Finally, the processed features are upsampled to the original resolution to generate the final segmentation result or other task output. It should be noted that the specific resolution values ​​mentioned above are only examples and do not constitute a specific limitation of the present invention. The specific structure of the LAE and MSFE modules will be combined with... Figure 7 and Figure 8 Please provide an explanation.

[0071] Figure 7 The diagram illustrates the structure of a lightweight adaptive extraction module for a sea ice segmentation model according to an embodiment of the present invention.

[0072] like Figure 7 As shown, the Lightweight Adaptive Extraction (LAE) module can include a lightweight extraction branch and an adaptive extraction branch.

[0073] The lightweight branch can use grouped convolution for downsampling, reducing the computational cost of the model. Grouped convolution reduces the number of parameters and computational cost of convolution operations by dividing the input channels into multiple groups for independent processing, while still effectively preserving the spatial information in the image. This reduces memory usage and accelerates the model's computation process. At the same time, the lightweight branch retains key spatial information of the image, enabling the network to extract important local features while maintaining high efficiency.

[0074] Adaptive branches can capture global context through average pooling and convolution operations, further enhancing the model's spatial understanding. Average pooling allows the module to extract global information from the image, while convolution provides a deeper level of feature abstraction. The attention weights generated by these operations are applied to the features extracted by the lightweight branch, and then weighted and fused element-wise. This weighted fusion process effectively enhances the model's focus on important regions, enabling it to concentrate on identifying key targets in segmentation tasks while ignoring unimportant background information. The fused output feature map is then combined with the original input features through residual connections, further enhancing information flow and preventing information loss within the network. This ensures smoother gradient propagation during model training. This design not only helps avoid overfitting but also improves the model's convergence speed and training stability.

[0075] Figure 8The diagram illustrates the structure of a multi-scale feature extraction module for a sea ice segmentation model according to an embodiment of the present invention.

[0076] like Figure 8 As shown, the Multi-Scale Attention Convergence (MSFE) module can include spatial attention and channel attention mechanisms. Through spatial and channel attention mechanisms, feature extraction and contextual understanding capabilities in image segmentation tasks are significantly improved.

[0077] Spatial attention mechanisms enable the model to focus on key regions, enhancing the extraction of details and important parts, while channel attention mechanisms optimize feature selection, helping the model focus on more discriminative channel features. By aggregating multi-scale information, MSFE can handle objects of different scales, improving the segmentation of image details and complex backgrounds. Furthermore, MSFE enhances the model's generalization ability. Through the fusion of multi-scale features, the MSFE module allows the sea ice segmentation model to learn richer and more diverse feature representations, reducing overfitting and improving adaptability in different scenarios. This allows the MSFE module to not only improve segmentation accuracy but also enhance the model's robustness, improving its performance on new datasets and in different environments, significantly enhancing the model's generalization ability and applicability.

[0078] Based on the above embodiments, the comparative experimental results of the sea ice segmentation method of the present invention with other common segmentation methods (such as segmentation methods based on U-Net, UKAN, DeeplabV3+, TransUnet, TransFuse, MobilenetV3, BisNet, DDRNet, and PIDNet models) are shown in Table 1. In Table 1, PA represents pixel accuracy, mPA represents average pixel accuracy, and mIoU represents average intersection-over-union ratio. As can be seen from Table 1, among the listed segmentation methods, the sea ice segmentation method of the present invention achieves the highest mIoU of 84.028 and mPA of 91.346. The results in Table 1 show that the sea ice segmentation method of the present invention effectively solves the challenge of sea ice segmentation in SAR images and achieves a significant improvement compared with existing methods.

[0079] Table 1

[0080]

[0081] Based on the above embodiments, the present invention has at least the following advantages compared with the prior art:

[0082] (1) Automatic correction of incident angle. The incident angle correction method for SAR imaging process proposed in this invention realizes full-process automated processing, which can meet the needs of large-scale, near real-time operational monitoring (such as polar sea ice monitoring) without manual intervention; the method is based on physical model driving, which effectively avoids the errors caused by traditional manual segmentation and statistical fitting, and has higher accuracy and robustness, and can adapt to SAR data of multiple platforms and multiple modes; at the same time, by completing the correction process in sync with the imaging algorithm, the processing efficiency is significantly improved, and standardized high-quality data input is provided for subsequent AI analysis.

[0083] (2) Effectively suppresses speckle noise while preserving texture details. This invention combines wavelet transform (DWT / IWT) for multi-scale feature modeling. DWT can decompose the input image into sub-bands of different frequencies, thereby preserving more high-frequency detail information, while IWT can gradually recover the global structure during reconstruction. This method automatically learns the optimal wavelet coefficient processing method, avoiding the limitations of traditional methods that rely on manually setting thresholds or basis functions, making the network more accurate in extracting and reconstructing features.

[0084] (3) Higher segmentation accuracy in complex scenes. By introducing the LAE and MSFE modules, this invention can retain more contextual information and high-resolution details on multi-scale feature maps, significantly reduce the impact of noise, and enhance the expressive power of semantic information. This makes the model perform better in complex scenes and significantly improves segmentation accuracy.

[0085] (4) Enhanced model generalization ability. This invention designs an MSFE module, which significantly improves feature representation ability through a dual aggregation mechanism of spatial and channel data. The MSFE module replaces the traditional skip connections, better supports multi-level information fusion of the decoder, and makes the model more stable in SAR sea ice images of different regions and categories, with significantly enhanced generalization ability.

[0086] Based on the aforementioned SAR sea ice segmentation method based on multi-scale feature extraction and fusion, this invention also provides a SAR sea ice segmentation device based on multi-scale feature extraction and fusion. The following will combine... Figure 9 The device is described in detail.

[0087] Figure 9 The diagram illustrates the structure of a SAR sea ice segmentation device based on multi-scale feature extraction and fusion according to an embodiment of the present invention.

[0088] like Figure 9 As shown, the SAR sea ice segmentation device 900 based on multi-scale feature extraction and fusion in this embodiment includes a correction module 910, a denoising module 920, and a segmentation module 930.

[0089] The correction module 910 can be used to correct the incident angle of SAR sea ice image data during the imaging process to obtain SAR image products of sea ice. In one embodiment, the correction module 910 can be used to perform the operation S110 described above, which will not be repeated here.

[0090] The denoising module 920 can be used to perform multi-scale feature decomposition and reconstruction of SAR image products using a deep learning network embedded with a wavelet transform module, and to perform channel enhancement using a residual channel attention mechanism in the deep learning network to obtain SAR image products with speckle noise suppression. In one embodiment, the denoising module 920 can be used to perform the operation S120 described above, which will not be repeated here.

[0091] The segmentation module 930 can be used to segment SAR image products after speckle noise suppression based on a sea ice segmentation model that integrates a lightweight adaptive extraction module and a multi-scale feature extraction module, to obtain the sea ice segmentation result. In one embodiment, the segmentation module 930 can be used to perform the operation S130 described above, which will not be repeated here.

[0092] For details regarding the relevant content, please refer to the previous text; it will not be repeated here.

[0093] According to embodiments of the present invention, any plurality of the above modules can be combined into one module, or any one of the modules can be split into multiple modules. Alternatively, at least a portion of the functionality of one or more of these modules can be combined with at least a portion of the functionality of other modules and implemented in one module. According to embodiments of the present invention, at least one of the above modules can be at least partially implemented as hardware circuitry, such as a field-programmable gate array (FPGA), a programmable logic array (PLA), a system-on-a-chip, a system-on-a-substrate, a system-on-package, an application-specific integrated circuit (ASIC), or implemented in hardware or firmware by any other reasonable means of integrating or packaging the circuitry, or implemented in any one of software, hardware, and firmware methods, or in a suitable combination of any of these. Alternatively, at least one of the above modules can be at least partially implemented as a computer program module, which, when run, can perform corresponding functions.

[0094] Figure 10 A block diagram of an electronic device suitable for implementing a SAR sea ice segmentation method based on multi-scale feature extraction fusion according to an embodiment of the present invention is shown schematically.

[0095] like Figure 10As shown, an electronic device 1000 according to an embodiment of the present invention includes a processor 1001, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage portion 1008 into a random access memory (RAM) 1003. The processor 1001 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 1001 may also include onboard memory for caching purposes. The processor 1001 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present invention.

[0096] RAM 1003 stores various programs and data required for the operation of electronic device 1000. Processor 1001, ROM 1002, and RAM 1003 are interconnected via bus 1004. Processor 1001 executes various operations of the method flow according to embodiments of the present invention by executing programs in ROM 1002 and / or RAM 1003. It should be noted that programs may also be stored in one or more memories other than ROM 1002 and RAM 1003. Processor 1001 may also execute various operations of the method flow according to embodiments of the present invention by executing programs stored in one or more memories.

[0097] According to an embodiment of the present invention, the electronic device 1000 may further include an input / output (I / O) interface 1005, which is also connected to a bus 1004. The electronic device 1000 may also include one or more of the following components connected to the input / output (I / O) interface 1005: an input section 1006 including a keyboard, mouse, etc.; an output section 1007 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 1008 including a hard disk, etc.; and a communication section 1009 including a network interface card such as a LAN card, modem, etc. The communication section 1009 performs communication processing via a network such as the Internet. A drive 1010 is also connected to the input / output (I / O) interface 1005 as needed. A removable medium 1011, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 1010 as needed so that computer programs read from it can be installed into the storage section 1008 as needed.

[0098] The present invention also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of the present invention.

[0099] According to embodiments of the present invention, a computer-readable storage medium may be a non-volatile computer-readable storage medium, such as including, but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of the present invention, a computer-readable storage medium may include ROM 1002 and / or RAM 1003 and / or one or more memories other than ROM 1002 and RAM 1003 described above.

[0100] Embodiments of the present invention also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code enables the computer system to implement the SAR sea ice segmentation method based on multi-scale feature extraction and fusion provided in the embodiments of the present invention.

[0101] When the computer program is executed by the processor 1001, it performs the functions defined in the system / apparatus of this invention. According to embodiments of the invention, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0102] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and may be downloaded and installed via the communication section 1009, and / or installed from a removable medium 1011. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0103] In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 1009, and / or installed from the removable medium 1011. When the computer program is executed by the processor 1001, it performs the functions defined in the system of this embodiment of the invention. According to embodiments of the invention, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0104] According to embodiments of the present invention, program code for executing the computer programs provided in the embodiments of the present invention can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0105] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0106] Those skilled in the art will understand that the features described in the various embodiments of the present invention can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in the present invention. In particular, the features described in the various embodiments of the present invention can be combined and / or combined in various ways without departing from the spirit and teachings of the present invention. All such combinations and / or combinations fall within the scope of the present invention.

[0107] The embodiments of the present invention have been described above. However, these embodiments are merely illustrative and not intended to limit the scope of the invention. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of the invention, and all such substitutions and modifications should fall within the scope of the invention.

Claims

1. A SAR sea ice segmentation method based on multi-scale feature extraction fusion, characterized in that, The method comprises the following steps: During imaging of SAR sea ice image data, the SAR sea ice image data is corrected in terms of incident angle to obtain a SAR image product of sea ice; A deep learning network embedded with a wavelet transform module is used to perform multi-scale feature decomposition and reconstruction on the SAR image product, and a residual channel attention mechanism is used in the deep learning network to perform channel enhancement, thereby obtaining a SAR image product after speckle noise suppression; The deep learning network embedded with the wavelet transform module is used to perform multi-scale feature decomposition and reconstruction on the SAR image product, which comprises the following steps: the SAR image product is input into the deep learning network, and deep-level features of the SAR image product are extracted by alternately using a first convolutional layer and a discrete wavelet transform module; an inverse wavelet transform module and a second convolutional layer are alternately used to fuse and reconstruct the deep-level features obtained by decomposition, thereby obtaining original structure information of the deep-level features; the residual channel attention mechanism is used to perform channel enhancement on each feature map extracted by the first convolutional layer, and the feature map after channel enhancement is transmitted to the corresponding second convolutional layer; A sea ice segmentation model integrated with a lightweight adaptive extraction module and a multi-scale feature extraction module is used to segment the SAR image product after speckle noise suppression, thereby obtaining a segmentation result of the sea ice.

2. The SAR sea ice segmentation method of claim 1, wherein, The step of correcting the SAR sea ice image data in terms of incident angle comprises the following steps: Radar parameters are parsed from a scene element file, and a correction coefficient is determined according to the radar parameters and a geometric relationship between slant range and incident angle; According to the correction coefficient, backscattering coefficients of the SAR sea ice image data at different incident angles are uniformly corrected to theoretical values at a reference incident angle.

3. The SAR sea ice segmentation method of claim 2, wherein, The step of determining the correction coefficient according to the radar parameters and the geometric relationship between slant range and incident angle comprises the following steps: According to the radar parameters, a central slant range and a distance resolution are calculated; According to the central slant range and the distance resolution, an incident angle corresponding to each column is calculated; The beam center incident angle is taken as the reference incident angle, and the correction coefficient is determined according to the beam center incident angle and the incident angle corresponding to each column.

4. The SAR sea ice segmentation method of claim 1, wherein, The residual channel attention mechanism comprises a 1×1 convolutional layer and a deep convolutional layer, the 1×1 convolutional layer is connected to the deep convolutional layer, and the deep convolutional layer is used to perform cross-channel local interaction on features after dimension reduction of the 1×1 convolutional layer.

5. The SAR sea ice segmentation method of claim 1, wherein, The step of segmenting the SAR image product after speckle noise suppression to obtain the segmentation result of the sea ice comprises the following steps: The SAR image product after speckle noise suppression is sequentially subjected to feature extraction, feature fusion, multi-scale information processing and high-resolution reconstruction to determine the segmentation result of the sea ice.

6. The SAR sea ice segmentation method of claim 5, wherein, The step of sequentially performing feature extraction, feature fusion, multi-scale information processing and high-resolution reconstruction on the SAR image product after speckle noise suppression comprises the following steps: An initial multi-scale feature map is extracted through a convolutional layer of the sea ice segmentation model; The lightweight adaptive extraction module is used to down-sample and enhance features of each initial multi-scale feature map, and different levels of feature maps are fused through a skip connection; The multi-scale feature extraction module is used to capture detail information from the fused feature map and fuse it with shallow features to complete multi-scale information processing of features; The features processed by the multi-scale information processing are up-sampled to the original resolution to obtain the segmentation result.

7. A SAR sea ice segmentation device based on multi-scale feature extraction fusion, characterized in that, It comprises: A correction module is configured to correct the incidence angle of the SAR sea ice image data during the imaging process of the SAR sea ice image data to obtain a SAR image product of sea ice; A denoising module is configured to use a deep learning network embedded with a wavelet transform module to decompose and reconstruct multi-scale features of the SAR image product, and use a residual channel attention mechanism in the deep learning network to enhance channels to obtain a SAR image product after suppressing speckle noise; the denoising module is also configured to input the SAR image product into the deep learning network, and extract deep features of the SAR image product by alternately using a first convolutional layer and a discrete wavelet transform module; An inverse wavelet transform module and a second convolutional layer are alternately used to fuse and reconstruct the deep features obtained by decomposition to obtain original structural information that retains deep features; The residual channel attention mechanism is used to enhance channels of each feature map extracted by the first convolutional layer, and the feature map after channel enhancement is transmitted to the corresponding second convolutional layer; A segmentation module is configured to segment the SAR image product after suppressing speckle noise based on a sea ice segmentation model integrated with a lightweight adaptive extraction module and a multi-scale feature extraction module to obtain a segmentation result of the sea ice.

8. An electronic device, comprising: It comprises: One or more processors; A storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors execute the method according to any one of claims 1 to 6.

Images