Target detection method and system based on SAR imaging and speckle removal
Through the multi-scale window Swin Transformer module and triple distillation model learning method, high-precision and integrated processing from synthetic aperture radar image imaging to target detection is achieved, solving the problems of low processing efficiency and error accumulation in the existing technology, and improving the Information capture capabilities for image imaging and decoherent spots.
Patent Information
- Application Number
- CN202411116786.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-15
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2044-08-15
AI Technical Summary
The prior art is difficult to achieve efficient and precise integrated processing from synthetic aperture radar image imaging, decoherent spots to target detection in complex scenarios, and traditional methods have problems of low processing efficiency and error accumulation.
The SAR imaging and decoherent patch module are constructed using the multi-scale window Swin Transformer module, and combined with the object detection module, the object detection model is trained through the triple distillation model learning method to realize real-time high-precision processing from SAR image imaging to object detection.
It improves the information capture capabilities of image imaging and decoherent spots, enhances image reconstruction capabilities, and realizes high-precision and integrated processing from SAR image imaging to object detection, improving processing efficiency and accuracy.
Smart Images

Figure CN119027684A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of image processing, and in particular to a target detection method and system based on SAR imaging and decoherence speckle. Background Art
[0002] Synthetic aperture radar (SAR) imaging technology mainly includes frequency domain processing, time domain processing and deep learning algorithms. In frequency domain processing, the range Doppler (RD) algorithm achieves high-resolution imaging by separating the range and azimuth dimensions, but does not consider the movement of range units, which limits the imaging quality. Therefore, improved algorithms are proposed to partially solve this problem, such as the quadratic range compression (SRC) algorithm and the frequency modulation scaling (CS) algorithm, but the improved algorithms have limited effects in large squint scenes. The ωK algorithm is further proposed, which effectively corrects the range migration through Stolt interpolation and can be applied to various squint angles. In time domain processing, the back projection algorithm (BPA) achieves accurate imaging by compensating the propagation delay of the pulse signal point by point, but there are problems of large computational complexity and low efficiency. In the application of deep learning technology, self-focusing imaging based on recurrent neural network (RNN) improves the accuracy and robustness of imaging by automatically learning features; in addition, fast back projection based on convolutional neural network (CNN) and SAR-GMTIm framework for moving targets have significantly improved the imaging speed and quality.
[0003] In the study of despeckle of synthetic aperture radar images, the main methods used include spatial filtering, transform domain filtering and deep learning. In spatial filtering, filters such as Lee, Kuan and Frost are used to reduce speckle by calculating the weighted average of pixel intensities in a local window. Although they perform well in uniform areas, they have poor edge effects and the effect decreases as the window expands. In transform domain filtering, wavelet domain filtering improves denoising by threshold processing after logarithmic transformation and wavelet decomposition, but it may cause distortion; the Bayesian wavelet threshold shrinkage algorithm improves the despeckle effect by assuming a Gaussian distribution. In deep learning technology, PSD, ID-CNN and Speckle2Void achieve denoising by learning the mapping relationship between noise and clear images; SAR-CNN uses homomorphic transformation and residual network to improve noise processing; GAN combined with RED-NET significantly improves denoising efficiency and quality.
[0004] In the ship detection of synthetic aperture radar images, traditional methods mainly include contrast information, template matching, statistical analysis and deep learning methods. Constant False Alarm Rate (CFAR) is the most commonly used synthetic aperture radar image target detection technology. This method has achieved good performance in some simple scenarios, but the performance of this method depends to a large extent on the construction of the sea clutter statistical model and the parameter estimation of the constructed model. Therefore, considering the inhomogeneity of clutter and various clutter models, a method based on heterogeneous distributions, such as alpha-stable distribution and gamma distribution, is proposed to fit variable ocean conditions.
[0005] However, with the continuous development of technology, these algorithms can no longer meet modern needs and are difficult to apply to various complex synthetic aperture radar image scenes. Previous studies have treated synthetic aperture radar image despeckle and target detection separately. The separate processing method requires multiple data transmissions and independent processing, resulting in low processing efficiency. In addition, there is a lack of information interaction and utilization between each link, and the processing results are limited by the local information processed at each stage, resulting in error accumulation in downstream tasks, and unable to provide fast and high-precision technology. Therefore, how to achieve real-time and high-precision integrated processing from synthetic aperture radar image imaging, despeckle to target detection tasks has become a difficult problem in the field of synthetic aperture radar technology research. Summary of the invention
[0006] In order to solve the above technical problems, the purpose of the present invention is to provide a target detection method and system based on SAR imaging and despeckle, which can realize the integrated processing from SAR image imaging, despeckle to target detection tasks in real time and with high precision.
[0007] The first technical solution adopted by the present invention is: a target detection method based on SAR imaging and despeckle, comprising the following steps:
[0008] Collect samples on high-resolution remote sensing images and construct training data sets;
[0009] Construct SAR imaging module based on multi-scale window Swin Transformer module;
[0010] Construct a speckle removal module based on the multi-scale window Swin Transformer module;
[0011] Constructing a target detection module, and combining the SAR imaging module and the speckle removal module to obtain a target detection model based on SAR imaging and speckle removal;
[0012] The target detection model based on SAR imaging and despeckle is trained based on the training data set and the triple distillation model learning method to obtain a trained target detection model based on SAR imaging and despeckle.
[0013] Furthermore, the training data set includes SAR images, despeckled SAR images, and target detection labels.
[0014] Furthermore, the SAR imaging module includes a convolutional layer, a residual block and a multi-scale window Swin Transformer module, wherein:
[0015] The multi-scale window Swin Transformer module is used to learn the global and local features of the SAR image;
[0016] The residual block introduces a skip connection to solve the potential gradient vanishing problem in the deep neural network and further extract feature information in the image;
[0017] The convolutional layer is used to remove speckle noise from SAR images.
[0018] Further, the multi-scale window Swin Transformer module includes a first subunit and a second subunit, wherein:
[0019] The first subunit includes a Layer Norm layer, a local feature extraction module and a first multi-window attention module;
[0020] The second subunit includes a Layer Norm layer, a local feature extraction module, and a second multi-window attention module;
[0021] The first multi-window attention module adopts a window multi-head self-attention mechanism;
[0022] The second multi-window attention module adopts a sliding window multi-head self-attention mechanism.
[0023] Furthermore, the first multi-window attention module and the second multi-window attention module use multiple windows of different sizes to perform self-attention calculations.
[0024] Further, the step of training the target detection model based on SAR imaging and despeckle based on the training data set and the triple distillation model learning method to obtain the trained target detection model based on SAR imaging and despeckle specifically includes:
[0025] The existing trained SAR imaging model, speckle decorrelation model and target detection model are used as the first teacher model, the second teacher model and the third teacher model respectively;
[0026] Inputting the training data set into the SAR imaging module and the first teacher model respectively to obtain a first prediction result and a first teacher model prediction result;
[0027] Constructing a first distillation loss function based on the first prediction result and the first teacher model prediction result;
[0028] Constructing an imaging loss function based on the first prediction result and the real image;
[0029] Inputting the training data set into the speckle removal module and the second teacher model respectively to obtain a second prediction result and a second teacher model prediction result;
[0030] Constructing a second distillation loss function based on the second prediction result and the second teacher model prediction result;
[0031] Constructing a speckle removal loss function based on the second prediction result and the real image;
[0032] Inputting the training data set into the target detection module and the third teacher model respectively to obtain a third prediction result and a third teacher model prediction result;
[0033] Constructing a third distillation loss function based on the third prediction result and the third teacher model prediction result;
[0034] Constructing a detection loss function based on the third prediction result and the target detection label;
[0035] Introducing an additional loss term, and assigning weights to the first distillation loss function, the imaging loss function, the second distillation loss function, the despeckle loss function, the third distillation loss function, and the detection loss function to obtain a global balanced loss function;
[0036] The parameters of the target detection model based on SAR imaging and despeckle are adjusted based on the global balance loss function to obtain a trained target detection model based on SAR imaging and despeckle.
[0037] The second technical solution adopted by the present invention is: a target detection system based on SAR imaging and despeckle, comprising:
[0038] Data acquisition module, used to collect samples on high-resolution remote sensing images and build training data sets;
[0039] SAR imaging module, used for imaging the input SAR echo signal;
[0040] A speckle removal module; used for performing speckle removal processing on the imaging result output by the SAR imaging module;
[0041] A target detection module is used to perform image detection on the SAR image after de-speckle removal;
[0042] The model training module trains the target detection model based on SAR imaging and despeckle based on a training data set and a triple distillation model learning method to obtain a trained target detection model based on SAR imaging and despeckle.
[0043] The present invention also proposes a target detection device based on SAR imaging and despeckle, comprising:
[0044] at least one processor;
[0045] at least one memory for storing at least one program;
[0046] When the at least one program is executed by the at least one processor, the at least one processor implements the target detection method based on SAR imaging and decorrelation as described above.
[0047] The beneficial effects of the method and system of the present invention are as follows: the present invention proposes a multi-scale window SwinTransformer module, and uses the module to construct a SAR imaging module and a despeckle module, which can capture the global and local relationships in the image, thereby obtaining richer information, helping to improve the model's ability to represent images and enhance image reconstruction capabilities; by combining the target detection module, the SAR imaging module and the despeckle module together, a target detection model based on SAR imaging and despeckle is constructed, and the model is trained using a triple distillation model training method to improve training accuracy, thereby obtaining a trained target detection model based on SAR imaging and despeckle. The trained target detection model based on SAR imaging and despeckle can realize the integrated processing of tasks from SAR image imaging, despeckle to target detection in real time and with high accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0048] Figure 1 It is a flow chart of the steps of a target detection method based on SAR imaging and despeckle of the present invention;
[0049] Figure 2 It is a structural block diagram of a target detection system based on SAR imaging and despeckle of the present invention;
[0050] Figure 3 It is a structural diagram of a target detection model based on SAR imaging and despeckle of a target detection method based on SAR imaging and despeckle of the present invention;
[0051] Figure 4It is a structural diagram of an imaging module of a target detection method based on SAR imaging and decoherence speckle of the present invention;
[0052] Figure 5 It is a multi-scale window SwinTransformer module structure diagram of a target detection method based on SAR imaging and decoherence speckle of the present invention;
[0053] Figure 6 It is a structural diagram of a local feature extraction module of a target detection method based on SAR imaging and decoherence speckle of the present invention;
[0054] Figure 7 It is a schematic diagram of triple distillation model learning and loss calculation of a target detection method based on SAR imaging and decoherence speckle of the present invention. DETAILED DESCRIPTION
[0055] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The step numbers in the following embodiments are only provided for the convenience of explanation and description, and the order between the steps is not limited in any way. The execution order of each step in the embodiment can be adaptively adjusted according to the understanding of those skilled in the art.
[0056] Reference Figure 1 The present invention provides a target detection method based on SAR imaging and despeckle, the method comprising the following steps:
[0057] S1. Collect samples on high-resolution remote sensing images and build a training data set;
[0058] Specifically, the training data set includes SAR images, despeckled SAR images, and target detection labels.
[0059] S2, build SAR imaging module based on multi-scale window Swin Transformer module;
[0060] Specifically, refer to Figure 4 The SAR imaging module includes two multi-scale window Swin Transformer modules (Multi scale Window Swin Transformer, MSW Swin Transformer), a convolutional layer (Conv) and a residual block.
[0061] The implementation process of the SAR imaging module specifically includes: first, the global and local features of the SAR image are learned through two multi-scale window Swin Transformer modules. The multi-scale window Swin Transformer module captures the global and local relationships in the image through a multi-scale window self-attention mechanism to obtain richer information. This helps to improve the model's ability to represent images and enhance image reconstruction capabilities. Then, four residual blocks are used to further extract feature information from the image. The residual block solves the potential gradient vanishing problem in deep neural networks by introducing jump connections, thereby ensuring the stability and efficiency of the network during deep training; at the same time, the residual block can accelerate the convergence of the training process and help the network learn more complex and abstract feature representations. Finally, the speckle noise of the SAR image is removed through two convolutional layers.
[0062] The self-attention mechanism in Transformer helps the model capture the global and local relationships in the image. This mechanism enables Transformer to perform well in some tasks that require understanding the global structure of the image. Since the Transformer model may face high computational complexity when processing large-scale images, Swin Transformer is proposed; Swin Transformer reduces the computational complexity to a certain extent and improves the computational efficiency of the model by introducing a multi-window self-attention mechanism. The Swin Transformer module can use the self-attention mechanism to capture the global and local context information in the image, which helps to distinguish noise and signals and better understand the structure and content of the image. However, due to the local effect and boundary effect of the self-attention mechanism of the Swin Transformer module with a fixed window size, it is easy to split a single target into multiple windows, resulting in inaccurate self-attention calculation. Therefore, based on the Swin Transformer module, the present invention proposes a multi-scale window Swin Transformer module, which introduces a multi-window attention module (MWA). Smaller windows help the model focus more on the subtle structures and details in the image, while larger windows are more suitable for obtaining global context information.
[0063] Reference Figure 5 , the multi-scale window Swin Transformer module comprises a first subunit and a second subunit, wherein:
[0064] The output of the first subunit is the input of the second subunit;
[0065] The first subunit includes a Layer Norm layer, a local feature extraction module and a first multi-window attention module;
[0066] The second subunit includes a Layer Norm layer, a local feature extraction module, and a second multi-window attention module;
[0067] The first multi-window attention module adopts a window multi-head self-attention mechanism;
[0068] The second multi-window attention module adopts a sliding window multi-head self-attention mechanism.
[0069] The first multi-window attention module and the second multi-window attention module each contain three branches, each branch uses three windows of different sizes, which are 4, 8, and 16 respectively. In the following SAR image despeckle task, obtaining more contextual information is crucial to improving model performance. Therefore, the present invention designs three windows of different sizes and performs self-attention calculation in each window; then, the similarity probability maps output by the self-attention mechanism of the three windows of different sizes are added and normalized.
[0070] Reference Figure 6 In the multi-scale window Swin Transformer module of the present invention, a localized feature extraction module (LFE) is used to replace the MLP layer. The local feature extraction module consists of two 1*1 convolutional layers and a 3*3 convolutional layer with a void rate r=2. The original MLP layer relies on linear transformation and has a weak ability to abstract features. The introduction of the local feature extraction module LFE to replace the MLP layer enables the network to better extract local features of the image in the task of removing speckle noise, and show stronger image detail texture capabilities.
[0071] S3, constructing a speckle removal module based on a multi-scale window Swin Transformer module;
[0072] Specifically, the speckle removal module is used to perform speckle removal processing on the imaging result output by the SAR imaging module, and its structure is the same as that of the SAR imaging module in step S2.
[0073] S4, constructing a target detection module, and combining the SAR imaging module and the speckle removal module to obtain a target detection model based on SAR imaging and speckle removal;
[0074] Specifically, refer to Figure 3, the encoding module, SAR imaging module, coherent speckle module and target detection module are combined into a single model to obtain a target detection model based on SAR imaging and decorrelation.
[0075] S5. Training the target detection model based on SAR imaging and despeckle based on the training data set and the triple distillation model learning method to obtain a trained target detection model based on SAR imaging and despeckle.
[0076] Specifically, refer to Figure 7 The triple distillation model learning method proposed in the present invention is a training method based on the "teacher-student network concept". It is simple, effective, and widely used in the industry. By distilling the feature representation knowledge learned by a complex network with strong learning ability and passing it to a network with small parameters and weak learning ability, distillation can provide soft label information that the student model cannot learn from the true value label. These soft label information contain inter-category information and feature representation knowledge that the teacher network can learn but the student network cannot learn, so the accuracy of the student network can generally be improved. In a specific embodiment of the present invention, the existing trained SAR imaging model, de-coherent speckle model, and target detection model are used as the first teacher model, the second teacher model, and the third teacher model to guide the SAR imaging module, the de-coherent speckle module, and the target detection module to perform learning and training, respectively. The specific process is as follows:
[0077] Inputting the training data set into the SAR imaging module and the first teacher model respectively to obtain a first prediction result and a first teacher model prediction result;
[0078] Constructing a first distillation loss function based on the first prediction result and the first teacher model prediction result;
[0079] Constructing an imaging loss function based on the first prediction result and the real image;
[0080] Inputting the training data set into the speckle removal module and the second teacher model respectively to obtain a second prediction result and a second teacher model prediction result;
[0081] Constructing a second distillation loss function based on the second prediction result and the second teacher model prediction result;
[0082] Constructing a speckle removal loss function based on the second prediction result and the real image;
[0083] Inputting the training data set into the target detection module and the third teacher model respectively to obtain a third prediction result and a third teacher model prediction result;
[0084] Constructing a third distillation loss function based on the third prediction result and the third teacher model prediction result;
[0085] A detection loss function is constructed based on the third prediction result and the target detection label.
[0086] The first distillation loss function, the imaging loss function, the second distillation loss function, the despeckle loss function, the third distillation loss function and the detection loss function together constitute the overall loss function of the target detection model based on SAR imaging and despeckle. Among them, the first distillation loss function is used to guide the SAR imaging module to learn the knowledge of the first teacher model; the second distillation loss function is used to guide the despeckle module to learn the knowledge of the second teacher model; the third distillation loss function is used to guide the target detection module to learn the knowledge of the third teacher model; the distillation loss function measures the difference between the prediction results of the student model and the prediction results of the teacher model, with the aim of enabling the student model to imitate the prediction behavior of the teacher model as much as possible. The imaging loss function, the despeckle loss function and the detection loss function are corresponding loss functions specifically designed for the three tasks of SAR imaging, despeckle and target detection, and are mainly used to measure the prediction accuracy of the model on the real label. The specific embodiment of the present invention uses the mean square error loss function at the pixel level as the imaging loss function and the despeckle loss function to measure the difference between the generated image and the real image; and uses the cross entropy loss function or the loss function specific to target detection as the detection loss function to measure the prediction accuracy of the model for the target.
[0087] In order to weigh the loss functions of each task in the entire deep learning model, the present invention assigns weights to the first distillation loss function, the imaging loss function, the second distillation loss function, the despeckle loss function, the third distillation loss function and the detection loss function, and performs weighted summation. The training importance between different tasks is balanced by adjusting the weights of them, and the weight coefficients can be adjusted according to actual conditions to achieve balance and optimization of each task. Generally, the weight coefficients can be determined by cross-validation and other techniques so that the model achieves optimal performance on each task. And in order to further enhance the role of global balance loss, the specific embodiments of the present invention combine the correlation and mutual influence between tasks, and introduce correlation penalty terms between tasks or cross loss terms between tasks as additional loss terms. The additional loss terms can help the model better understand the relationship between tasks, thereby achieving better balance and comprehensive effects.
[0088] Reference Figure 2 The present invention provides a target detection system based on SAR imaging and despeckle, comprising:
[0089] Data acquisition module, used to collect samples on high-resolution remote sensing images and build training data sets;
[0090] SAR imaging module, used for imaging the input SAR echo signal;
[0091] A speckle removal module; used for performing speckle removal processing on the imaging result output by the SAR imaging module;
[0092] A target detection module is used to perform image detection on the SAR image after de-speckle removal;
[0093] The model training module trains the target detection model based on SAR imaging and despeckle based on a training data set and a triple distillation model learning method to obtain a trained target detection model based on SAR imaging and despeckle.
[0094] The contents of the above method embodiments are all applicable to the present system embodiments. The functions specifically implemented by the present system embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.
[0095] A target detection device based on SAR imaging and despeckle:
[0096] at least one processor;
[0097] at least one memory for storing at least one program;
[0098] When the at least one program is executed by the at least one processor, the at least one processor implements the target detection method based on SAR imaging and decorrelation as described above.
[0099] The contents of the above method embodiments are all applicable to the present device embodiments. The functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.
[0100] A storage medium stores instructions executable by a processor, wherein the instructions executable by the processor are used to implement the target detection method based on SAR imaging and de-coherence speckle as described above when executed by the processor.
[0101] The contents of the above method embodiments are all applicable to the present storage medium embodiments. The functions specifically implemented by the present storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.
[0102] The above is a specific description of the preferred implementation of the present invention, but the invention is not limited to the embodiments. Those skilled in the art may make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.
Claims
1. A target detection method based on SAR imaging and despeckle, characterized in that: The following steps are involved: Collect samples on high-resolution remote sensing images and construct training data sets; Construct SAR imaging module based on multi-scale window Swin Transformer module; Construct a speckle removal module based on the multi-scale window Swin Transformer module; Constructing a target detection module, and combining the SAR imaging module and the speckle removal module to obtain a target detection model based on SAR imaging and speckle removal; The target detection model based on SAR imaging and despeckle is trained based on the training data set and the triple distillation model learning method to obtain a trained target detection model based on SAR imaging and despeckle.
2. The target detection method based on SAR imaging and despeckle according to claim 1, characterized in that: The training data set includes a SAR image, a despeckled SAR image, and a target detection label.
3. The target detection method based on SAR imaging and despeckle according to claim 1, characterized in that: The SAR imaging module includes a convolutional layer, a residual block and a multi-scale window Swin Transformer module, wherein: The multi-scale window Swin Transformer module is used to learn the global and local features of the SAR image; The residual block introduces a skip connection to solve the potential gradient vanishing problem in the deep neural network and further extract feature information in the image; The convolutional layer is used to remove speckle noise from SAR images.
4. The target detection method based on SAR imaging and despeckle according to claim 1, characterized in that: The multi-scale window Swin Transformer module comprises a first subunit and a second subunit, wherein: The first subunit includes a LayerNorm layer, a local feature extraction module and a first multi-window attention module; The second subunit includes a LayerNorm layer, a local feature extraction module, and a second multi-window attention module; The first multi-window attention module adopts a window multi-head self-attention mechanism; The second multi-window attention module adopts a sliding window multi-head self-attention mechanism.
5. The target detection method based on SAR imaging and despeckle according to claim 4, characterized in that: The first multi-window attention module and the second multi-window attention module use multiple windows of different sizes to perform self-attention calculations.
6. The target detection method based on SAR imaging and despeckle according to claim 1, characterized in that: The step of training the target detection model based on SAR imaging and despeckle based on the training data set and the triple distillation model learning method to obtain the trained target detection model based on SAR imaging and despeckle specifically includes: The existing trained SAR imaging model, speckle decorrelation model and target detection model are used as the first teacher model, the second teacher model and the third teacher model respectively; Inputting the training data set into the SAR imaging module and the first teacher model respectively to obtain a first prediction result and a first teacher model prediction result; Constructing a first distillation loss function based on the first prediction result and the first teacher model prediction result; Constructing an imaging loss function based on the first prediction result and the real image; Inputting the training data set into the speckle removal module and the second teacher model respectively to obtain a second prediction result and a second teacher model prediction result; Constructing a second distillation loss function based on the second prediction result and the second teacher model prediction result; Constructing a speckle removal loss function based on the second prediction result and the real image; Inputting the training data set into the target detection module and the third teacher model respectively to obtain a third prediction result and a third teacher model prediction result; Constructing a third distillation loss function based on the third prediction result and the third teacher model prediction result; Constructing a detection loss function based on the third prediction result and the target detection label; Introducing an additional loss term, and assigning weights to the first distillation loss function, the imaging loss function, the second distillation loss function, the despeckle loss function, the third distillation loss function, and the detection loss function to obtain a global balanced loss function; The parameters of the target detection model based on SAR imaging and despeckle are adjusted based on the global balance loss function to obtain a trained target detection model based on SAR imaging and despeckle.
7. A target detection system based on SAR imaging and despeckle, characterized in that: include: Data acquisition module, used to collect samples on high-resolution remote sensing images and build training data sets; SAR imaging module, used for imaging the input SAR echo signal; A speckle removal module is used to perform speckle removal processing on the imaging result output by the SAR imaging module; A target detection module is used to perform image detection on the SAR image after de-speckle removal; The model training module trains the target detection model based on SAR imaging and despeckle based on a training data set and a triple distillation model learning method to obtain a trained target detection model based on SAR imaging and despeckle.
8. A target detection device based on SAR imaging and despeckle, characterized in that: include: at least one processor; at least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the target detection method based on SAR imaging and despeckle as described in any one of claims 1 to 6.
Citation Information
Patent Citations
SAR image speckle filtering and target detection integrated method and device
CN110717876A
SAR (Synthetic Aperture Radar) image denoising method based on swin transform and wavelet transform
CN117036191A
A computer-implemented method of enhancing object detection in a digital image of known underlying structure, and corresponding module, data processing apparatus and computer program
WO2023047118A1
