Terrain classification method and system based on SAR altimeter delay Doppler image

Abstract

The invention discloses a terrain classification method and system based on an SAR altimeter delay Doppler image, and the method comprises the following steps: carrying out the calculation based on given DEM data, obtaining a terrain factor, carrying out the analysis of the terrain factor, and determining a terrain type; establishing an SAR altimeter echo model, and generating a delay Doppler image through simulation based on given DEM data; taking a terrain category obtained by terrain factor analysis as a label, labeling a corresponding delay Doppler image, inputting the labeled delay Doppler image into a convolutional neural network for training, and enabling the convolutional neural network to automatically extract image features and perform terrain classification; and automatically carrying out terrain classification on the new delay Doppler image through the trained convolutional neural network. According to the method, the feature information of the SAR altimeter delay Doppler image is fully utilized, and accurate classification of the terrain is realized. Through comprehensive analysis of 13 terrain description factors, the terrain features are comprehensively described, and the classification reliability is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of synthetic aperture radar, and particularly relates to a terrain classification method and system based on SAR altimeter delay Doppler images. Background Art

[0002] SAR (Synthetic Aperture Radar), that is, synthetic aperture radar, is an active ground observation system that can be installed on flight platforms such as airplanes and satellites, and has the ability to observe the ground all day and all weather. The SAR altimeter can be used as an auxiliary system to provide high-precision terrain data and support for navigation and positioning.

[0003] Modern SAR altimeter systems mainly rely on altitude sequences to match with a preset digital elevation model (DEM) during positioning. However, this method has the following problems: on the one hand, in an environment with interference, simply relying on altitude sequence matching is prone to positioning errors; on the other hand, when there are deviations in the preset DEM, it will also affect the positioning accuracy. Therefore, how to improve the reliability of positioning in a signal-limited environment is an urgent problem to be solved in this field.

[0004] Terrain classification, as an important auxiliary means, can provide additional terrain information support for navigation and positioning. Existing research shows that accurate terrain type recognition can significantly improve the robustness of positioning. However, current terrain classification methods are mainly based on single feature descriptions or only use altitude information for classification, making it difficult to comprehensively describe terrain features and resulting in insufficient classification accuracy. In addition, traditional classification methods have problems with insufficient generalization ability when dealing with complex terrains.

[0005] Based on the above background, the present invention proposes a new terrain classification method based on SAR altimeter delay Doppler images. This method extracts multi-dimensional terrain description factors, combines the characteristics of the delay Doppler images of the SAR altimeter, and uses deep learning technology to achieve automatic terrain classification, providing a new solution for improving the reliability of navigation and positioning in a signal-limited environment. Summary of the Invention

[0006] The purpose of the present invention is to provide a terrain classification method and system based on SAR altimeter delay Doppler images to solve the problem of insufficient terrain classification accuracy in a GPS signal-limited environment.

[0007] According to the first aspect of the embodiments of the present invention, a terrain classification method based on SAR altimeter delay Doppler images is provided, including the following steps: Calculate terrain factors based on given DEM data, and perform terrain factor analysis to determine terrain categories; Establish a SAR altimeter echo model and generate delayed Doppler images through simulation based on given DEM data; The terrain categories obtained by terrain factor analysis are used as labels to annotate the corresponding delayed Doppler images, and the annotated delayed Doppler images are input into the convolutional neural network for training, so that the convolutional neural network can automatically extract image features and perform terrain classification. The new delayed Doppler images are automatically classified into terrain categories using the trained convolutional neural network.

[0008] Furthermore, the terrain factors include: elevation standard deviation, skewness coefficient, kurtosis coefficient, Fisher information content, slope, Hurst parameter of fractional Brownian motion process, roughness, fluctuation abundance, elevation entropy, correlation length, correlation index, elevation code distortion and elevation difference between points.

[0009] Furthermore, the calculation of terrain factors based on given DEM data includes the following sub-steps: Preprocess the DEM data and remove DEM blocks containing outliers; The preprocessed DEM data is calculated to obtain the terrain factors.

[0010] Furthermore, terrain factor analysis includes the following sub-steps: Conduct correlation analysis on terrain factors, calculate the correlation coefficient matrix, and eliminate highly correlated terrain factors with correlation coefficients greater than 0.95; Perform KMO test on the remaining terrain factors to obtain KMO values; Based on the comparison between the KMO value and the critical value, it is determined whether the remaining terrain factors are suitable for dimensionality reduction. If suitable, the remaining terrain factors are reduced in dimension through principal component analysis to obtain the dimensionality reduction results. The silhouette coefficient method is used to determine the optimal number of clusters and thus the terrain category.

[0011] Furthermore, the SAR altimeter echo model includes: a transmission signal model, an echo signal model and a compressed signal model.

[0012] Furthermore, the convolutional neural network contains: Input layer: receives delayed Doppler image data.

[0013] Multiple convolutional layers: Use learnable filters to extract spatial features.

[0014] ReLU activation layer: introduces nonlinear features.

[0015] Pooling layer: reduces feature dimension and extracts main features.

[0016] Fully connected layer: combines high-level features.

[0017] Softmax output layer: Output the probability distribution of each terrain category.

[0018] According to the second aspect of the embodiments of the present invention, there is provided a terrain classification system based on SAR altimeter delay-Doppler images, including: A terrain determination module, configured to calculate terrain factors based on given DEM data, perform terrain factor analysis, and determine terrain categories; A graph generation module, configured to establish an SAR altimeter echo model and generate delay-Doppler images through simulation based on given DEM data; A training module, configured to use the terrain categories obtained by terrain factor analysis as labels to annotate the corresponding delay-Doppler images, and input the annotated delay-Doppler images into a convolutional neural network for training, so that the convolutional neural network can automatically extract image features and perform terrain classification; An automatic classification module, configured to automatically perform terrain classification on new delay-Doppler images through the trained convolutional neural network.

[0019] According to the third aspect of the embodiments of the present invention, there is provided an electronic device, including a processor and a memory; Wherein, the memory is used to store one or more computer programs; when the one or more computer programs stored in the memory are executed by the processor, the electronic device can implement the terrain classification method based on SAR altimeter delay-Doppler images as described in the first aspect of the embodiments of the present invention.

[0020] According to the fourth aspect of the embodiments of the present invention, there is provided a computer storage medium. The computer-readable storage medium includes a computer program, and when the computer program runs on an electronic device, the electronic device is enabled to execute the terrain classification method based on SAR altimeter delay-Doppler images as described in the first aspect of the embodiments of the present invention.

[0021] A terrain classification method and system based on SAR altimeter delay-Doppler images according to the embodiments of the present invention have the following beneficial effects: 1. The present invention proposes a new terrain classification method, which makes full use of the characteristic information of SAR altimeter delay-Doppler images to achieve accurate terrain classification. Through comprehensive analysis of 13 terrain description factors, the terrain characteristics are comprehensively characterized, and the reliability of classification is improved.

[0022] 2. The present invention adopts a method combining feature dimensionality reduction and clustering analysis, effectively reducing data redundancy and improving the efficiency of feature extraction. Through KMO test and cumulative variance contribution rate analysis, the rationality of the dimensionality reduction process is ensured.

[0023] 3. The present invention automatically extracts features of the delay-Doppler image using a convolutional neural network, avoiding the limitations of manual feature design. Experimental results show that the accuracy of this method in the three-class terrain classification task reaches 90.67%, with a relatively high classification accuracy.

[0024] It is to be understood that both the foregoing general description and the following detailed description are exemplary and intended to provide further explanation of the claimed technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is a schematic diagram of the overall process of the terrain classification method of the present invention; Figure 2 is an example diagram of the high-resolution DEM dataset used in the present invention; Figure 3 is a correlation coefficient matrix diagram among 13 terrain description factors of the present invention; Figure 4 is a cumulative variance contribution rate curve diagram of the terrain factor dimensionality reduction analysis of the present invention; Figure 5 is an analysis result diagram of determining the optimal number of clusters using the silhouette coefficient method in the present invention; Figure 6 is an example diagram of the three-dimensional scatter clustering result of the data after dimensionality reduction in the present invention; Figure 7 is an example diagram of the delay-Doppler image corresponding to the first type of terrain in the present invention; Figure 8 is an example diagram of the delay-Doppler image corresponding to the second type of terrain in the present invention; Figure 9 is an example diagram of the delay-Doppler image corresponding to the third type of terrain in the present invention; Figure 10 is a schematic diagram of the convolutional neural network structure designed in the present invention; Figure 11 is a classification confusion matrix diagram of the CNN model of the present invention on the test set.

[0026] Figure 12 is a structural block diagram of a terrain classification system based on SAR altimeter delay-Doppler images according to an embodiment of the present invention.

[0027] Figure 13 is a structural block diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0028] The following will further describe the preferred embodiments of the present invention in detail with reference to the accompanying drawings.

[0029] First, in combination with Figures 1 to 11Describe a terrain classification method based on SAR altimeter delay-Doppler images according to an embodiment of the present invention, which is used to solve the problem of insufficient terrain classification accuracy in an environment with limited GPS signals and has a wide range of application scenarios.

[0030] As Figures 1 to 11 shown, a terrain classification method based on SAR altimeter delay-Doppler images according to an embodiment of the present invention has the following steps: Step S1: Calculate terrain factors based on given DEM data, perform terrain factor analysis, and determine the terrain category.

[0031] Among them, the terrain factors include: standard deviation of elevation, skewness coefficient, kurtosis coefficient, Fisher information content, slope, Hurst parameter of fractional Brownian motion process, roughness, fluctuation abundance, elevation entropy, correlation length, correlation index, elevation code distortion, and elevation difference between points. The specific calculation formulas are as follows: 1. Standard deviation of elevation (Esigma):

[0032] Where: is the elevation value of the i-th point, is the average elevation value.

[0033] 2. Skewness coefficient (CSkew):

[0034] Where: is the total number of samples.

[0035] 3. Kurtosis coefficient (CKurto):

[0036] 4. Fisher information content (FIC):

[0037] Where: and are the gradient operators in the x direction and y direction respectively.

[0038] 5. Slope (DSlope):

[0039] 6. Hurst parameter (FBMH):

[0040] 7. Roughness (DRough):

[0041] Where: M and N are the number of rows and columns of the data matrix, respectively.

[0042] 8. Fluctuation Abundance (DFAbund):

[0043] Where: is the roughness.

[0044] 9. Elevation Entropy (Entropy):

[0045] Where: is the probability of the th elevation interval, is the total number of elevation intervals.

[0046] 10. Correlation Length (SCLeng):

[0047] Where is the correlation coefficient at position and is the indicator function.

[0048] 11. Correlation Index (CIndex):

[0049] 12. Elevation Code Distortion (DCode):

[0050] Where: is the number of different quantized elevation values represents the quantized elevation value.

[0051] 13. Elevation Difference between Points (Hr):

[0052] Where: and represent the maximum and minimum elevations in the area.

[0053] And it should be noted that the selection of terrain factors is based on the following considerations: ① The elevation standard deviation (Esigma) reflects the overall undulation degree of the terrain; ② The skewness coefficient (CSkew) and kurtosis coefficient (CKurto) describe the asymmetry and peakedness of the elevation distribution; ③ The Fisher information content (FIC) and slope (DSlope) characterize the gradient features of the terrain; ④ The Hurst parameter (FBMH) and roughness (DRough) characterize the detailed features of the surface of the earth; ⑤ The fluctuation abundance (DFAbund) and elevation entropy (Entropy) represent the complexity of the terrain; ⑥ The correlation length (SCLeng) and correlation index (CIndex) reflect the spatial correlation of the terrain; ⑦ The elevation code distortion (DCode) and elevation difference between points (Hr) describe the discrete characteristics of elevation.

[0054] Furthermore, in this embodiment, the terrain factors calculated based on the given DEM data include the following sub-steps: Step S1.1: Preprocess the DEM data to remove the DEM blocks containing outliers, which can ensure the quality of the data.

[0055] Step S1.2: In this embodiment, calculate the terrain factors from the preprocessed DEM data.

[0056] Furthermore, the terrain factor analysis includes the following sub-steps: Step S1.3: Conduct a correlation analysis on the terrain factors, calculate the correlation coefficient matrix, and remove the highly correlated terrain factors with a correlation coefficient greater than 0.95; Step S1.4: Conduct a KMO test on the remaining terrain factors to obtain the KMO value; In this embodiment, the formula for conducting the KMO test is:

[0057] Where: is the value, is the correlation coefficient between variable and variable , is the partial correlation coefficient between variable and variable .

[0058] Step S1.5: Based on the comparison result of the KMO value and the critical value, determine whether the remaining terrain factors are suitable for dimensionality reduction processing. If suitable, conduct dimensionality reduction processing on the remaining terrain factors through principal component analysis to obtain the dimensionality reduction result; Step S1.6: Use the silhouette coefficient method to determine the optimal number of clusters, thereby determining the terrain categories.

[0059] In this embodiment, the formula for the optimal number of clusters is:

[0060] Among them is the sample and the average distance from other samples of the same type, is the sample and the average distance from the nearest other type of sample.

[0061] Step S2: Establish an SAR altimeter echo model and generate a delay-Doppler image through simulation based on the given DEM data.

[0062] Furthermore, in this embodiment, the SAR altimeter echo model includes: a reasonably designed transmitted signal model, an echo signal model established considering terrain scattering characteristics, and a compressed signal model.

[0063] Establish the transmitted signal model:

[0064] Among them: is the transmitted signal, is the baseband signal, is the carrier frequency, is the time delay.

[0065] Establish the received echo signal model:

[0066] Among them: is the received echo signal, is the transmitted power, is the antenna gain, is the wavelength, is the th scatterer area, is the th scatterer average scattering coefficient, is the th scatterer and radar distance, is the propagation loss, is the speed of light.

[0067] Obtain the compressed signal model:

[0068] Among them: is the compressed signal, is the pulse width, is the frequency modulation slope, .

[0069] Step S3: Using the terrain categories obtained from the terrain factor analysis as labels, label the corresponding delay-Doppler images, and input the labeled delay-Doppler images into a convolutional neural network for training, so that the convolutional neural network can automatically extract image features and perform terrain classification.

[0070] Step S4: Automatically perform terrain classification on new delay-Doppler images through the trained convolutional neural network.

[0071] Furthermore, in this embodiment, the network includes two parts: a feature extraction layer and a classification layer. The feature extraction layer consists of multiple convolutional blocks, and each convolutional block contains a convolutional layer, an activation function, and a pooling layer. The classification layer contains a fully connected layer, a Dropout layer, and a Softmax output layer.

[0072] It should be noted that the convolutional neural network has: 1. Multi-level feature extraction ability; 2. Adaptive parameter learning mechanism; 3. Network structure suitable for processing two-dimensional image data; 4. Reliable classification decision output.

[0073] The method of the present invention will be further described below through specific application embodiments and the accompanying drawings of the specification.

[0074] The data of the specific application embodiment of the present invention is from a dataset obtained by randomly sampling global DEM data, and a DEM example is Figure 2 as shown. The process of the specific application embodiment includes the following steps: Preprocess the DEM data. This step mainly performs quality inspection on the original DEM data and eliminates DEM blocks containing abnormal DSlope values.

[0075] Calculate terrain description factors. According to the calculation formulas of the foregoing 13 terrain description factors, extract features from each DEM data block.

[0076] First, perform a correlation analysis on the 13 terrain description factors, calculate the correlation coefficient matrix, as Figure 3 shown. The analysis results show that the correlation coefficients of the three factors, FIC, DCode, and Hr, with other factors exceed 0.95. To avoid information redundancy, these three highly correlated factors are eliminated.

[0077] Perform a KMO test on the remaining 10 factors, and obtain a KMO value of 0.77, which is greater than the critical value of 0.6, indicating that the data is suitable for dimensionality reduction processing; perform dimensionality reduction on the terrain description factors through principal component analysis. As Figure 4 shown, when the number of principal component factors is greater than or equal to 3, the cumulative variance contribution rate exceeds the threshold of 85%. Therefore, 3 principal component factors are selected as the dimensionality reduction result.

[0078] Use the silhouette coefficient method to determine the optimal number of clusters. AsFigure 5 As shown, the silhouette coefficient is calculated under different numbers of clusters. The results show that when the number of clusters is 3, the silhouette coefficient reaches the maximum value of 0.68. Therefore, the terrain is divided into 3 categories. The three-dimensional scatter plot distribution of the data after dimensionality reduction is as Figure 6 shown.

[0079] In this embodiment, the following parameter settings are adopted: the center frequency of the transmitted signal is 9.6 GHz, the pulse width is 20 μs, and the sampling frequency is 100 MHz. For three different types of terrain, the generated DDIs are respectively as Figure 7 , 8 , and as shown in Figure 9. It can be observed that the DDIs of different terrain types have obvious characteristic differences.

[0080] Convolutional neural network training. As the network structure shown in Figure 10 , it includes the following parts: the first convolutional block uses 32 3×3 convolutional kernels, the second convolutional block uses 64 3×3 convolutional kernels, and the fully connected layer contains 128 neurons. Using a batch size of 64 and a learning rate of 1×10 -4 for 20 rounds of training, finally achieving an accuracy of 90.67% in the three-category terrain classification task. The confusion matrix is as Figure 11 shown.

[0081] These results verify the effectiveness and reliability of the method of the present invention in the terrain classification task.

[0082] The above has described a terrain classification method based on SAR altimeter delay-Doppler images according to an embodiment of the present invention in conjunction with the attached Figures 1 to 11 drawings. Further, the present invention can also be applied to a terrain classification system based on SAR altimeter delay-Doppler images.

[0083] As Figure 12 shown, according to the second aspect of the embodiment of the present invention, there is provided a terrain classification system based on SAR altimeter delay-Doppler images, including: A terrain determination module 100, configured to calculate terrain factors based on given DEM data, perform terrain factor analysis, and determine terrain categories; A graph generation module 200, configured to establish an SAR altimeter echo model and generate delay-Doppler images through simulation based on given DEM data; A training module 300, configured to use the terrain categories obtained by terrain factor analysis as labels to label the corresponding delay-Doppler images, and input the labeled delay-Doppler images into a convolutional neural network for training, so that the convolutional neural network can automatically extract image features and perform terrain classification; An automatic classification module 400, configured to automatically perform terrain classification on new delay-Doppler images through the trained convolutional neural network.

[0084] The above in combination with the attached Figure 12 describes a terrain classification system based on SAR altimeter delay-Doppler images according to an embodiment of the present invention. Further, the present invention can also be applied to an electronic device.

[0085] As Figure 13 shown, according to the third aspect of the embodiment of the present invention, an electronic device is provided. The electronic device may include: one or more processors 1401, a memory 1402, and one or more computer programs 1403. The above-mentioned components can be connected through one or more communication buses. Among them, the one or more computer programs 1403 are stored in the memory 1402 and configured to be executed by the one or more processors 1401. The one or more computer programs 1403 include instructions, and the above instructions can cause the electronic device to execute the terrain classification method based on SAR altimeter delay-Doppler images according to the first aspect of the embodiment of the present invention.

[0086] The above in combination with the attached Figure 13 describes an electronic device according to an embodiment of the present invention. Further, the present invention can also be applied to a computer storage medium.

[0087] According to the fourth aspect of the embodiment of the present invention, a computer storage medium is provided. The computer-readable storage medium includes a computer program, and when the computer program runs on an electronic device, it causes the electronic device to execute the terrain classification method based on SAR altimeter delay-Doppler images as described in the first aspect of the embodiment of the present invention.

[0088] Through the description of the above embodiments, those skilled in the art can clearly understand that the embodiments of the present application can be implemented by hardware, or by firmware, or by a combination thereof. When implemented in software, the above functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium. The computer-readable medium includes computer storage media and communication media, where the communication media includes any medium that facilitates the transfer of a computer program from one place to another. The storage media can be any available medium that can be accessed by a computer. By way of example but not limitation: the computer-readable medium can include RAM, ROM, electrically erasable programmable read only memory (EEPROM), compact disc read-Only memory (CD-ROM), or other optical disc storage, magnetic disk storage media, or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer. In addition. Any connection can suitably be a computer-readable medium. For example, if the software is transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, wireless, and microwave are included in the definition of the medium. As used in the embodiments of the present application, disk and disc include compact disc (CD), laser disc, optical disc, digital video disc (DVD), floppy disk, and Blu-ray disc, where disks typically reproduce data magnetically, while discs reproduce data optically with a laser. The above combinations should also be included within the scope of protection of the computer-readable medium.

[0089] It should be noted that in this specification, the terms "include", "comprise", or any other variant thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or device that includes a series of elements includes not only those elements but also other elements that are not expressly listed, or also includes elements that are inherent to such process, method, article, or device. Without further limitation, an element defined by the statement "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or device that includes the element.

[0090] Although the content of the present invention has been described in detail through the above preferred embodiments, it should be recognized that the above description should not be construed as a limitation of the present invention. After those skilled in the art have read the above content, various modifications and alternatives to the present invention will be obvious. Therefore, the protection scope of the present invention should be defined by the appended claims.

Claims

1. A terrain classification method based on SAR altimeter delayed Doppler image, characterized in that: The following steps are included: Calculate terrain factors based on given DEM data, perform terrain factor analysis, and determine terrain categories; Establish a SAR altimeter echo model and generate delayed Doppler images through simulation based on given DEM data; The terrain categories obtained by terrain factor analysis are used as labels to annotate the corresponding delayed Doppler images, and the annotated delayed Doppler images are input into the convolutional neural network for training, so that the convolutional neural network can automatically extract image features and perform terrain classification. The new delayed Doppler images are automatically classified into terrain categories using the trained convolutional neural network.

2. The terrain classification method based on SAR altimeter delayed Doppler image as claimed in claim 1, characterized in that: The terrain factors include: elevation standard deviation, skewness coefficient, kurtosis coefficient, Fisher information content, slope, Hurst parameter of fractional Brownian motion process, roughness, fluctuation abundance, elevation entropy, correlation length, correlation index, elevation code distortion and elevation difference between points.

3. The terrain classification method based on SAR altimeter delayed Doppler image as claimed in claim 1 or 2, characterized in that: The method of calculating the terrain factor based on the given DEM data comprises the following sub-steps: Preprocess the DEM data and remove DEM blocks containing outliers; The preprocessed DEM data is calculated to obtain the terrain factors.

4. The terrain classification method based on SAR altimeter delayed Doppler image as claimed in claim 2, characterized in that: The terrain factor analysis comprises the following sub-steps: Conduct correlation analysis on terrain factors, calculate the correlation coefficient matrix, and eliminate highly correlated terrain factors with correlation coefficients greater than 0.95; Perform KMO test on the remaining terrain factors to obtain KMO values; Based on the comparison between the KMO value and the critical value, it is determined whether the remaining terrain factors are suitable for dimensionality reduction. If suitable, the remaining terrain factors are reduced in dimension through principal component analysis to obtain the dimensionality reduction results. The silhouette coefficient method is used to determine the optimal number of clusters and thus the terrain category.

5. The terrain classification method based on SAR altimeter delayed Doppler image as claimed in claim 2, characterized in that: The SAR altimeter echo model includes: a transmission signal model, an echo signal model and a compressed signal model.

6. The terrain classification method based on SAR altimeter delayed Doppler image as claimed in claim 1, characterized in that: The convolutional neural network includes: an input layer, multiple convolutional layers, a ReLU activation layer, a pooling layer, a fully connected layer and a Softmax output layer.

7. A terrain classification system based on SAR altimeter delayed Doppler images, characterized in that: Include: The terrain determination module is used to calculate the terrain factors based on the given DEM data, and to perform terrain factor analysis to determine the terrain category; A graphics generation module is used to establish a SAR altimeter echo model and generate a delayed Doppler image through simulation based on given DEM data; A training module is used to use the terrain categories obtained by terrain factor analysis as labels to annotate the corresponding delayed Doppler images, and input the annotated delayed Doppler images into the convolutional neural network for training, so that the convolutional neural network can automatically extract image features and perform terrain classification; The automatic classification module is used to automatically classify the terrain of new delayed Doppler images through the trained convolutional neural network.

8. An electronic device, characterized in that: including a processor and a memory; Wherein, the memory is used to store one or more computer programs; when the one or more computer programs stored in the memory are executed by the processor, the electronic device is able to implement the terrain classification method based on the SAR altimeter delayed Doppler image as described in any one of claims 1 to 6.

9. A computer storage medium, characterized in that The computer-readable storage medium includes a computer program, and when the computer program is executed on an electronic device, the electronic device executes the terrain classification method based on the SAR altimeter delayed Doppler image as claimed in any one of claims 1 to 6.

Images