Forest canopy height estimation method and system, terminal and storage medium

Abstract

The invention discloses a forest canopy height estimation method and system, a terminal and a storage medium, and the method comprises the steps: obtaining GEDI data of a forest, carrying out the preprocessing, removing an abnormal value, removing a non-forest region, and making a data set according to a non-forest sample and a forest sample; the method comprises the following steps: acquiring SAR image data to form an interference image, generating a coherent image, generating a multi-temporal coherence upper triangular matrix according to the coherent image, and converting the coherence upper triangular matrix corresponding to each sampling point into a coherence multi-temporal sequence; inputting the processed coherence multi-temporal sequence as a characteristic variable into an LSTM regression model, and training the LSTM regression model by using a data set; and obtaining image data of a forest to be identified, inputting the image data into the trained LSTM regression model, and outputting the forest canopy height of the forest to be identified. According to the method, the accuracy, reliability and efficiency of forest height estimation are improved, and forest monitoring is more efficient and wider.

Description

Technical Field

[0001] The present invention relates to the field of computer application technology, and in particular to a forest canopy height estimation method, system, terminal and computer-readable storage medium. Background Art

[0002] As a vital terrestrial ecosystem on Earth, forests play a decisive role in the global carbon cycle. As one of the forest structure parameters, forest canopy height is an important indicator for studying aboveground biomass, and is also a key indicator for studying forest productivity and biodiversity. With the development of remote sensing technology, especially the application of synthetic aperture radar (SAR) technology, forest structure can be monitored and evaluated more accurately. SAR technology has been widely used in forest structure research due to its advantages such as high resolution, ability to work day and night, and insensitivity to atmospheric conditions. In particular, interferometric synthetic aperture radar (InSAR) and polarimetric interferometric synthetic aperture radar (PolInSAR) are recognized as effective means of measuring forest canopy height due to their deep penetration in long-wave bands and cross-polarization modes. In addition, InSAR coherence shows great potential in land cover classification and vegetation classification, and can significantly improve classification performance. Changes in InSAR coherence are not only related to different land cover types, but also closely related to changes in the land surface, such as natural processes such as vegetation growth.

[0003] In the field of forest canopy height measurement, existing technologies mainly focus on the use of SAR data, especially data obtained by satellite systems such as Sentinel-1 and TanDEM-X. These technologies use time series of SAR images to analyze changes in forest structure, among which InSAR coherence is less used. In existing studies, SAR data, especially C-band SAR data, combined with ground reference data, have been used to predict forest height. Existing technologies have made some progress in studying forest canopy height using InSAR data and GEDI (Global Ecosystem Dynamics Investigation) measurement data, such as analyzing InSAR data through random forest regression models and deep learning networks, and combining Sentinel-1 and Sentinel-2 data to improve the accuracy of land cover classification and forest height estimation.

[0004] However, existing technologies still have some limitations. First, traditional InSAR-based methods usually rely on ground-based measured data for calibration and verification, which is not only time-consuming and labor-intensive, but may not be able to obtain comprehensive coverage due to limitations of ground conditions. Second, existing technologies often ignore the dynamic changes of InSAR coherence when processing SAR data, which limits their application in capturing seasonal changes in forest structure. In addition, when combining multi-source remote sensing data for forest structure inversion, existing methods may have problems with data fusion and consistency, affecting the final estimation accuracy. These limitations indicate that existing technologies may have failed to fully utilize the potential of deep learning models in InSAR multi-temporal coherence data analysis, especially in the continuous monitoring and prediction of forest canopy height. More advanced methods are needed to improve the accuracy and efficiency of forest height estimation.

[0005] Therefore, the prior art still needs to be improved and developed. Summary of the invention

[0006] The main purpose of the present invention is to provide a forest canopy height estimation method, system, terminal and computer-readable storage medium, aiming to solve the problem that the prior art fails to fully utilize the potential of deep learning models in InSAR multi-temporal coherence data analysis and cannot improve the accuracy and efficiency of forest height estimation.

[0007] To achieve the above object, the present invention provides a method for estimating the height of a forest canopy, the method comprising the following steps:

[0008] Acquire GEDI data of forests, preprocess the GEDI data to remove outliers and exclude non-forest areas, and create a data set based on non-forest samples and forest samples;

[0009] Acquire SAR image data, form an interference image according to the SAR image data, generate a coherence image according to the interference image, generate a multi-phase coherence upper triangular matrix according to the coherence image, and convert the coherence upper triangular matrix corresponding to each sampling point into a coherence multi-phase sequence;

[0010] Inputting the processed coherent multi-phase sequence as a feature variable into the LSTM regression model, and training the LSTM regression model using the data set to obtain a trained LSTM regression model;

[0011] Obtain image data of the forest to be identified, input the image data of the forest to be identified into the trained LSTM regression model, and output the forest canopy height of the forest to be identified.

[0012] Optionally, the forest canopy height estimation method, wherein the GEDI data of the forest is obtained, the GEDI data is preprocessed to remove outliers and exclude non-forest areas, and a data set is prepared based on non-forest samples and forest samples, specifically comprising:

[0013] GEDIL2A elevation data with the same time range as the coherence data were obtained through NASA's Global Ecosystem Dynamics Survey project;

[0014] Distinguishing land cover types through a global land cover map, removing outliers in the GEDIL2A elevation data, and obtaining intermediate GEDIL2A elevation data;

[0015] Non-forest samples and forest samples in the intermediate GEDIL2A elevation data were selected and used as data sets for the classification model.

[0016] Optionally, the method for estimating forest canopy height, wherein the GEDI data of the forest is obtained, the GEDI data is preprocessed to remove outliers and exclude non-forest areas, and a data set is prepared based on non-forest samples and forest samples, and then further comprises:

[0017] Acquire a multispectral image of a Sentinel-2 satellite, and calculate a vegetation index based on the multispectral image;

[0018] The vegetation index is used as a feature vector, and a classification model is constructed based on a support vector machine algorithm, wherein the classification model is used to filter out non-forest areas and only retain GEDI data points that are all forests within the detection range;

[0019] A relative height index is extracted based on the GEDI echo, and the relative height index is used as the actual forest canopy height.

[0020] Optionally, in the forest canopy height estimation method, the vegetation index includes: enhanced vegetation index, normalized difference vegetation index, red edge vegetation index and difference vegetation index.

[0021] Optionally, the forest canopy height estimation method, wherein the acquiring of SAR image data, forming an interference image according to the SAR image data, generating a coherent image according to the interference image, generating a multi-phase coherence upper triangular matrix according to the coherence image, and converting the coherence upper triangular matrix corresponding to each sampling point into a coherence multi-phase sequence, specifically comprises:

[0022] Acquire Sentinel-1ASAR image data with a revisit time of a preset time, perform interference on the Sentinel-1ASAR image data in pairs to form an interference image, and generate a corresponding coherent image using a multi-window method;

[0023] The calculation of interference coherence is:

[0024]

[0025] Where γ represents the interference coherence, ranging from 0 to 1, S1 and S2 represent two co-polarized SAR complex images, E{·} represents the mathematical expectation, * represents the complex conjugate operator, |S1| represents the amplitude of S1, and |S2| represents the amplitude of S2;

[0026] From the set of coherent image points, select the coherent points with GEDI footprints within the resolution range;

[0027] After screening, each sampling point is arranged in the order of the interference interval time to generate a multi-temporal coherence upper triangular matrix;

[0028] If there are multiple GEDI footprints, the tree height values ​​of the multiple GEDI footprints are weighted averaged according to the distance to determine the GEDI tree height value of the coherence point where the GEDI footprint exists;

[0029] Among them, each sampling point has a multi-temporal coherence upper triangular matrix and a corresponding GEDI tree height value;

[0030] In the upper triangular matrix of coherence, the values ​​with the same interferometric time interval are distributed diagonally, and for the vegetation area, the relatively high coherence is maintained close to the main diagonal;

[0031] The upper triangular part of the coherence upper triangular matrix is ​​extracted along the diagonal line, and the upper triangular part is converted into coherence sequences of unequal lengths and different time spans. The coherence upper triangular matrix corresponding to each sampling point is converted into a coherence multi-phase sequence.

[0032] Optionally, the forest canopy height estimation method, wherein the processed coherence multi-phase sequence is input into an LSTM regression model as a feature variable, and the LSTM regression model is trained using the data set to obtain a trained LSTM regression model, specifically includes:

[0033] The processed coherent multi-phase sequence is input into the LSTM regression model as a feature variable, and the prediction results of the LSTM regression model are tested using the preprocessed GEDI sampling data;

[0034] The data set is divided into a training set, a validation set and a test set in proportion, the training set is used to train the LSTM regression model so that the LSTM regression model learns the mapping relationship from the coherence multi-phase sequence to the canopy height, the validation set is used to adjust the parameters and hyperparameters of the LSTM regression model during the training process to avoid overfitting, and the test set is independent of the training process and is used to ultimately evaluate the generalization ability of the LSTM regression model;

[0035] After the LSTM regression model is trained using the training set and the validation set, the LSTM regression model is tested using the test set, and a trained LSTM regression model is obtained when the test requirements are met.

[0036] Optionally, in the forest canopy height estimation method, the LSTM regression model is a two-layer LSTM regression model, each layer contains 256 units, and uses ReLU as an activation function, and a Dropout layer is added;

[0037] The loss function for the LSTM regression model training is the mean square error, which is calculated as:

[0038]

[0039] Among them, MSE represents the mean square error, y i represents the actual canopy height of the ith sample, It indicates the canopy height of the i-th sample predicted by the model, and N indicates the number of samples;

[0040] The evaluation indicators R square and root mean square error RMSE are used as the judgment criteria for the final model to predict the accuracy of canopy height, which is calculated as:

[0041]

[0042] Among them, R 2 represents the evaluation index R square sum, RMSE represents the root mean square error, Represents the average value of canopy height.

[0043] In addition, to achieve the above-mentioned purpose, the present invention also provides a forest canopy height estimation system, wherein the forest canopy height estimation system comprises:

[0044] A data processing module is used to obtain GEDI data of forests, pre-process the GEDI data, remove outliers and exclude non-forest areas, and make a data set based on non-forest samples and forest samples;

[0045] A sequence generation module is used to acquire SAR image data, form an interference image according to the SAR image data, generate a coherence image according to the interference image, generate a multi-phase coherence upper triangular matrix according to the coherence image, and convert the coherence upper triangular matrix corresponding to each sampling point into a coherence multi-phase sequence;

[0046] A model training module, used for inputting the processed coherent multi-phase sequence as a feature variable into the LSTM regression model, and training the LSTM regression model using the data set to obtain a trained LSTM regression model;

[0047] The height estimation module is used to obtain the image data of the forest to be identified, input the image data of the forest to be identified into the trained LSTM regression model, and output the forest canopy height of the forest to be identified.

[0048] In addition, to achieve the above-mentioned purpose, the present invention also provides a terminal, wherein the terminal includes: a memory, a processor, and a forest canopy height estimation program stored in the memory and executable on the processor, and when the forest canopy height estimation program is executed by the processor, the steps of the forest canopy height estimation method as described above are implemented.

[0049] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a forest canopy height estimation program, and when the forest canopy height estimation program is executed by a processor, the steps of the forest canopy height estimation method as described above are implemented.

[0050] In the present invention, GEDI data of the forest is obtained, the GEDI data is preprocessed, outliers are removed and non-forest areas are eliminated, and a data set is prepared according to non-forest samples and forest samples; SAR image data is obtained, an interference image is formed according to the SAR image data, a coherent image is generated according to the interference image, a multi-phase coherence upper triangular matrix is ​​generated according to the coherent image, and the coherence upper triangular matrix corresponding to each sampling point is converted into a coherence multi-phase sequence; the processed coherence multi-phase sequence is input into the LSTM regression model as a feature variable, and the LSTM regression model is trained using the data set to obtain a trained LSTM regression model; image data of the forest to be identified is obtained, the image data of the forest to be identified is input into the trained LSTM regression model, and the forest canopy height of the forest to be identified is output. The present invention can better capture the dynamic changes of forest structure by combining the changes in InSAR coherence and the accurate measurement of GEDI, improve the accuracy, reliability and efficiency of forest height estimation, reduce the dependence on ground measured data, reduce the cost and difficulty of data collection, and make forest monitoring more efficient and extensive. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 is a flow chart of a preferred embodiment of the forest canopy height estimation method of the present invention;

[0052] Figure 2 It is a schematic diagram of a forest canopy height estimation process based on InSAR time series coherence and LSTM in a preferred embodiment of the forest canopy height estimation method of the present invention;

[0053] Figure 3 It is a structural diagram of a preferred embodiment of the forest canopy height estimation system of the present invention;

[0054] Figure 4 It is a structural diagram of a preferred embodiment of the terminal of the present invention. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the present invention clearer and more specific, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0056] Aiming at the limitations of the existing technology in estimating forest canopy height, the present invention proposes a new method, that is, using InSAR coherence or its changes and the forest canopy height values ​​measured by NASA Global Ecosystem Dynamics Investigation (GEDI) to train a deep learning model Long Short-Term Memory (LSTM) to estimate the forest canopy height. The LiDAR instrument carried by GEDI has provided accurate forest canopy height data since 2018, which can be used as an alternative source of ground truth data. The existing technology usually relies on ground-based measured data for calibration, which has problems of low efficiency and high cost, and may not reflect the dynamic changes of forest structure in a timely manner. In addition, the existing technology also faces challenges in processing SAR data with irregular time intervals. By combining the changes in InSAR coherence and the precise measurements of GEDI, the dynamic changes of forest structure can be better captured, and the accuracy and reliability of estimation can be improved; secondly, this method can reduce the dependence on ground-based measured data, reduce the cost and difficulty of data collection, and make forest monitoring more efficient and extensive.

[0057] The forest canopy height estimation method described in the preferred embodiment of the present invention is as follows: Figure 1 and Figure 2 As shown, the forest canopy height estimation method comprises the following steps:

[0058] Step S10, obtaining GEDI data of forests, preprocessing the GEDI data, removing outliers and excluding non-forest areas, and making a data set based on non-forest samples and forest samples.

[0059] Specifically, the first is the GEDI data acquisition and screening strategy. Through the NASA Global Ecosystem Dynamics Survey Project, GEDIL2A elevation data (i.e., space-borne lidar altimetry data GEDI L2A level data) with the same time range as the coherence data is obtained; GEDI data is divided into different levels, from the most original data to finished data that has been processed and analyzed more. Among them, the L1A level data contains the original GEDI lidar waveform data, which has not been geolocated or corrected. The L1B level data is waveform data containing geolocation, which is the result of preliminary processing of the L1A data and provides the geographic coordinate information of the waveform. The L2A level data required in the method of the present invention is generated by analyzing the peaks and distributions in the L1B waveform data, and contains altitude and height indicators derived from the geolocated waveform. These data provide ground elevation and relative height indicators related to the crown structure. These data provide more detailed surface and vegetation structure information, which are suitable for applications such as vegetation height and biomass estimation. GEDI contains three lasers and a full waveform recording lidar instrument working in the near-infrared region. Individual lasers are fired at 60-meter intervals along the track and have a circular detection area of ​​approximately 25 meters. Therefore, each data point represents a GEDI waveform return metric for every 25-meter diameter. Since GEDI data provides more accurate vegetation information, it is used as a data source to verify the prediction accuracy. For each data point, the relative height metric RH100 based on the L2A processing algorithm is extracted, which is the 100th percentile of the energy return height relative to the ground, representing the height difference from the highest point reflection (usually the top of the canopy) to the lowest mode detected reflection (usually the ground). This value can be regarded as an approximation of the canopy height. However, GEDI measurement data in urban, bare land and water areas are unreliable, and non-forested areas often exhibit high coherence, resulting in prediction errors. Therefore, it is necessary to preprocess the GEDI data to remove these outliers.

[0060] The land cover types are distinguished by using a global land cover map (e.g., the Global Land Cover (GLC) map provided by the European Space Agency), and the outliers in the GEDIL2A elevation data are removed, that is, most of the outliers are removed to obtain the intermediate GEDIL2A elevation data; non-forest samples and forest samples in the intermediate GEDIL2A elevation data are selected and used as the data set for the classification model. That is, since the GLC image cannot completely exclude all non-forest areas in the detection area of ​​the GEDI sampling point, a large number of non-forest surfaces still remain in the screened area, which directly affects the purity of the data and the accuracy of subsequent analysis. Therefore, it is necessary to further eliminate non-forest areas from the remaining GEDI point set. To achieve this, non-forest and forest samples can be manually selected as the data set for the classification model.

[0061] The Sentinel-2 satellite has 13 spectral bands, among which four bands (B2, B3, B4 and B8) with a spatial resolution of 10 meters correspond to different vegetation reflectance characteristics. Using the reflectance values ​​of these bands, some commonly used vegetation indices can be calculated, such as the Enhanced Vegetation Index (EVI), the Normalized Difference Vegetation Index (NDVI), the Red Edge Vegetation Index (RVI) and the Difference Vegetation Index (DVI). The present invention uses the B2, B3, B4 and B8 bands of Sentinel-2 and the NDVI, EVI, RVI and DVI calculated according to the four bands as feature variables, and constructs a classification model based on the Support Vector Machine (SVM) algorithm. This classification model is used to filter out non-forest areas and only retain GEDI data points that are all forests within the detection range (the GLC map can only do a coarse screening based on the longitude and latitude of the data points, and cannot remove non-forest areas within the detection range of the data points. Therefore, after the screening of the GLC map, SVM is used here to do a fine screening. The combination of the two ensures that the data points are all forest areas within the detection range). In addition, considering the possibility of abnormal forest canopy heights in the preprocessed data set, the GEDI data is further screened. For example, GEDI cannot accurately predict the height of sparse vegetation below 3 meters. Therefore, the forest canopy height of GEDI samples with RH100 < 3 meters is assigned a value of 0.

[0062] Based on the GEDI echo (GEDI echo refers to the signal received by the GEDI lidar sensor. When the GEDI laser pulse is emitted from the International Space Station and hits the earth's surface, part of the light will be reflected back. These reflected signals carry information about the characteristics of the earth's surface, including the vertical structure of vegetation and the terrain of the ground), a relative height index is extracted and used as the actual forest canopy height. The relative height index is extracted based on the GEDI echo, mainly by analyzing the reflection pattern in the lidar waveform, determining the highest and lowest reflection points in the waveform, and then calculating the height difference between the vegetation canopy and the ground.

[0063] Step S20, acquiring SAR image data, forming an interference image according to the SAR image data, generating a coherent image according to the interference image, generating a multi-phase coherence upper triangular matrix according to the coherence image, and converting the coherence upper triangular matrix corresponding to each sampling point into a coherence multi-phase sequence.

[0064] Specifically, multi-view coherent image generation and processing uses Sentinel-1A data obtained from the European Space Agency's Copernicus Global Earth Observation Project as one of the data sources, with a revisit period of 12 days. Sentinel-1A single-look complex (SLC) image uses cross-polarization (Vertical polarization Horizontalpolarization, VH) and co-polarization (Vertical polarization, VV) polarization modes, and an interferometric wide swath (IW) mode with a ground resolution of 5 meters × 20 meters. The co-polarization VV data of Sentinel-1A is selected for calculation. In order to obtain coherence data, Sentinel-1ASAR image data with a revisit time of 12 days are obtained, and the Sentinel-1ASAR image data are paired and interfered to form an interference image, and a multi-window method of 20×5 pixels is used to generate the corresponding coherent image.

[0065] Among them, the interferometric coherence is the normalized mutual correlation coefficient between two registered SAR images, ranging from 0 to 1, and the calculation formula is as follows:

[0066]

[0067] Where γ represents the interference coherence, which ranges from 0 to 1, S1 and S2 represent two registered co-polarization SAR complex images, E{·} represents the mathematical expectation, * represents the complex conjugate operator, |S1| represents the amplitude of S1, and |S2| represents the amplitude of S2.

[0068] From the coherent image point set, select the coherence points with GEDI footprints within the resolution range; after screening, arrange each sampling point in the order of interference interval time to generate a multi-temporal coherence upper triangular matrix; if there are multiple GEDI footprints, weighted average the tree height values ​​of multiple GEDI footprints according to the distance to determine the GEDI tree height value of the coherence point with GEDI footprints; therefore, after the above screening strategy, each sampling point has a multi-temporal coherence upper triangular matrix and a corresponding GEDI tree height value.

[0069] Considering that in the coherence upper triangular matrix, the values ​​with the same interference time interval are distributed diagonally, and for the vegetation area, the relatively high coherence is maintained close to the main diagonal; therefore, the upper triangular part of the coherence upper triangular matrix is ​​extracted along the diagonal, and the upper triangular part is converted into coherence sequences of unequal lengths and different time spans, thereby converting the coherence upper triangular matrix corresponding to each sampling point into a coherence multi-phase sequence.

[0070] In order to make the sequence length consistent, these features are padded with equal length and time encoding is cleverly added to make full use of this information in the analysis. At the same time, a masking layer is used to mark the padded values ​​as masks to ensure that the model ignores these padded values ​​during calculation. In this way, each sequence not only maintains its original temporal dynamic characteristics, but can also be uniformly input into the model for processing.

[0071] Step S30: input the processed coherent multi-phase sequence as a feature variable into the LSTM regression model, and use the data set to train the LSTM regression model to obtain a trained LSTM regression model.

[0072] Specifically, the LSTM algorithm was used to train and test the forest canopy height estimation model (i.e. Figure 2The LSTM regression model in the LSTM, LSTM, Long Short-Term Memory, long short-term memory network), first, the processed coherent multi-phase sequence is input into the LSTM regression model as a feature variable, and the prediction results of the LSTM regression model are tested using the preprocessed GEDI sampling data. Secondly, the data set is divided into a training set (70%), a validation set (15%) and a test set (15%) in proportion. The training set is used to train the LSTM regression model so that the LSTM regression model learns the mapping relationship from the coherent multi-phase sequence to the canopy height. The validation set is used to adjust the parameters and hyperparameters of the LSTM regression model during the training process to avoid overfitting. The test set is independent of the training process and is used to finally evaluate the generalization ability of the LSTM regression model. After the LSTM regression model is trained using the training set and the validation set, the LSTM regression model is tested using the test set, and after meeting the test requirements, a trained LSTM regression model is obtained.

[0073] The LSTM regression model is a two-layer LSTM regression model, each layer contains 256 units, and uses ReLU as the activation function. At the same time, a Dropout layer is added to prevent overfitting, and the output of a part of neurons is randomly discarded to enhance the generalization ability of the model. Finally, the model is trained and tested. The optimizer in the training process selects Adam, which automatically adjusts the learning rate to help the model converge faster. The early stopping strategy is used to monitor the loss changes of the validation set. When the loss on the validation set no longer improves, the training is stopped in advance to prevent overfitting.

[0074] The loss function for the LSTM regression model training is the mean squared error (MSE), which is calculated as:

[0075]

[0076] Among them, MSE represents the mean square error, y i represents the actual canopy height of the ith sample, It indicates the canopy height of the i-th sample predicted by the model, and N indicates the number of samples.

[0077] The input shape of the model is determined by the time step and number of features of the training data. In addition, the evaluation indicators R square and root mean square error RMSE (Root Mean Square Error, RMSE) are used as the judgment criteria for the final model to predict the accuracy of canopy height, calculated as:

[0078]

[0079] Among them, R 2represents the evaluation index R square sum, RMSE represents the root mean square error, Represents the average value of canopy height.

[0080] Step S40: Obtain image data of the forest to be identified, input the image data of the forest to be identified into the trained LSTM regression model, and output the forest canopy height of the forest to be identified.

[0081] Finally, after the model training is completed, a trained LSTM regression model is obtained, which can be directly used to directly estimate the forest canopy height of the forest to be identified, thereby improving the accuracy and reliability of forest canopy height estimation.

[0082] Beneficial effects of the present invention:

[0083] (1) A detailed GEDI data acquisition and preprocessing strategy was proposed. Using NASA's GEDI L2A data, the relative height index RH100 was extracted to obtain accurate forest canopy height. At the same time, the GLC map and SVM algorithm were introduced in combination with the data extracted by Sentinel-2 to assist in screening forest areas and filter out interference data from non-forest surfaces to ensure the purity of GEDI data.

[0084] (2) By analyzing the multi-temporal coherence of Sentinel-1 ASAR data, including the extraction of the upper triangular part of the coherence matrix and the addition of time encoding, as well as equal-length front-filling processing for unequal-length sequences, these feature engineering steps enhance the model's ability to capture temporal dynamics, which helps to capture the dynamic changes of forest structure.

[0085] (3) Using a two-layer LSTM regression model, the forest canopy height estimation model is trained, the processed coherence multi-phase series is used as input features, and the GEDI sampling data is combined for prediction. The early stopping strategy and Dropout layer are used to prevent overfitting and improve the generalization ability of the model.

[0086] Furthermore, the Sentinel-1 InSAR data used in the present invention may consider using other types of InSAR data, such as TanDEM-X, ALOS-2 PALSAR-2 or future SAR satellite systems. These data may have different bands, resolutions or revisit periods, which may affect the performance of the model; the LSTM model can be replaced by other types of deep learning models, such as Gated Recurrent Unit (GRU), Transformer or Convolutional LSTM, which may have different advantages in processing time series data; after model prediction, spatial smoothing or filtering techniques can be used to reduce prediction noise, or merged with other remote sensing data (such as optical images) to further improve the estimation accuracy of forest canopy height; at the same time, the method of the present invention can also be applied to a variety of scenarios, such as surface biomass estimation, changes in land cover types, etc.; in addition, you can also try to use transfer learning, that is, first pre-train the model on a large-scale data set, and then fine-tune it on data in a specific area to improve the adaptability of the model in the new area.

[0087] Furthermore, if Figure 3 As shown, based on the above-mentioned forest canopy height estimation method, the present invention also provides a forest canopy height estimation system accordingly, wherein the forest canopy height estimation system comprises:

[0088] A data processing module 51 is used to obtain GEDI data of forests, pre-process the GEDI data, remove outliers and exclude non-forest areas, and create a data set based on non-forest samples and forest samples;

[0089] A sequence generation module 52 is used to acquire SAR image data, form an interference image according to the SAR image data, generate a coherence image according to the interference image, generate a multi-phase coherence upper triangular matrix according to the coherence image, and convert the coherence upper triangular matrix corresponding to each sampling point into a coherence multi-phase sequence;

[0090] A model training module 53, used for inputting the processed coherence multi-phase sequence as a feature variable into the LSTM regression model, and training the LSTM regression model using the data set to obtain a trained LSTM regression model;

[0091] The height estimation module 54 is used to obtain image data of the forest to be identified, input the image data of the forest to be identified into the trained LSTM regression model, and output the forest canopy height of the forest to be identified.

[0092] Furthermore, if Figure 4As shown, based on the above-mentioned forest canopy height estimation method and system, the present invention also provides a terminal accordingly, and the terminal includes a processor 10, a memory 20 and a display 30. Figure 4 Only some components of the terminal are shown, but it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0093] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or memory of the terminal. In other embodiments, the memory 20 may also be an external storage device of the terminal, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the terminal. Furthermore, the memory 20 may also include both an internal storage unit of the terminal and an external storage device. The memory 20 is used to store application software and various types of data installed on the terminal, such as the program code of the installation terminal. The memory 20 may also be used to temporarily store data that has been output or is to be output. In one embodiment, a forest canopy height estimation program 40 is stored on the memory 20, and the forest canopy height estimation program 40 can be executed by the processor 10, thereby realizing the forest canopy height estimation method in the present application.

[0094] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor or other data processing chip, used to run the program code or process data stored in the memory 20, such as executing the forest canopy height estimation method.

[0095] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, etc. The display 30 is used to display information on the terminal and to display a visual user interface. The components 10-30 of the terminal communicate with each other via a system bus.

[0096] In one embodiment, when the processor 10 executes the forest canopy height estimation program 40 in the memory 20 , the steps of the forest canopy height estimation method are implemented.

[0097] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a forest canopy height estimation program, and when the forest canopy height estimation program is executed by a processor, the steps of the forest canopy height estimation method as described above are implemented.

[0098] In summary, the present invention provides a forest canopy height estimation method, system, terminal and storage medium, the method comprising: obtaining GEDI data of a forest, preprocessing the GEDI data, removing outliers and excluding non-forest areas, and preparing a data set according to non-forest samples and forest samples; obtaining SAR image data, forming an interference image according to the SAR image data, generating a coherent image according to the interference image, generating a multi-phase coherence upper triangular matrix according to the coherent image, and converting the coherence upper triangular matrix corresponding to each sampling point into a coherence multi-phase sequence; inputting the processed coherence multi-phase sequence as a feature variable into an LSTM regression model, training the LSTM regression model using the data set, and obtaining a trained LSTM regression model; obtaining image data of a forest to be identified, inputting the image data of the forest to be identified into the trained LSTM regression model, and outputting the forest canopy height of the forest to be identified. By combining the changes in InSAR coherence and the precise measurement of GEDI, the present invention can better capture the dynamic changes of forest structure, improve the accuracy, reliability and efficiency of forest height estimation, reduce the dependence on ground measured data, reduce the cost and difficulty of data collection, and make forest monitoring more efficient and extensive.

[0099] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the existence of other identical elements in the process, method, article or terminal including the element.

[0100] Of course, those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing related hardware (such as a processor, a controller, etc.) through a computer program, and the program can be stored in a computer-readable storage medium that can be read by a computer, and the program can include the processes of the above-mentioned method embodiments when executed. The computer-readable storage medium can be a memory, a disk, an optical disk, etc.

[0101] It should be understood that the application of the present invention is not limited to the above examples. For ordinary technicians in this field, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A method for estimating forest canopy height, characterized in that: The forest canopy height estimation method comprises: Acquire GEDI data of forests, preprocess the GEDI data to remove outliers and exclude non-forest areas, and create a data set based on non-forest samples and forest samples; Acquire SAR image data, form an interference image according to the SAR image data, generate a coherence image according to the interference image, generate a multi-phase coherence upper triangular matrix according to the coherence image, and convert the coherence upper triangular matrix corresponding to each sampling point into a coherence multi-phase sequence; Inputting the processed coherent multi-phase sequence as a feature variable into the LSTM regression model, and training the LSTM regression model using the data set to obtain a trained LSTM regression model; Obtain image data of the forest to be identified, input the image data of the forest to be identified into the trained LSTM regression model, and output the forest canopy height of the forest to be identified.

2. The method for estimating forest canopy height according to claim 1, characterized in that: The obtaining of forest GEDI data, preprocessing of the GEDI data, removing outliers and excluding non-forest areas, and making a data set based on non-forest samples and forest samples specifically includes: GEDIL2A elevation data with the same time range as the coherence data were obtained through NASA's Global Ecosystem Dynamics Survey project; Distinguishing land cover types through a global land cover map, removing outliers in the GEDIL2A elevation data, and obtaining intermediate GEDIL2A elevation data; Non-forest samples and forest samples in the intermediate GEDIL2A elevation data were selected and used as data sets for the classification model.

3. The method for estimating forest canopy height according to claim 2, characterized in that: The method further comprises: obtaining GEDI data of forests, preprocessing the GEDI data, removing outliers and excluding non-forest areas, and preparing a data set based on non-forest samples and forest samples, and then further comprising: Acquire a multispectral image of a Sentinel-2 satellite, and calculate a vegetation index based on the multispectral image; The vegetation index is used as a feature vector, and a classification model is constructed based on a support vector machine algorithm, wherein the classification model is used to filter out non-forest areas and only retain GEDI data points that are all forests within the detection range; A relative height index is extracted based on the GEDI echo, and the relative height index is used as the actual forest canopy height.

4. The method for estimating forest canopy height according to claim 3, characterized in that: The vegetation index includes: enhanced vegetation index, normalized vegetation index, red edge vegetation index and difference vegetation index.

5. The method for estimating forest canopy height according to claim 3, characterized in that: The acquiring of SAR image data, forming an interference image according to the SAR image data, generating a coherent image according to the interference image, generating a multi-phase coherence upper triangular matrix according to the coherence image, and converting the coherence upper triangular matrix corresponding to each sampling point into a coherence multi-phase sequence specifically includes: Acquire Sentinel-1ASAR image data with a revisit time of a preset time, perform interference on the Sentinel-1ASAR image data in pairs to form an interference image, and generate a corresponding coherent image using a multi-window method; The calculation of interference coherence is: Where γ represents the interference coherence, ranging from 0 to 1, S1 and S2 represent two co-polarized SAR complex images, E{·} represents the mathematical expectation, * represents the complex conjugate operator, |S1| represents the amplitude of S1, and |S2| represents the amplitude of S2; From the set of coherent image points, select the coherent points with GEDI footprints within the resolution range; After screening, each sampling point is arranged in the order of the interference interval time to generate a multi-temporal coherence upper triangular matrix; If there are multiple GEDI footprints, the tree height values ​​of the multiple GEDI footprints are weighted averaged according to the distance to determine the GEDI tree height value of the coherence point where the GEDI footprint exists; Among them, each sampling point has a multi-temporal coherence upper triangular matrix and a corresponding GEDI tree height value; In the upper triangular matrix of coherence, the values ​​with the same interferometric time interval are distributed diagonally, and for the vegetation area, the relatively high coherence is maintained close to the main diagonal; The upper triangular part of the coherence upper triangular matrix is ​​extracted along the diagonal line, and the upper triangular part is converted into coherence sequences of unequal lengths and different time spans. The coherence upper triangular matrix corresponding to each sampling point is converted into a coherence multi-phase sequence.

6. The method for estimating forest canopy height according to claim 5, characterized in that: The processed coherent multi-phase sequence is input as a feature variable into the LSTM regression model, and the LSTM regression model is trained using the data set to obtain a trained LSTM regression model, specifically including: The processed coherent multi-phase sequence is input into the LSTM regression model as a feature variable, and the prediction results of the LSTM regression model are tested using the preprocessed GEDI sampling data; The data set is divided into a training set, a validation set and a test set in proportion, the training set is used to train the LSTM regression model so that the LSTM regression model learns the mapping relationship from the coherence multi-phase sequence to the canopy height, the validation set is used to adjust the parameters and hyperparameters of the LSTM regression model during the training process to avoid overfitting, and the test set is independent of the training process and is used to ultimately evaluate the generalization ability of the LSTM regression model; After the LSTM regression model is trained using the training set and the validation set, the LSTM regression model is tested using the test set, and a trained LSTM regression model is obtained when the test requirements are met.

7. The method for estimating forest canopy height according to claim 6, characterized in that: The LSTM regression model is a double-layer LSTM regression model, each layer contains 256 units, uses ReLU as the activation function, and adds a Dropout layer; The loss function for the LSTM regression model training is the mean square error, which is calculated as: Among them, MSE represents the mean square error, y i represents the actual canopy height of the ith sample, It indicates the canopy height of the i-th sample predicted by the model, and N indicates the number of samples; The evaluation indicators R square and root mean square error RMSE are used as the judgment criteria for the final model to predict the accuracy of canopy height, which is calculated as: Among them, R 2 represents the evaluation index R square sum, RMSE represents the root mean square error, Represents the average value of canopy height.

8. A forest canopy height estimation system, characterized in that: The forest canopy height estimation system comprises: A data processing module is used to obtain GEDI data of forests, pre-process the GEDI data, remove outliers and exclude non-forest areas, and make a data set based on non-forest samples and forest samples; A sequence generation module is used to acquire SAR image data, form an interference image according to the SAR image data, generate a coherence image according to the interference image, generate a multi-phase coherence upper triangular matrix according to the coherence image, and convert the coherence upper triangular matrix corresponding to each sampling point into a coherence multi-phase sequence; A model training module, used for inputting the processed coherent multi-phase sequence as a feature variable into the LSTM regression model, and training the LSTM regression model using the data set to obtain a trained LSTM regression model; The height estimation module is used to obtain the image data of the forest to be identified, input the image data of the forest to be identified into the trained LSTM regression model, and output the forest canopy height of the forest to be identified.

9. A terminal, characterized in that: The terminal includes: a memory, a processor, and a forest canopy height estimation program stored in the memory and executable on the processor. When the forest canopy height estimation program is executed by the processor, the steps of the forest canopy height estimation method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a forest canopy height estimation program, and when the forest canopy height estimation program is executed by a processor, the steps of the forest canopy height estimation method according to any one of claims 1 to 7 are implemented.

Images