Vegetation distribution extraction method, device and equipment based on dual-orbit radar satellite

Abstract

The present invention relates to the technical field of remote sensing image processing, and discloses a vegetation distribution extraction method, device and equipment based on a dual-orbit radar satellite. The vegetation distribution extraction method based on a dual-orbit radar satellite includes: obtaining a target image set of a target area; the target image set represents a dual-orbit radar image set obtained through a preset cloud-based geospatial analysis platform and processed by multi-temporal filtering; collecting vector target samples of the target area; calculating statistical feature parameters of the target image set along the time scale according to multiple preset time periods, and generating a first mathematical statistical feature set according to the statistical feature parameters; screening out a second mathematical statistical feature set corresponding to the vector target sample in the first mathematical statistical feature set; and classifying and processing the second mathematical statistical feature set to obtain a target vegetation distribution result. The present invention improves the extraction efficiency of vegetation distribution results.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing image processing, and in particular to a vegetation distribution extraction method, device and equipment based on a dual-orbit radar satellite. Background Art

[0002] Vegetation is an important part of the ecosystem. Vegetation coverage is an important basis for understanding the state and function of the ecosystem, such as primary productivity, environmental carrying capacity, and soil erosion intensity. At the same time, in the context of climate change, obtaining the multi-year distribution of vegetation on a large scale is of great significance for monitoring ecosystem changes and sustainable development. At the same time, considering that traditional methods of obtaining vegetation distribution, such as ground survey sampling, consume huge manpower and material resources. Therefore, it is more feasible to use remote sensing imagery to extract large-scale vegetation distribution.

[0003] In related technologies, surface vegetation coverage is monitored through optical remote sensing images. For example, the vegetation information in remote sensing images can be quickly extracted based on the normalized difference vegetation index in the red light band and the near-infrared band. However, cloud occlusion seriously interferes with the acquisition process of optical remote sensing images. Radar images such as synthetic aperture radar can penetrate clouds to continuously monitor the surface and have high application potential for extracting vegetation distribution, but their inherent speckle noise reduces the accuracy of extracting vegetation distribution from a single synthetic aperture radar image. Multiple data are needed to suppress noise interference and maintain the original details of the data.

[0004] Traditional surface cover distribution extraction based on remote sensing images often requires preprocessing of the data to make it analysis-ready data, and then combining multiple data to extract the surface cover distribution. This process involves a large number of platform interactive calculations, which is time-consuming, has low extraction efficiency, and is difficult to implement. Summary of the invention

[0005] In view of this, the present invention provides a vegetation distribution extraction method, device and equipment based on dual-orbit radar satellite to solve the problem of low efficiency in extracting vegetation distribution results.

[0006] In the first aspect, the present invention provides a vegetation distribution extraction method based on a dual-orbit radar satellite, comprising: obtaining a target image set of a target area; the target image set represents a dual-orbit radar image set obtained through a preset cloud-based geospatial analysis platform and subjected to multi-temporal filtering processing; collecting vector target samples of the target area; wherein the vector training samples represent target samples in the form of vector points converted from vegetation samples and non-vegetation samples; calculating statistical feature parameters of the target image set along the time scale according to multiple preset time periods, and generating a first mathematical and statistical feature set for multiple preset time periods based on the statistical feature parameters; screening out a second mathematical and statistical feature set corresponding to the vector target samples from the first mathematical and statistical feature set; and classifying the second mathematical and statistical feature set to obtain a target vegetation distribution result.

[0007] The present invention obtains a dual-track radar target image set that has been processed by multi-temporal filtering through a preset cloud-based geospatial analysis platform. First, the preset cloud-based geospatial analysis platform stores a large amount of data in the cloud and integrates a distributed cloud computing interface. Therefore, when the present invention obtains data, it does not need to download and process data locally, and can directly obtain the target image set on the preset cloud-based geospatial analysis platform. Secondly, the target image set obtained by the present invention has been processed by multi-temporal filtering to reduce the interference of speckle noise and retain the texture details of the target image. The present invention collects vector target samples in the target area, calculates statistical feature parameters of the target image set along the time scale according to multiple preset time periods, and the statistical feature parameters are used to highlight the unique dynamics of vegetation. According to the statistical feature parameters, a first mathematical statistical feature set of multiple preset time periods is generated. The present invention highlights the unique dynamics of vegetation that is different from other land types according to the first mathematical statistical feature set, and further reduces the interference of noise. The present invention selects a second mathematical statistical feature set corresponding to the vector target sample in the first mathematical statistical feature set; the second mathematical statistical feature set is classified and processed to obtain the target vegetation distribution result. Compared with the related technologies, the present invention does not need to perform excessive preprocessing on the data, does not need to call other platforms for interaction, and the obtained high vegetation distribution results are more in line with the actual situation, is less time-consuming, and is simple to operate, thereby improving the extraction efficiency of vegetation distribution results.

[0008] In an optional embodiment, the target image set includes an ascending image set and a descending image set; the statistical characteristic parameters of the target image set along the time scale are calculated according to multiple preset time periods, including: using a first preset function and a second preset function to calculate a first maximum value, a first minimum value and a first mean of the ascending image set, and calculating a second maximum value, a second minimum value and a second mean of the descending image set; calculating a first characteristic parameter based on the first maximum value, the first minimum value and the first mean; wherein the first characteristic parameter includes a first median, a first range, a first standard deviation, a first coefficient of variation, a first kurtosis data and a first skewness data; calculating a second characteristic parameter based on the second maximum value, the second minimum value and the second mean; wherein the second characteristic parameter includes a second median, a second range, a second standard deviation, a second coefficient of variation, a second kurtosis data and a second skewness data; and the first characteristic parameter and the second characteristic parameter constitute the statistical characteristic parameter.

[0009] The target image set of the present invention includes an ascending orbit image set and a descending orbit image set. Images in different directions make the target image collection more accurate and more in line with the actual scene. The present invention calculates the maximum, minimum and mean values, calculates the first characteristic parameter and the second characteristic parameter according to the maximum, minimum and mean values, and forms a statistical characteristic parameter from the first characteristic parameter and the second characteristic parameter, thereby further reducing noise interference, thereby making the statistical characteristic parameters more representative and more accurate, and highlighting the unique dynamic changes of vegetation that are different from other land types.

[0010] In an optional implementation, generating a first mathematical statistical feature set for a plurality of preset time periods according to the statistical feature parameters includes: superimposing a first feature parameter and a second feature parameter in the statistical feature parameters to obtain the first mathematical statistical feature set.

[0011] In an optional embodiment, the second mathematical and statistical feature set is classified and processed to obtain a target vegetation distribution result, including: inputting the second mathematical and statistical feature set into a trained supervised classification model according to multiple preset time periods to obtain vegetation distribution results for multiple preset time periods in the target area; superimposing the vegetation distribution results for multiple preset time periods in the target area to obtain the target vegetation distribution result.

[0012] The present invention extracts vegetation distribution results according to a supervised classification model. Since the supervised classification model has been trained in large quantities, the extracted target vegetation distribution results are more accurate.

[0013] In an optional embodiment, a target image set of a target area is obtained, including: importing the boundary of the target area into a preset cloud-based geospatial analysis platform, obtaining ascending and descending ground distance detection images of a preset resolution in a preset time period; performing multi-temporal filtering on the ground distance detection images, and obtaining a target image set based on a set of processed ground distance detection images.

[0014] In an optional embodiment, vector target samples of a target area are collected, including: collecting a sample set of the target area based on a digital image provided by a preset cloud-based geospatial analysis platform; wherein the sample set includes a plurality of vegetation samples labeled with a first preset value and a plurality of non-vegetation samples labeled with a second preset value; and converting a format of the sample set into a vector format to obtain a vector target sample.

[0015] The present invention converts the format of the sample set into a vector format, thereby making the sample set more stable and occupying less storage space.

[0016] In an optional embodiment, the vegetation distribution extraction method based on dual-orbit radar satellite also includes a training process for a supervised classification model. The process of training the supervised classification model includes: obtaining a historical mathematical and statistical feature set, inputting the historical mathematical and statistical feature set into the supervised classification model, and training the supervised classification model, wherein the input of the supervised classification model is the historical mathematical and statistical feature set, and the output of the supervised classification model is the vegetation distribution results for multiple preset time periods.

[0017] In a second aspect, the present invention provides a vegetation distribution extraction device based on a dual-orbit radar satellite, comprising:

[0018] A data acquisition module is used to acquire a target image set of a target area; the target image set represents a dual-track radar image set obtained through a preset cloud-based geospatial analysis platform and processed by multi-temporal filtering; a sample acquisition module is used to collect vector target samples of the target area; wherein the vector training samples represent target samples in the form of vector points converted from vegetation samples and non-vegetation samples; a first mathematical and statistical feature set determination module is used to calculate the statistical feature parameters of the target image set along the time scale according to multiple preset time periods, and generate a first mathematical and statistical feature set for multiple preset time periods based on the statistical feature parameters; a second mathematical and statistical feature set determination module is used to screen out a second mathematical and statistical feature set corresponding to the vector target sample from the first mathematical and statistical feature set; a vegetation distribution result determination module is used to classify the second mathematical and statistical feature set to obtain a target vegetation distribution result.

[0019] In a third aspect, the present invention provides a computer device comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the vegetation distribution extraction method based on a dual-orbit radar satellite according to the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0020] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the vegetation distribution extraction method based on a dual-orbit radar satellite according to the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 is a schematic flow chart of a vegetation distribution extraction method based on a dual-orbit radar satellite according to an embodiment of the present invention;

[0023] Figure 2 is a flow chart of another vegetation distribution extraction method based on dual-orbit radar satellites according to an embodiment of the present invention;

[0024] Figure 3 is a flow chart of another vegetation distribution extraction method based on dual-orbit radar satellites according to an embodiment of the present invention;

[0025] Figure 4 is a schematic diagram of vegetation distribution results in the Yarlung Zangbo River Basin according to an embodiment of the present invention;

[0026] Figure 5 is a structural block diagram of a vegetation distribution extraction device based on a dual-orbit radar satellite according to an embodiment of the present invention;

[0027] Figure 6 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0029] The embodiment of the present invention provides a vegetation distribution extraction method based on a dual-orbit radar satellite, which achieves the effect of accurately extracting vegetation distribution results by processing mathematical and statistical features.

[0030] According to an embodiment of the present invention, an embodiment of a vegetation distribution extraction method based on a dual-orbit radar satellite is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that shown here.

[0031] In this embodiment, a vegetation distribution extraction method based on a dual-orbit radar satellite is provided, which can be used in a server. Figure 1 is a flow chart of a vegetation distribution extraction method based on a dual-orbit radar satellite according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0032] Step S101, obtaining a target image set of a target area; the target image set refers to a dual-track radar image set obtained through a preset cloud-based geospatial analysis platform and processed by multi-temporal filtering.

[0033] Among them, the target area represents the area where the vegetation distribution results are to be analyzed, and the range of the target area can be customized according to needs; the target image set includes an ascending orbit image set and a descending orbit image set, and the target image set can be a Sentinel-1 ground distance detection image set.

[0034] In some optional embodiments, obtaining a target image set of the target area includes: importing the boundary of the target area into a preset cloud-based geospatial analysis platform, obtaining ascending and descending ground distance detection images of a preset resolution in a preset time period; performing multi-temporal filtering on the ground distance detection images, and obtaining a target image set based on a set of processed ground distance detection images.

[0035] The preset cloud geospatial analysis platform may be Google Earth Engine, a cloud computing platform for processing satellite images and other earth observation data, the preset time period may be one year, and the multiple preset time periods may be multiple years, such as 2015 to 2022, and other similar preset time periods may be explained in this interpretation. The preset resolution may be customized according to actual needs.

[0036] Exemplarily, the boundary of the Yarlung Zangbo River Basin is imported into Google Earth Engine, and the ascending orbit ground distance detection images and descending orbit ground distance detection images between 2015 and 2022 are obtained, totaling 10110 images. The 10110 images are integrated to obtain an image set, and the image set is subjected to multi-temporal (LEE SIGMA) filtering to obtain the Sentinel-1 image set.

[0037] Step S102 , collecting vector target samples of the target area; wherein the vector training samples represent target samples in the form of vector points converted from vegetation samples and non-vegetation samples.

[0038] In some optional embodiments, collecting vector target samples of the target area includes: collecting a sample set of the target area based on digital images provided by a preset cloud-based geospatial analysis platform; wherein the sample set includes a plurality of vegetation samples labeled with a first preset value and a plurality of non-vegetation samples labeled with a second preset value; and converting the format of the sample set into a vector format to obtain the vector target samples.

[0039] Among them, the first preset value can be 1, the second preset value can be 0, the vegetation sample represents the sample of the ground with vegetation, and the non-vegetation sample represents the sample of the ground without vegetation. By way of example, the non-vegetation sample can be a bare land sample, a water body sample, and an impermeable surface sample. The ratio of the number of multiple vegetation samples to multiple non-vegetation samples can be 1 to 1.

[0040] In some optional implementations, labels are set for the vector target samples, and the vector target samples are uploaded to a useful data (Assets) module of a preset cloud-based geospatial analysis platform for subsequent application to the vector target samples.

[0041] Step S103, calculating statistical characteristic parameters of the target image set along the time scale according to a plurality of preset time periods, and generating a first mathematical statistical feature set of the plurality of preset time periods according to the statistical characteristic parameters.

[0042] Among them, the statistical characteristic parameters of the target image set along the time scale are calculated year by year, the maximum value, minimum value and mean value are calculated according to the target image set based on the first preset function and the second preset function, and the statistical characteristic parameters are calculated according to the maximum value, minimum value and mean value.

[0043] In some optional implementations, generating a first mathematical statistical feature set for a plurality of preset time periods according to the statistical feature parameters includes: superimposing a first feature parameter and a second feature parameter in the statistical feature parameters to obtain the first mathematical statistical feature set.

[0044] The superposition process includes combining the first characteristic parameter and the second characteristic parameter. For example, the statistical characteristic parameter includes 6 first characteristic parameters and 6 second characteristic parameters. The first characteristic parameters and the second characteristic parameters are superimposed to obtain 12 first mathematical statistical characteristic sets.

[0045] Step S104: Filter out a second mathematical statistical feature set corresponding to the vector target sample from the first mathematical statistical feature set.

[0046] In an embodiment of the present invention, the first mathematical and statistical feature set includes characteristic parameters of all positions in the target area, and the second mathematical and statistical feature set corresponding to the vector target sample is screened out from the first mathematical and statistical feature set to achieve the vegetation distribution result for the vector target sample.

[0047] Step S105: classify the second mathematical and statistical feature set to obtain a target vegetation distribution result.

[0048] In some optional embodiments, the second set of mathematical and statistical features is input into a trained supervised classification model according to multiple preset time periods to obtain vegetation distribution results for multiple preset time periods of the target area, and the vegetation distribution results for multiple preset time periods of the target area are superimposed to obtain target vegetation distribution results.

[0049] Exemplarily, the second mathematical and statistical feature set is input into the trained supervised classification model year by year to obtain the vegetation distribution result of each year, and the vegetation distribution result of each year is combined together to form the target vegetation distribution result.

[0050] The vegetation distribution extraction method based on dual-orbit radar satellite provided in this embodiment is a dual-orbit radar target image set that has been processed by multi-temporal filtering through a preset cloud-based geospatial analysis platform. First, the preset cloud-based geospatial analysis platform stores a large amount of data in the cloud and integrates a distributed cloud computing interface. Therefore, when the embodiment of the present invention obtains data, it does not need to download and process the data locally, and can directly obtain the target image set on the preset cloud-based geospatial analysis platform. Secondly, the target image set obtained by the embodiment of the present invention has been processed by multi-temporal filtering to reduce the interference of speckle noise and retain the texture details of the target image. The embodiment of the present invention collects vector target samples in the target area, calculates statistical feature parameters of the target image set along the time scale according to multiple preset time periods, and the statistical feature parameters are used to highlight the unique dynamics of vegetation. According to the statistical feature parameters, a first mathematical statistical feature set of multiple preset time periods is generated. The embodiment of the present invention highlights the unique dynamics of vegetation that is different from other land types according to the first mathematical statistical feature set, and further reduces the interference of noise. The embodiment of the present invention selects a second mathematical statistical feature set corresponding to the vector target sample from the first mathematical statistical feature set; the second mathematical statistical feature set is classified and processed to obtain the target vegetation distribution result. Compared with the related art, the embodiment of the present invention does not need to perform excessive preprocessing on the data, and does not need to call other platforms for interaction. The obtained high vegetation distribution results are more in line with the actual situation, are less time-consuming, simple to operate, and improve the extraction efficiency of vegetation distribution results.

[0051] In this embodiment, a vegetation distribution extraction method based on a dual-orbit radar satellite is provided, which can be used for the above-mentioned server. Figure 2 FIG. 1 is a flow chart of another vegetation distribution extraction method based on dual-orbit radar satellite according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps:

[0052] Step S201, obtain a target image set of the target area; the target image set refers to a dual-track radar image set obtained through a preset cloud-based geospatial analysis platform and processed by multi-temporal filtering. Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.

[0053] Step S202, collect vector target samples of the target area; wherein the vector training samples represent target samples in the form of vector points converted from vegetation samples and non-vegetation samples. Figure 1 Step S102 of the illustrated embodiment will not be described in detail here.

[0054] Step S203, calculating statistical characteristic parameters of the target image set along the time scale according to a plurality of preset time periods, and generating a first mathematical statistical feature set of the plurality of preset time periods according to the statistical characteristic parameters.

[0055] The target image set includes an ascending image set and a descending image set in a single preset time period.

[0056] Specifically, the above step S203 includes:

[0057] Step S2031, using the first preset function and the second preset function, calculate the first maximum value, the first minimum value and the first mean value of the ascending orbit image set, and calculate the second maximum value, the second minimum value and the second mean value of the descending orbit image set.

[0058] The first preset function is reduce(), the second preset function is ee.Reducer(), and the formula for calculating the first maximum value is:

[0059] max 1 =D 1 .reduce(ee.Reducer.max());

[0060] Among them, max 1 represents the first maximum value, D 1 It represents the ascending image set in the target image set; reduce() represents the first preset function; ee.Reducer() represents the second preset function.

[0061] The formula for calculating the first minimum is:

[0062] min 1 =D 1 .reduce(ee.Reducer.min());

[0063] Among them, min 1 Indicates the first minimum value, D 1 It represents the ascending image set in the target image set; reduce() represents the first preset function; ee.Reducer() represents the second preset function.

[0064] The formula for calculating the first mean is:

[0065] mean 1 =D 1 .reduce(ee.Reducer.mean());

[0066] Among them, min 1 represents the first mean, D 1 It represents the ascending image set in the target image set; reduce() represents the first preset function; ee.Reducer() represents the second preset function.

[0067] The formula for calculating the second maximum value is:

[0068] max 2 =D 2 .reduce(ee.Reducer.max());

[0069] Among them, max 2 Indicates the second maximum value, D 2 It represents the reduced track image set in the target image set; reduce() represents the first preset function; ee.Reducer() represents the second preset function.

[0070] The formula for calculating the second minimum is:

[0071] min 2 =D 2 .reduce(ee.Reducer.min());

[0072] Among them, min 2 Indicates the second minimum value, D 2 It represents the reduced track image set in the target image set; reduce() represents the first preset function; ee.Reducer() represents the second preset function.

[0073] The formula for calculating the second mean is:

[0074] mean 2 =D 2 .reduce(ee.Reducer.mean());

[0075] Among them, min 2 represents the second mean, D 2 It represents the reduced track image set in the target image set; reduce() represents the first preset function; ee.Reducer() represents the second preset function.

[0076] Step S2032, calculating a first characteristic parameter according to the first maximum value, the first minimum value and the first mean value.

[0077] Among them, the first characteristic parameters include a first median, a first range, a first standard deviation, a first coefficient of variation, a first kurtosis data, and a first skewness data.

[0078] Step S2033, calculating a second characteristic parameter according to the second maximum value, the second minimum value and the second mean value.

[0079] The second characteristic parameters include a second median, a second range, a second standard deviation, a second coefficient of variation, a second kurtosis data, and a second skewness data.

[0080] Among them, the formula for calculating the first median is:

[0081] median 1 =D 1 .reduce(ee.Reducer.median());

[0082] Among them, median 1 Denotes the first median; D 1 It represents the ascending image set in the target image set; reduce() represents the first preset function; ee.Reducer() represents the second preset function.

[0083] The formula for calculating the second median is:

[0084] median 2 =D 2 .reduce(ee.Reducer.median());

[0085] Among them, median 2 Indicates the second median; D 2 It represents the reduced track image set in the target image set; reduce() represents the first preset function; ee.Reducer() represents the second preset function.

[0086] The formula for calculating the first range is:

[0087] range 1 =max 1 .subtract(min 1 );

[0088] Among them, range 1 Indicates the first range; max 1 Indicates the first maximum value; min 1 represents the first minimum value; subtract() represents the third preset function.

[0089] The formula for calculating the second range is:

[0090] range 2 =max 2 .subtract(min 2 );

[0091] Among them, range 2 Indicates the second extreme; max 2 Indicates the second maximum value; min 2 represents the second minimum value; subtract() represents the third preset function.

[0092] The formula for calculating the first standard deviation is:

[0093] stdDev 1 =D 1 .reduce(ee.Reducer.stdDev());

[0094] Among them, stdDev 1 Denotes the first standard deviation; D 1 It represents the ascending image set in the target image set; reduce() represents the first preset function; ee.Reducer() represents the second preset function.

[0095] The formula for calculating the second standard deviation is:

[0096] stdDev 2 =D 2 .reduce(ee.Reducer.stdDev());

[0097] Among them, stdDev 2 Denotes the second standard deviation; D 2 It represents the reduced track image set in the target image set; reduce() represents the first preset function; ee.Reducer() represents the second preset function.

[0098] The formula for calculating the first coefficient of variation is:

[0099] coefficient 1 =stdDev 1 .divide(mean 1 );

[0100] Among them, coefficient 1 Indicates the first coefficient of variation; stdDev 1 Indicates the first standard deviation; mean 1 represents the first mean; divide() represents the fourth preset function.

[0101] The formula for calculating the second coefficient of variation is:

[0102] coefficient 2 =stdDev 2 .divide(mean 2 );

[0103] Among them, coefficient 2 Indicates the second coefficient of variation; stdDev 2 Indicates the second standard deviation; mean 2 Represents the second mean; divide() represents the fourth preset function.

[0104] The formula for calculating the first kurtosis is:

[0105] kurtosis 1 =D 1 .reduce(ee.Reducer.kurtosis());

[0106] Among them, kurtosis 1 represents the first kurtosis; D 1 It represents the ascending image set in the target image set; reduce() represents the first preset function; ee.Reducer() represents the second preset function.

[0107] The formula for calculating the second kurtosis is:

[0108] kurtosis 2 =D 2 .reduce(ee.Reducer.kurtosis());

[0109] Among them, kurtosis 2 represents the second kurtosis; D 2 It represents the reduced track image set in the target image set; reduce() represents the first preset function; ee.Reducer() represents the second preset function.

[0110] The formula for calculating the first skewness is:

[0111] skew 1 =D 1 .reduce(ee.Reducer.skew());

[0112] Among them, skew 1 represents the first skewness; D 1 It represents the ascending image set in the target image set; reduce() represents the first preset function; ee.Reducer() represents the second preset function.

[0113] The formula for calculating the second skewness is:

[0114] skew 2 =D 2 .reduce(ee.Reducer.skew());

[0115] Among them, skew 2 Indicates the second skewness; D 2 It represents the reduced track image set in the target image set; reduce() represents the first preset function; ee.Reducer() represents the second preset function.

[0116] Step S2034, forming a statistical characteristic parameter from the first characteristic parameter and the second characteristic parameter.

[0117] Step S2035: generating a first mathematical statistical feature set for a plurality of preset time periods according to the statistical feature parameters.

[0118] In the embodiment of the present invention, 12 statistical feature parameters form a first mathematical statistical feature set for a plurality of preset time periods.

[0119] Step S204: Filter out a second set of mathematical statistics features corresponding to the vector target sample from the first set of mathematical statistics features. Figure 1 Step S104 of the illustrated embodiment will not be described in detail here.

[0120] Step S205: classify the second mathematical and statistical feature set to obtain a target vegetation distribution result.

[0121] Specifically, the above step S205 includes:

[0122] Step S2051: input the second mathematical and statistical feature set into the trained supervised classification model according to multiple preset time periods to obtain vegetation distribution results of the target area in multiple preset time periods.

[0123] Step S2052, superimposing the vegetation distribution results of multiple preset time periods in the target area to obtain the target vegetation distribution result.

[0124] In an embodiment of the present invention, the second mathematical and statistical feature set is input into the trained supervised classification model year by year to obtain the annual vegetation distribution results, and the vegetation distribution results of each year are combined based on the add() function provided by Google Earth Engine to form the target vegetation distribution results.

[0125] In some optional implementations, the vegetation distribution result may be in the form of a label, the label of the vegetation distribution is greater than 0, and the label of other surface cover distributions is 0, and multiple labels are combined together to obtain the target vegetation distribution result.

[0126] In some optional embodiments, the vegetation distribution extraction method based on dual-orbit radar satellite also includes a training process for a supervised classification model. The process of training the supervised classification model includes: obtaining a historical mathematical and statistical feature set, inputting the historical mathematical and statistical feature set into the supervised classification model, and training the supervised classification model, wherein the input of the supervised classification model is the historical mathematical and statistical feature set, and the output of the supervised classification model is the vegetation distribution results for multiple preset time periods.

[0127] Among them, the historical statistical feature parameters in the historical mathematical and statistical feature set are named respectively to obtain a name set, and the historical samples are converted into patterns that can be recognized by the supervised classification model based on the sampleRegions() function provided on Google Earth Engine. The supervised classification model is trained based on the train() function provided on Google Earth Engine. The random forest model (RF) is selected as a typical supervised classification model, and the number of decision trees is set to 1000. The historical statistical feature parameters are input into the supervised classification model, and the supervised classification model is trained.

[0128] The vegetation distribution extraction method based on dual-orbit radar satellite provided in this embodiment comprises an ascending orbit image set and a descending orbit image set in the target image set. Images in different directions make the target image collection more accurate and more in line with the actual scene. The present invention calculates the maximum, minimum and mean values, calculates the first characteristic parameter and the second characteristic parameter according to the maximum, minimum and mean values, and forms a statistical characteristic parameter from the first characteristic parameter and the second characteristic parameter, thereby further reducing the interference of noise, thereby making the statistical characteristic parameters more representative and more accurate, and highlighting the unique dynamic changes of vegetation that are different from other land types.

[0129] In this embodiment, a vegetation distribution extraction method based on a dual-orbit radar satellite is provided, which can be used for the above-mentioned server. Figure 3 FIG. 4 is a flow chart of another vegetation distribution extraction method based on a dual-orbit radar satellite according to an embodiment of the present invention. Figure 3 As shown in the figure, the process includes: remote sensing data acquisition, supervised classification model construction and result output.

[0130] Among them, remote sensing data acquisition includes: obtaining the Sentinel-1 dataset based on a preset cloud-based geospatial analysis platform, and determining the annual time series of the ascending image set and the annual time series of the descending image set based on the Sentinel-1 dataset, wherein the number of annual time series of the ascending image set and the number of annual time series of the descending image set is equal to the number of images in the year, and the number of series is equal to the number of years.

[0131] The construction of the supervised classification model includes: obtaining the original format training samples according to the preset cloud-based geospatial analysis platform, obtaining the target training samples after format conversion, obtaining the mathematical statistics set of ascending orbit images and the mathematical statistics set of descending orbit images according to the time series within the year of the ascending orbit image set and the time series within the year of the descending orbit image set, wherein the number of features in the mathematical statistics set of ascending orbit images and the mathematical statistics set of descending orbit images are both 6, with a total of 12 statistical feature parameters, and the supervised classification model is trained according to the mathematical statistics set of ascending orbit images, the mathematical statistics set of descending orbit images and the target training samples.

[0132] The result output includes: obtaining the mathematical statistics set of ascending orbit images and the mathematical statistics set of descending orbit images to be analyzed, inputting the mathematical statistics set of ascending orbit images and the mathematical statistics set of descending orbit images into the supervised classification model year by year, obtaining the vegetation distribution results year by year, and superimposing the vegetation distribution results year by year to obtain the vegetation distribution results for many years. For example, Figure 4 The following is a schematic diagram of vegetation distribution in the Yarlung Zangbo River Basin.

[0133] In this embodiment, a vegetation distribution extraction device based on a dual-orbit radar satellite is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0134] This embodiment provides a vegetation distribution extraction device based on a dual-orbit radar satellite. Figure 5 As shown, including:

[0135] The data acquisition module 501 is used to acquire a target image set of a target area; the target image set represents a dual-track radar image set obtained through a preset cloud-based geospatial analysis platform and processed by multi-temporal filtering.

[0136] The sample collection module 502 is used to collect vector target samples of the target area; wherein the vector training samples represent target samples in the form of vector points converted from vegetation samples and non-vegetation samples.

[0137] The first mathematical and statistical feature set determination module 503 is used to calculate the statistical feature parameters of the target image set along the time scale according to multiple preset time periods, and generate the first mathematical and statistical feature sets of the multiple preset time periods according to the statistical feature parameters.

[0138] The second mathematical and statistical feature set determination module 504 is used to select a second mathematical and statistical feature set corresponding to the vector target sample from the first mathematical and statistical feature set.

[0139] The vegetation distribution result determination module 505 is used to classify the second mathematical and statistical feature set to obtain a target vegetation distribution result.

[0140] In some optional implementations, the data acquisition module 501 includes:

[0141] The ground distance detection image acquisition unit is used to import the boundary of the target area into a preset cloud-based geospatial analysis platform to obtain the ascending and descending ground distance detection images of a preset resolution in a preset time period.

[0142] The target image set determination unit is used to detect the image set according to the ground distance to obtain the target image set.

[0143] In some optional embodiments, the sample collection module 502 includes:

[0144] The sample set determination unit is used to collect a sample set of the target area based on the digital image provided by the preset cloud geospatial analysis platform; wherein the sample set includes a plurality of vegetation samples whose labels are the first preset value and a plurality of non-vegetation samples whose labels are the second preset value.

[0145] The format conversion unit is used to convert the format of the sample set into a vector format to obtain a vector target sample.

[0146] In some optional implementations, the first mathematical and statistical feature set determination module 503 includes:

[0147] The numerical calculation unit is used to calculate the first maximum value, the first minimum value and the first mean value of the ascending orbit image set, and calculate the second maximum value, the second minimum value and the second mean value of the descending orbit image set by using the first preset function and the second preset function.

[0148] The first characteristic parameter determination unit is used to calculate the first characteristic parameter according to the first maximum value, the first minimum value and the first mean; wherein the first characteristic parameter includes the first median, the first range, the first standard deviation, the first coefficient of variation, the first kurtosis data and the first skewness data.

[0149] The first characteristic parameter determination unit is used to calculate the second characteristic parameter according to the second maximum value, the second minimum value and the second mean; wherein the second characteristic parameter includes the second median, the second range, the second standard deviation, the second coefficient of variation, the second kurtosis data and the second skewness data.

[0150] The statistical characteristic parameter determination unit is used to form a statistical characteristic parameter from the first characteristic parameter and the second characteristic parameter.

[0151] The superposition unit is used to superimpose the first characteristic parameter and the second characteristic parameter in the statistical characteristic parameters to obtain a first mathematical statistical feature set.

[0152] In some optional implementations, the vegetation distribution result determination module 505 includes:

[0153] The model classification unit is used to input the second mathematical and statistical feature set into the trained supervised classification model according to multiple preset time periods to obtain vegetation distribution results of multiple preset time periods in the target area.

[0154] The target vegetation distribution result determination unit is used to superimpose the vegetation distribution results of multiple preset time periods in the target area to obtain the target vegetation distribution result.

[0155] In some optional implementations, the vegetation distribution extraction device based on dual-orbit radar satellites further includes:

[0156] The model training unit is used to obtain a set of historical mathematical and statistical features, input the set of historical mathematical and statistical features into a supervised classification model, and train the supervised classification model, wherein the input of the supervised classification model is the set of historical mathematical and statistical features, and the output of the supervised classification model is the vegetation distribution results of multiple preset time periods.

[0157] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0158] The vegetation distribution extraction device based on dual-orbit radar satellite in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0159] The embodiment of the present invention also provides a computer device having the above Figure 5 The vegetation distribution extraction device based on dual-orbit radar satellite is shown.

[0160] See also Figure 6 , Figure 6 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 6 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 6 A processor 10 is taken as an example.

[0161] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0162] The memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0163] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0164] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0165] The computer device further comprises a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0166] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0167] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A vegetation distribution extraction method based on dual-orbit radar satellite, characterized in that: The method comprises: Acquire a target image set of the target area; the target image set represents a dual-track radar image set obtained through a preset cloud-based geospatial analysis platform and processed by multi-temporal filtering; wherein the preset cloud-based geospatial analysis platform is Google Earth Engine; Collecting vector target samples of the target area; wherein the vector target samples represent target samples in the form of vector points converted from vegetation samples and non-vegetation samples; Calculate the statistical characteristic parameters of the target image set along the time scale according to a plurality of preset time periods, and generate a first mathematical statistical characteristic set of the plurality of preset time periods according to the statistical characteristic parameters; wherein the first mathematical statistical characteristic set includes characteristic parameters of all positions in the target area; Filtering out a second mathematical statistical feature set corresponding to the vector target sample from the first mathematical statistical feature set; Performing classification processing on the second mathematical and statistical feature set to obtain a target vegetation distribution result; The classifying process of the second mathematical and statistical feature set to obtain a target vegetation distribution result includes: Inputting the second set of mathematical and statistical features into the trained supervised classification model according to the multiple preset time periods to obtain vegetation distribution results of the multiple preset time periods of the target area; wherein the supervised classification model is trained based on a preset function provided on the Google Earth Engine; Superimposing the vegetation distribution results of the target area in the plurality of preset time periods to obtain the target vegetation distribution result; The target vegetation distribution result is in the form of a label.

2. The method according to claim 1, characterized in that The target image set includes an ascending orbit image set and a descending orbit image set; and calculating the statistical characteristic parameters of the target image set along the time scale according to a plurality of preset time periods includes: Using a first preset function and a second preset function, calculating a first maximum value, a first minimum value, and a first mean value of the ascending track image set, and calculating a second maximum value, a second minimum value, and a second mean value of the descending track image set; Calculate a first characteristic parameter according to the first maximum value, the first minimum value and the first mean; wherein the first characteristic parameter includes a first median, a first range, a first standard deviation, a first coefficient of variation, a first kurtosis data and a first skewness data; Calculate the second characteristic parameters according to the second maximum value, the second minimum value and the second mean; wherein the second characteristic parameters include the second median, the second range, the second standard deviation, the second coefficient of variation, the second kurtosis data and the second skewness data; The statistical characteristic parameter is composed of the first characteristic parameter and the second characteristic parameter.

3. The method according to claim 2, characterized in that The step of generating a first set of mathematical statistical features for the plurality of preset time periods according to the statistical feature parameters comprises: The first characteristic parameter and the second characteristic parameter in the statistical characteristic parameters are superimposed to obtain the first mathematical statistical feature set.

4. The method according to any one of claims 1 to 3, characterized in that The step of acquiring a target image set of a target area includes: Importing the boundary of the target area into the preset cloud-based geospatial analysis platform, and obtaining ground distance detection images of ascending and descending orbits at a preset resolution in the preset time period; The ground distance detection images are subjected to multi-temporal filtering processing, and a target image set is obtained according to a set of processed ground distance detection images.

5. The method according to any one of claims 1 to 3, characterized in that The collecting of vector target samples of the target area includes: According to the digital image provided by the preset cloud-based geospatial analysis platform, a sample set of the target area is collected; wherein the sample set includes a plurality of vegetation samples whose labels are a first preset value and a plurality of non-vegetation samples whose labels are a second preset value; The format of the sample set is converted into a vector format to obtain the vector target sample.

6. The method according to claim 1, characterized in that The method also includes a process of training the supervised classification model, and the process of training the supervised classification model includes: Obtain a historical mathematical and statistical feature set, input the historical mathematical and statistical feature set into the supervised classification model, and train the supervised classification model, wherein the input of the supervised classification model is the historical mathematical and statistical feature set, and the output of the supervised classification model is the vegetation distribution results of the multiple preset time periods.

7. A vegetation distribution extraction device based on dual-orbit radar satellite, characterized in that: The device comprises: A data acquisition module, used to acquire a target image set of a target area; the target image set represents a dual-track radar image set obtained through a preset cloud-based geospatial analysis platform and processed by multi-temporal filtering; wherein the preset cloud-based geospatial analysis platform is Google Earth Engine; A sample collection module, used for collecting vector target samples of the target area; wherein the vector target samples represent target samples in the form of vector points converted from vegetation samples and non-vegetation samples; A first mathematical and statistical feature set determination module is used to calculate the statistical feature parameters of the target image set along the time scale according to a plurality of preset time periods, and generate a first mathematical and statistical feature set of the plurality of preset time periods according to the statistical feature parameters; wherein the first mathematical and statistical feature set includes feature parameters of all positions in the target area; A second mathematical and statistical feature set determination module, configured to filter out a second mathematical and statistical feature set corresponding to the vector target sample from the first mathematical and statistical feature set; A vegetation distribution result determination module, used for classifying the second mathematical and statistical feature set to obtain a target vegetation distribution result; The classifying process of the second mathematical and statistical feature set to obtain a target vegetation distribution result includes: Inputting the second set of mathematical and statistical features into the trained supervised classification model according to the multiple preset time periods to obtain vegetation distribution results of the multiple preset time periods of the target area; wherein the supervised classification model is trained based on a preset function provided on the Google Earth Engine; Superimposing the vegetation distribution results of the target area in the plurality of preset time periods to obtain the target vegetation distribution result; The target vegetation distribution result is in the form of a label.

8. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the vegetation distribution extraction method based on a dual-orbit radar satellite as described in any one of claims 1 to 6 by executing the computer instructions.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the vegetation distribution extraction method based on dual-orbit radar satellites according to any one of claims 1 to 6.

Images