Method and device for inverting forest height based on single-baseline InSAR (Interferometric Synthetic Aperture Radar)

Abstract

The invention discloses a forest height inversion method and device based on a single-baseline InSAR, and the method comprises the steps: carrying out the interference processing of the observation data of the single-baseline InSAR, obtaining an interference complex coherence coefficient, and calculating a coherence compensation factor; calculating a local effective vertical wave number and a local incident angle; sub-aperture decomposition is carried out on the main image and the auxiliary image, multiple pairs of sub-view SLC images are obtained, and interference processing is carried out; least square coherent straight line fitting is carried out on the multiple sub-view interference complex coherence coefficients, and the earth surface phase is judged; the sub-apparent interference complex coherence coefficient farthest from the earth surface phase is regarded as a sub-apparent interference complex coherence coefficient # imgabs0 # with the minimum earth surface scattering, and the earth surface scattering is ignored for solving to obtain an initial forest height; and judging the sub-apparent interference complex coherence coefficient farthest from # imgabs1 #, and inverting the forest height by combining the two sub-apparent interference complex coherence coefficients, thereby realizing inversion of the forest height by single-baseline single-polarization InSAR data.

Description

Technical Field

[0001] The present invention relates to the field of forestry remote sensing of synthetic aperture radar interferometry (InSAR), and particularly to a method and device for forest height inversion based on synthetic aperture radar interferometric images. Background Art

[0002] Forest height plays a crucial role in the ecosystem and is a key factor in regulating the carbon balance and climate. The InSAR (Interferometric Synthetic Aperture Radar) technology has the capabilities of continuous, cloud-penetrating, large-scale observation, and at the same time can penetrate into the interior of the forest, which is a preferred observation means compared with optical remote sensing and lidar (LiDAR). Most of the current spaceborne SAR systems collect large-scale observation data in a single polarization mode, and mostly use the bistatic observation or formation flight mode. Inverting forest height based on single-baseline single-polarization InSAR has always been a technical difficulty in the field of forestry remote sensing.

[0003] The challenges faced in inverting forest height from single-baseline single-polarization InSAR data mainly come from the fact that the observation information provided by the observation data is less. In the single-polarization working mode, only two observables can be provided after interferometric processing, namely the coherence amplitude and the interference phase, which are difficult to meet the requirements of solving complex physical models. For example, based on the widely used Random Volume over Ground (RVoG) model, four parameters including forest height, extinction coefficient, surface phase, and ground volume ratio need to be solved simultaneously. It is difficult to solve the four unknown parameters based on the two observables provided by single-baseline single-polarization InSAR. Moreover, the InSAR DEM generated in the forest area by InSAR data is troubled by "inability to penetrate", which results in uncertain forest height signals in the DEM (Digital Elevation Model) and cannot be converted into accurate surface phases. In addition, even if accurate surface phases are provided, the single-baseline single-polarization InSAR observed complex coherence coefficient only provides two observables (coherence amplitude and interference phase), and it is impossible to achieve a balance in the process of solving the three unknowns of forest height, ground volume ratio, and extinction coefficient.

[0004] In summary, it is necessary to develop a single-baseline single-polarization InSAR forest height inversion scheme that does not rely on external data assistance, can overcome the dependence on multi-polarization, multi-baseline observation data or other platform observation data, and achieve refined inversion of forest height under the constraint of physical models. Summary of the Invention

[0005] In view of the deficiencies in the above-mentioned prior art, the purpose of the present invention is to provide a method and device for inverting forest height based on single-baseline InSAR, realizing a forest height inversion method that does not rely on multi-baseline, multi-polarization or other platform observation data assistance.

[0006] In a first aspect, a method for inverting forest height based on single-baseline InSAR is provided, including the following steps: S1: Obtain single-polarization single-baseline InSAR data of the observation area, perform interference processing to obtain an interference complex coherence coefficient, and calculate a coherence compensation factor; S2: Calculate an effective vertical wave number through baseline parameter information, and perform local correction to obtain a local incident angle and a local effective vertical wave number; S3: Perform azimuth sub-aperture decomposition on the master and slave images of the original InSAR data to obtain multiple pairs of sub-view SLC images; S4: For the multiple sub-view interference complex coherence coefficients obtained by performing interference processing on the multiple pairs of sub-view SLC images, perform coherent amplitude correction on the multiple sub-view interference complex coherence coefficients based on the coherence compensation factor, then spread the points into the complex plane, and then perform least squares coherent straight line fitting on the multiple sub-view interference complex coherence coefficients, and combine the positive and negative judgment of the effective vertical wave number to obtain the surface phase; S5: In the unit circle of the complex plane, find the sub-view interference complex coherence coefficient that is farthest from the surface phase, regard it as the sub-view interference complex coherence coefficient with the least surface scattering, and solve for the initial forest height based on the random ground scatterer two-layer model and ignoring the surface scattering; S6: In the unit circle of the complex plane, find the sub-view interference complex coherence coefficient that is farthest from the sub-view interference complex coherence coefficient with the least surface scattering, combine the above two sub-view interference complex coherence coefficients, and inversely calculate the final forest height based on the random ground scatterer two-layer model by establishing a non-linear iterative inversion framework.

[0007] Further, step S1 specifically includes: Obtain single-polarization single-baseline InSAR data of the observation area and perform interference processing: ; In the formula, and respectively represent the master image and the slave image participating in the interference, represents conjugate multiplication of the master image and the slave image, represents obtaining the expected value; the interference complex coherence coefficient is a complex number, expressed as , and are the interference phase and the coherence amplitude respectively; The calculation includes the signal-to-noise ratio de-coherence factor And the system quantization decoherence factor The coherence compensation factor, and correct the coherence amplitude of the interferometric complex coherence coefficient. The expression for correcting the coherence amplitude of the observed interferometric complex coherence coefficient is .

[0008] Furthermore, step S2 specifically includes: Obtain the nominal incident angle of the electromagnetic wave signal through the baseline parameters and geometric relationships during InSAR data observation , and obtain the effective vertical wave number , and the expression is: ; In the formula, is the effective vertical wave number, represents the radar wave wavelength, and respectively represent the slant range and the vertical baseline length; when the InSAR system is dual-station data, , when the InSAR system is repeat-pass station data, ; Use an external DEM product to calculate the local slope , and perform local correction on the nominal incident angle to obtain the local incident angle : ; Based on the following formula, perform local correction on the effective vertical wave number to obtain the local effective vertical wave number : .

[0009] Furthermore, perform azimuth sub-aperture decomposition on the master and slave images of the original InSAR data. Specifically: perform azimuth fast Fourier transform on the master and slave images of the original InSAR data respectively to convert them to the frequency domain; in the frequency domain, based on a preset sub-aperture segmentation strategy, calculate the azimuth spectral center and the spectral bandwidth of the image after sub-aperture decomposition; after windowing and filtering each sub-aperture spectrum, perform two-dimensional inverse fast Fourier transform to reconstruct multiple sub-view images of the master and slave images respectively; if finally obtain sub-view master images and sub-view slave images, then through pairwise pairing, sub-view SLC image interference pairs can be generated.

[0010] Furthermore, step S4 specifically includes: After obtaining multiple sub - view interference complex coherence coefficients through interference processing of multiple pairs of sub - view SLC images, use the signal - to - noise ratio decoherence factor and the system quantization decoherence factor to perform coherent amplitude correction on the multiple sub - view interference complex coherence coefficients; Expand the multiple sub - view interference complex coherence coefficients after the above - mentioned coherent amplitude correction onto the complex plane; Use total least - squares constraint to perform coherent straight - line fitting on the multiple sub - view interference complex coherence coefficients within the unit circle of the complex plane, and obtain the phases and of the two intersection points of the coherent straight line and the unit circle of the complex plane. Determine that one of them is the surface phase point through the magnitude relationship between the positive and negative values of the effective vertical wave number and the phases of the two obtained intersection points. The judgment criterion is: when > 0, if , then is , if , then is ; when < 0, if , then is , if , then is ; where, represents the effective vertical wave number; represents taking the phase of a complex number; represents the surface phase; represents multiplied by the conjugate of . The i in the formula is the expression of the imaginary part.

[0011] Furthermore, step S5 specifically includes: Within the unit circle of the complex plane, find the sub - view interference complex coherence coefficient farthest from the surface phase point by means of traversal search. The expression is: ; In the formula, represents the th sub - view interference complex coherence coefficient; The sub - view interference complex coherence coefficient is considered to be the sub - view interference complex coherence coefficient with the least surface scattering. At this time, an initial forest height is obtained by the method of ignoring the ground - object amplitude ratio. The expression is: ; In the formula, represents the surface phase, Indicates The theoretical coherence of the corresponding model Indicates the corresponding terrain amplitude ratio, which can be understood as the ratio of ground scattering energy to volume scattering energy; Indicates the forest height; Indicates the extinction coefficient; Indicates the local effective vertical wavenumber; Indicates the local incident angle; Solving the above equation gives the initial forest height.

[0012] Furthermore, step S6 specifically includes: Judging the sub - look interferometric complex coherence coefficient with the minimum distance containing surface scattering within the unit circle of the complex plane The farthest sub - look interferometric complex coherence coefficient , and the expression is as follows: ; In the formula, Indicates the th sub - look interferometric complex coherence coefficient; Within the unit circle of the complex plane, referring to the positions of the surface phase points and , since the distances from different sub - look interferometric complex coherence coefficients to the surface phase are different, the projection points on the coherence line can be considered to have different terrain amplitude ratios. The terrain amplitude ratio corresponding to the th sub - look interferometric complex coherence coefficient can be expressed as an expression that can achieve different terrain amplitude ratios corresponding to different sub - looks: ; In the formula, represents the distance from to the sub - look interferometric complex coherence coefficient with the minimum surface scattering , is the distance from to the surface phase ; Indicates the distance from to the actual pure volume scattering . By means of geometric relationships, the terrain amplitude ratio can be established as an expression of .

[0013] For each sub - look complex interferometric coherence coefficient , the following functional relationship is established based on the random terrain scattering two - layer model: ; In the formula, Indicates the surface phase, and are the local effective vertical wavenumber and the local incident angle, respectively; thus, for any sub - look interferometric complex coherence , there are three unknowns ; represents the forest height; represents the extinction coefficient; Combining the above two sub - look interferometric complex coherence coefficients and , based on the random volume scattering two - layer model, the forest height is inverted by establishing a non - linear iterative inversion framework, and the expression is: ; Use the non - linear iterative algorithm to solve the above formula. Take the initial forest height obtained in step S5 as the initial value and input it into the non - linear iterative algorithm for solution to obtain the final forest height.

[0014] On the second aspect, a device for inverting forest height based on single - baseline InSAR is provided, including: An interferometric data pre - processing module, which is used to obtain single - polarization single - baseline InSAR data of the observation area, perform interferometric processing to obtain interferometric complex coherence coefficients, and calculate the coherence compensation factor; A vertical wavenumber calculation and correction module, which is used to calculate the effective vertical wavenumber through baseline parameter information and perform local correction to obtain the local incident angle and the local effective vertical wavenumber; A sub - aperture decomposition module, which is used to perform azimuth sub - aperture decomposition on the master and slave images of the original InSAR data to obtain multiple pairs of sub - look SLC images; A surface phase extraction module, which performs interferometric processing on multiple pairs of sub - look SLC images to obtain multiple sub - look interferometric complex coherence coefficients. Based on the coherence compensation factor, the multiple sub - look interferometric complex coherence coefficients are corrected in coherence amplitude and then spread to the complex plane. Then, the multiple sub - look interferometric complex coherence coefficients are subjected to least - squares coherent line fitting, and the surface phase is obtained by combining the positive and negative judgment of the effective vertical wavenumber; An initial forest height estimation module, which is used to find the sub - look interferometric complex coherence coefficient that is farthest from the surface phase in the unit circle of the complex plane, regard it as the sub - look interferometric complex coherence coefficient with the least surface scattering, and solve for the initial forest height based on the random volume scattering two - layer model by ignoring the surface scattering; A forest height inversion module, which is used to find the sub - look interferometric complex coherence coefficient that is farthest from the sub - look interferometric complex coherence coefficient with the least surface scattering in the unit circle of the complex plane, combine the above two sub - look interferometric complex coherence coefficients, and invert the final forest height based on the random volume scattering two - layer model by establishing a non - linear iterative inversion framework.

[0015] The present invention discloses a method and device for forest height inversion based on single - baseline InSAR, and innovatively proposes a technical solution that can achieve forest height inversion only with single - baseline and single - polarization InSAR data. The present invention breaks through the dependence of the prior art on high - precision prior information or multi - source auxiliary data (such as high - precision DTM data or observation data from other platforms), constructs an innovative inversion framework by deeply mining the scattering characteristics of InSAR observation data itself, and significantly improves the inversion accuracy of the canopy height in forest areas. In terms of data requirements, the present invention realizes a self - contained technical breakthrough, and can complete high - precision forest parameter inversion only relying on single - baseline InSAR data without the need to rely on auxiliary observation data from other platforms. In terms of technical performance, the present invention shows obvious accuracy advantages compared with traditional methods, providing a new high - precision solution for forest resource monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0017] Figure 1 It is a flowchart of a method for forest height inversion based on single - baseline InSAR provided by an embodiment of the present invention; Figure 2 It is a schematic diagram of the sub - view interference complex coherence coefficient and coherence line fitting in the unit circle of the complex plane provided by an embodiment of the present invention; Figure 3 It is a forest height inversion result diagram provided by an embodiment of the present invention; among them, (a) represents the forest height inversion result obtained by the present invention, and (b) represents the airborne LiDAR CHM; Figure 4 It is a scatter plot for accuracy verification of the forest height inversion result obtained by an example of the present invention and the airborne LiDAR CHM. The color of the statistical scatter points represents the corresponding number of pixels. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be described in detail below. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

[0019] To better illustrate the technical solution of the present invention, taking a test area as an example, a scene of X-band bistatic InSAR data is selected, and airborne LiDAR DTM data is collected as the ground truth for verification, and the present invention is further described in detail; note that the specific implementation described herein is only used to explain the present invention and is not used to limit the present invention.

[0020] Among the data of the test area selected by the present invention, the airborne LiDAR DTM data has high precision and can be used as the ground truth to verify the results of the technical method of the present invention.

[0021] An embodiment of the present invention provides a method for inverting forest height based on single-baseline InSAR, and the process is as Figure 1 shown, including the following steps: S1: Obtain single-polarization single-baseline InSAR data of the observation area and perform interference processing to obtain an interference complex coherence coefficient, calculate a coherence compensation factor, and correct the coherence amplitude of the interference complex coherence coefficient.

[0022] Specifically, step S1 includes: S11: Obtain single-polarization single-baseline InSAR data of the observation area and perform interference processing: ; In the formula, and respectively represent the main image and the secondary image participating in the interference, represents conjugate multiplication of the main image and the secondary image, represents obtaining the expected value; the interference complex coherence coefficient is a complex number, expressed as , and are the interference phase and the coherence amplitude respectively, is the natural constant, and i is the expression of the imaginary part; the interference phase related to the terrain elevation contains the surface phase .

[0023] S12: Calculate the coherence compensation factor including the signal-to-noise ratio dephasing factor and the system quantization dephasing factor introduced during the data processing, and correct the coherence amplitude of the interference complex coherence coefficient. The expression for correcting the coherence amplitude of the observed interference complex coherence coefficient is .

[0024] It is mainly caused by the insufficient power ratio of signal to noise (SNR), manifested as a decrease in coherence, and can be corrected by filtering. For InSAR images, there is an equivalent zero signal-to-noise ratio (Noise Equivalent SigmaZero, NESZ) in both the range and azimuth directions, which is composed of polynomials. For the master and slave images, the signal-to-noise ratio on each polarization channel is calculated from the backscattering coefficient and the estimated NESZ value, and then the signal-to-noise ratio decorrelation factor is calculated by the following formula: ; In the formula, and represent the signal-to-noise ratios of the master image and the slave image respectively, and their expressions can both be: , ; In the formula, and represent the equivalent zero signal-to-noise ratios of the master image and the slave image respectively, and represent the backscattering coefficients of the master image and the slave image respectively.

[0025] System quantization decorrelation factor is a constant. Generally, = 0.98 can meet the application requirements of most radar images. Of course, in other embodiments, other constants close to 1 can also be taken, such as 0.95 - 0.99, etc.

[0026] S2: Calculate the effective vertical wavenumber through the baseline parameter information and perform local correction to obtain the local incident angle and the local effective vertical wavenumber.

[0027] Specifically, step S2 includes: S21: Obtain the nominal incident angle of the electromagnetic wave signal through the baseline parameters and geometric relationships during InSAR data observation, and calculate the effective vertical wavenumber ; In the formula, is the effective vertical wavenumber, which can characterize the altimetry sensitivity of the InSAR signal; represents the radar wave wavelength, and represent the slant range and the vertical baseline length respectively; when the InSAR system is a bistatic data system, when the InSAR system is a repeat-pass data system, ; S22: Calculate the local slope using an external DEM product , and its expression is: ; In the formula, is the local slope at pixel ; is the elevation value of the DEM product at pixel ; is the pixel resolution in the range direction; S23: After calculating the local slope , perform local correction on the nominal incident angle to obtain the local incident angle : ; S24: Based on the following formula, perform local correction on the effective vertical wave number to obtain the local effective vertical wave number : .

[0028] S3: Perform azimuth sub-aperture decomposition on the master and slave images of the original InSAR data to obtain multiple pairs of sub-view SLC images.

[0029] In this embodiment, performing azimuth sub-aperture decomposition on the master and slave images of the original InSAR specifically includes: performing azimuth fast Fourier transform on the master image and the slave image of the original InSAR data respectively to convert them to the frequency domain; in the frequency domain, based on a preset sub-aperture segmentation strategy, calculate the azimuth spectral center and spectral bandwidth of the image after sub-aperture decomposition; after windowing and filtering each sub-aperture spectrum, perform two-dimensional inverse fast Fourier transform to reconstruct multiple sub-view images of the master image and the slave image respectively; if finally obtaining sub-view master images and sub-view slave images, then sub-view SLC image interference pairs can be generated through pairwise pairing.

[0030] The number and resolution of the segmented sub-view SLC images are determined by the segmentation bandwidth, and the azimuth spectral centers of the segmented sub-view SLC images are respectively: , ; In the formula, represents the azimuth spectral center of the d-th sub-view SLC image; N represents the number of sub-view SAR images segmented along the azimuth direction; represents the segmentation bandwidth. In this example, =5 is selected, and the segmentation bandwidth is = 0.5, and the center frequencies of the bandwidth spectrum are: -0.250, -0.125, 0, 0.125, 0.250.

[0031] S4: For multiple sub - view interference complex coherence coefficients obtained by performing interference processing on multiple pairs of sub - view SLC images, after performing coherent amplitude correction on the multiple sub - view interference complex coherence coefficients, they are spread points into the complex plane (complex number plane), and then the least - squares coherent straight - line fitting is performed on the multiple sub - view interference complex coherence coefficients. The surface phase is obtained by combining the positive and negative judgments of the effective vertical wave number.

[0032] Specifically, step S4 includes: S41: After obtaining multiple sub - view interference complex coherence coefficients by performing interference processing on multiple pairs of sub - view SLC images, use the signal - to - noise ratio de - coherence factor and the system quantization de - coherence factor to perform coherent amplitude correction on the multiple sub - view interference complex coherence coefficients; S42: Spread the multiple sub - view interference complex coherence coefficients after the above - mentioned coherent amplitude correction into the complex plane. For any sub - view interference complex coherence coefficient, the phase range is 0 - , and the amplitude range is 0 - 1; S43: Use the total least - squares constraint to perform coherent straight - line fitting on the multiple sub - view interference complex coherence coefficients within the unit circle of the complex plane, and obtain the phases and of the two intersection points of the coherent straight - line and the unit circle of the complex plane. Figure 2 The figure shows the schematic diagram of the sub - view interference complex coherence coefficients within the unit circle of the complex plane and the coherent straight - line fitting; one of them is determined as the surface phase point by judging the size relationship between the positive and negative values of the effective vertical wave number and the phases of the two obtained intersection points. The judgment criterion is: ; Specifically, it can be interpreted as: when > 0, if , then is , if , then is ; when 0, if , then is , if , then is ; where, represents the effective vertical wave number; represents taking the phase of a complex number; and respectively represent the phases corresponding to the two intersection points; represents the surface phase; denote and conjugate multiply, where i in the formula is the expression of the imaginary part.

[0033] S5: In the unit circle of the complex plane, find the sub - view interferometric complex coherence coefficient that is farthest from the surface phase, and regard it as the sub - view interferometric complex coherence coefficient with the least surface scattering. Based on the random volume scattering two - layer model and ignoring the surface scattering, solve to obtain the initial forest height.

[0034] Specifically, step S5 includes: S51: In the unit circle of the complex plane, by means of traversal search, find the sub - view interferometric complex coherence coefficient that is farthest from the surface phase point The expression is: ; ; In the formula, denotes the th sub - view interferometric complex coherence coefficient; S52: The sub - view interferometric complex coherence coefficient found above that is farthest from the surface phase point is considered to be the sub - view interferometric complex coherence coefficient with the least surface scattering. At this time, use the method of ignoring the volume amplitude ratio to obtain an initial forest height. The process of obtaining the initial forest height is to solve the problem shown in the following expression: ; ; In the formula, denotes the surface phase information, denotes the theoretical coherence of the corresponding model, denotes the corresponding volume amplitude ratio, which can be understood as the ratio of ground scattering energy to volume scattering energy; denotes the forest height; denotes the extinction coefficient; denotes the local effective vertical wavenumber; denotes the local incident angle; The model corresponding to solving the above formula is the random volume scattering two - layer model, and the expression of the model is: ; In the formula, denotes the surface phase, denotes the pure volume decoherence, denotes the corresponding volume amplitude ratio, which can be understood as the ratio of ground scattering energy to volume scattering energy; For any sub - view interference pair, the pure volume decoherence can be expressed as a function of the forest height and the extinction coefficient .

[0035] S6: In the unit circle of the complex plane, find the sub - view interferometric complex coherence coefficient that is farthest from the sub - view interferometric complex coherence coefficient with the least surface scattering. Combine the above two sub - view interferometric complex coherence coefficients, and based on the random volume scattering two - layer model, establish a non - linear iterative inversion framework to invert the final forest height.

[0036] Specifically, step S6 includes: S61: In the unit circle of the complex plane, judge the sub - view interferometric complex coherence coefficient that is farthest from the sub - view interferometric complex coherence coefficient with the least surface scattering through geometric relationships The farthest sub - view interferometric complex coherence coefficient , and the expression is as follows: ; In the formula, represents the th sub - view interferometric complex coherence coefficient; S62: In the unit circle of the complex plane, referring to the positions of the surface phase and , since the distances from different sub - view interferometric complex coherence coefficients to the surface phase are different, the projection points of them on the coherence line can be considered to have different volume amplitude ratios. The volume amplitude ratio corresponding to the th sub - view interferometric complex coherence coefficient can be expressed as an expression that can realize the volume amplitude ratios corresponding to different sub - views: ; In the formula, represents 's distance to the sub - view interferometric complex coherence coefficient with the least surface scattering , is 's distance to the surface phase , and both of these distances can be calculated according to the above - mentioned geometric relationships; represents 's distance to the actual pure volume scattering . By means of geometric relationships, the volume amplitude ratio can be established as an expression.

[0037] S63: For each sub - view interferometric complex coherence coefficient , based on the random volume scattering two - layer model, establish the following functional relationship: ; In the formula, represents the surface phase, and are the local effective vertical wavenumber and the local incident angle respectively; thus, for any sub - look interferometric complex coherence , corresponding to three unknowns ; S64: Combine the above two sub - look interferometric complex coherence coefficients and , and based on the random volume scattering two - layer model, establish a non - linear iterative inversion framework to invert the forest height. The expression is: ; To solve the above formula, use a non - linear iterative algorithm to solve it. Take the initial forest height obtained in step S5 as the initial value and input it into the non - linear iterative algorithm for solution to obtain the final forest height.

[0038] In this example, the finally obtained forest height is as Figure 3 shown, Figure 3 in which (a) is the forest height inversion result obtained by the present invention, Figure 3 in which (b) is the airborne LiDAR canopy height model (Conapy Height Model, CHM, that is, the forest height). Then, use the LiDAR CHM as the true value for accuracy evaluation. The evaluation indicators include the following: (coefficient of determination), RMSE (root mean square error). The statistical results of the accuracy indicators are as Figure 4 shown. is 0.77, indicating a high correlation. The RMSE is 2.65 m, which proves that the technical method of the present invention is effective and can improve the forest height inversion accuracy of single - baseline InSAR.

[0039] The embodiment of the present invention also provides a device for inverting forest height based on single - baseline InSAR, including: An interferometric data pre - processing module, which acquires single - polarization single - baseline InSAR data of the observation area, performs interferometric processing to obtain interferometric complex coherence coefficients and corresponding coherence amplitudes, calculates a coherence compensation factor, and corrects the coherence amplitude; A vertical wavenumber calculation and correction module, which is used to calculate the effective vertical wavenumber through baseline parameter information and perform local correction to obtain the local incident angle and the local effective vertical wavenumber; A sub - aperture decomposition module, which is used to perform azimuth - direction sub - aperture decomposition on the master and slave images of the original InSAR data to obtain multiple pairs of sub - look SLC images; The surface phase extraction module performs interference processing on multiple pairs of sub - viewed SLC images to obtain multiple sub - viewed interference complex coherence coefficients. After performing coherent amplitude correction on the multiple sub - viewed interference complex coherence coefficients, they are spread into the complex plane. Then, the least - squares coherent line fitting is performed on the multiple sub - viewed interference complex coherence coefficients, and the surface phase is obtained by combining the positive and negative judgments of the effective vertical wavenumber. The initial forest height estimation module is used to find the sub - viewed interference complex coherence coefficient that is farthest from the surface phase within the unit circle of the complex plane, regard it as the sub - viewed interference complex coherence coefficient with the least surface scattering, and solve for the initial forest height based on the random ground - body scattering two - layer model by neglecting the surface scattering. The forest height inversion module is used to find the sub - viewed interference complex coherence coefficient that is farthest from the sub - viewed interference complex coherence coefficient with the least surface scattering within the unit circle of the complex plane, combine the above two sub - viewed interference complex coherence coefficients, and inversely obtain the final forest height based on the random ground - body scattering two - layer model by establishing a non - linear iterative inversion framework.

[0040] It should be understood that the functional unit modules in each embodiment of the present invention can be concentrated in one processing unit, or each unit module can exist physically alone, or two or more unit modules can be integrated into one unit module, and can be implemented in the form of hardware or software.

[0041] It can be understood that the same or similar parts in the above embodiments can be referred to each other, and the content not detailed in some embodiments can be referred to the same or similar content in other embodiments.

[0042] Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for inverting forest height based on single - baseline InSAR, characterized in that, It includes the following steps: S1: Obtain the single-polarization single-baseline InSAR data of the observation area, perform interference processing to obtain the interference complex coherence coefficient, and calculate the coherence compensation factor; S2: Calculate the effective vertical wave number through the baseline parameter information, and perform local correction to obtain the local incident angle and the local effective vertical wave number; S3: Decompose the master and slave images of the original InSAR data in the azimuth sub-aperture to obtain multiple pairs of sub-view SLC images; S4: For the multiple sub-view interference complex coherence coefficients obtained by performing interference processing on the multiple pairs of sub-view SLC images, perform coherence amplitude correction on the multiple sub-view interference complex coherence coefficients based on the coherence compensation factor, then spread the points into the complex plane, and then perform least squares coherent straight line fitting on the multiple sub-view interference complex coherence coefficients, and combine the positive and negative judgment of the effective vertical wave number to obtain the surface phase; S5: In the unit circle of the complex plane, find the sub-view interference complex coherence coefficient that is farthest from the surface phase, regard it as the sub-view interference complex coherence coefficient with the least surface scattering, and solve the initial forest height based on the random ground scatter two-layer model and ignoring the surface scattering; S6: In the unit circle of the complex plane, find the sub-view interference complex coherence coefficient that is farthest from the sub-view interference complex coherence coefficient with the least surface scattering, combine the above two sub-view interference complex coherence coefficients, and invert the final forest height based on the random ground scatter two-layer model by establishing a non-linear iterative inversion framework; 2. The method for inverting forest height based on single-baseline InSAR according to claim 1, wherein Step S1 specifically includes: Obtain the single-polarization single-baseline InSAR data of the observation area and perform interference processing: ； In the formula, and respectively represent the main image and the secondary image participating in the interference, represents conjugate multiplication of the main image and the secondary image, represents obtaining the expected value; the interference complex coherence coefficient is a complex number and is expressed as , and are the interference phase and the coherence amplitude respectively; Calculate the coherence compensation factor including the signal-to-noise ratio decoherence factor and the system quantization decoherence factor ​ 3. The method for inverting forest height based on single-baseline InSAR according to claim 1, wherein Step S2 specifically includes: Obtain the nominal incident angle of the electromagnetic wave signal based on the baseline parameters and geometric relationships during InSAR data observation and calculate the effective vertical wave number The expression is as follows: ； wherein, is the effective vertical wave number, represents the radar wave wavelength, and respectively represent the slant range and the vertical baseline length; when the InSAR system is bistatic data, , when the InSAR system is repeat-pass data, ; Calculating local slope using external DEM products and performing local correction on the nominal incident angle to obtain the local incident angle : ； Based on the following formula, the effective vertical wave number is locally corrected to obtain the local effective vertical wave number as follows: 。 4. The method for inverting forest height based on single-baseline InSAR according to claim 1, wherein In step S3, azimuth sub-aperture decomposition is performed on the master and slave images of the original InSAR data, specifically: performing azimuth fast Fourier transform on the master image and the slave image of the original InSAR data respectively to convert them to the frequency domain; in the frequency domain, based on a preset sub-aperture segmentation strategy, calculating the azimuth spectral center of the image after sub-aperture decomposition and the spectral bandwidth ; after windowing and filtering each sub-aperture spectrum, performing inverse two-dimensional fast Fourier transform to reconstruct multiple sub-view images of the master image and the slave image respectively; if finally obtaining sub-view master images and sub-view slave images, then through pairwise pairing, sub-view SLC image interference pairs can be generated.

5. The method for inverting forest height based on single-baseline InSAR according to claim 1, characterized in that Step S4 specifically includes: After obtaining multiple sub - view interference complex coherence coefficients by performing interference processing on multiple pairs of sub - view SLC images, use the signal - to - noise ratio decoherence factor and the system quantization decoherence factor to perform coherent amplitude correction on the multiple sub - view interference complex coherence coefficients; Spread the multiple sub-view interference complex coherence coefficients after the above coherence amplitude correction onto the complex plane; Using the total least squares constraint, perform coherent straight line fitting on multiple sub - view interference complex coherence coefficients within the unit circle of the complex plane to obtain the phases of the two intersection points of the coherent straight line and the unit circle of the complex plane. and , and determine one of them as the surface phase point based on the magnitude relationship between the positive and negative values of the effective vertical wave number and the phases of the two intersection points obtained above. The judgment criterion is: when > 0, if , then is , if , then is ; when < 0, if , then is , if , then is ; where represents the effective vertical wave number; represents taking the phase of a complex number; represents the surface phase; represents multiplied by the conjugate of , and i in the formula is the expression of the imaginary part.

6. The method for inverting forest height based on single-baseline InSAR according to claim 1, wherein Step S5 specifically includes: Within the unit circle of the complex plane, by means of traversal search, find the sub-aperture interferometric complex coherence coefficient that is farthest from the surface phase point , and the expression is: ； In the formula, represents the th sub-view interference complex coherence coefficient; represents the surface phase; Sub - view interferometric complex coherence coefficient It is considered as the sub - view interferometric complex coherence coefficient with the least surface scattering. At this time, an initial forest height is obtained by the method of ignoring the terrain amplitude ratio, and the expression is as follows: ； In the formula, represents the corresponding model theoretical coherence; represents the corresponding terrane amplitude ratio; represents the forest height; represents the extinction coefficient; represents the local effective vertical wave number; represents the local incident angle; Solve the above formula to obtain the initial forest height.

7. The method for inverting forest height based on single-baseline InSAR according to claim 1, wherein Step S6 specifically includes: Judging the sub - view interferometric complex coherence coefficient with the minimum ground surface scattering distance through geometric relationship within the unit circle of the complex plane The farthest sub - view interferometric complex coherence coefficient , and the expression is as follows: ； In the formula, represents the th sub-view interference complex coherence coefficient; In the unit circle of the complex plane, the reference surface phase and Since the distances from the complex coherence coefficients of different sub-view interference to the surface phase are different, their projection points on the coherence line can be considered to have different ground amplitude ratios. Complex coherence coefficient of sub-view interference The corresponding ground amplitude ratio It can be expressed as an expression that can realize the ground body amplitude ratio corresponding to different sub-views: ； In the formula, represents the distance to the sub - view interference complex coherence coefficient with the minimum surface scattering ; is the distance to the surface phase point ; represents the distance to the actual pure volume scattering . By means of geometric relationships, the ground object amplitude ratio can be established as an expression. For each sub-view interference complex coherence coefficient the following functional relationship is established based on the randomly stratified scattering two-layer model: ； In the formula, represents the surface phase, and are the local effective vertical wave number and the local incident angle respectively; thus, for any sub - view interference complex coherence , there are three corresponding unknowns ; represents the forest height; represents the extinction coefficient; Combining the above two sub-view interference complex coherence coefficients and , based on the random volume scattering two-layer model, a non-linear iterative inversion framework is established to invert the forest height, and the expression is as follows: ； Use a non-linear iterative algorithm to solve the above formula, take the initial forest height obtained in step S5 as the initial value, input it into the non-linear iterative algorithm for solution, and obtain the final forest height.

8. An apparatus for retrieving forest height based on single-baseline InSAR, characterized in that, It includes: An interference data preprocessing module, which is used to obtain the single-polarization single-baseline InSAR data of the observation area, perform interference processing to obtain the interference complex coherence coefficient, and calculate the coherence compensation factor; A vertical wave number calculation and correction module, which is used to calculate the effective vertical wave number through the baseline parameter information, and perform local correction to obtain the local incident angle and the local effective vertical wave number; A sub-aperture decomposition module, which is used to decompose the master and slave images of the original InSAR data in the azimuth sub-aperture to obtain multiple pairs of sub-view SLC images; A surface phase extraction module, for the multiple sub-view interference complex coherence coefficients obtained by performing interference processing on the multiple pairs of sub-view SLC images, perform coherence amplitude correction on the multiple sub-view interference complex coherence coefficients based on the coherence compensation factor, then spread the points into the complex plane, and then perform least squares coherent straight line fitting on the multiple sub-view interference complex coherence coefficients, and combine the positive and negative judgment of the effective vertical wave number to obtain the surface phase; An initial forest height estimation module is used to find the sub - view interferometric complex coherence coefficient that is farthest from the surface phase within the unit circle of the complex plane, regard it as the sub - view interferometric complex coherence coefficient with the least surface scattering, and obtain the initial forest height by solving based on the random volume scattering two - layer model and ignoring the surface scattering; A forest height inversion module is used to find the sub - view interferometric complex coherence coefficient that is farthest from the sub - view interferometric complex coherence coefficient with the least surface scattering within the unit circle of the complex plane, combine the above two sub - view interferometric complex coherence coefficients, and inversely obtain the final forest height based on the random volume scattering two - layer model by establishing a non - linear iterative inversion framework.

Images