River ice thickness inversion method and system based on multiple polarization parameters and texture information

Abstract

This invention provides a method and system for inverting river ice thickness based on multiple polarization parameters and texture information. The inversion method includes: acquiring polarization data from a synthetic aperture radar and measured information of a preset area; determining first processed image information and first feature information including polarization parameter information and texture information based on the polarization data; determining pseudo-color image information and establishing a correspondence between the pseudo-color image information and the measured information based on the first processed image information; determining river ice classification information using the pseudo-color image information; determining the average river ice thickness information of the same river ice classification information within the preset area based on the measured information, the first processed image information, and the first feature information; determining an empirical model for river ice thickness based on the average river ice thickness information and the first feature information; and inverting the river ice thickness based on the empirical model and the first feature information. This method enables the inversion of river ice thickness corresponding to different river ice types using multiple parameters, improving the accuracy of the inversion.

Description

Technical Field

[0001] This invention relates to the field of microwave remote sensing technology, and in particular to a method and system for inverting river ice thickness based on multipolarization parameters and texture information. Background Technology

[0002] Polarimetric synthetic aperture radar (PSAR) possesses all-weather, all-terrain detection capabilities. However, current ice accumulation observation methods are limited by factors such as visual distance, observation location, traffic conditions, and weather conditions, resulting in insufficient accuracy and range. Furthermore, ice accumulation thickness is typically retrieved using a single parameter, the backscattering coefficient, leading to poor accuracy. Therefore, a multi-parameter thermal ice thickness retrieval method based on PSAR data is of significant importance.

[0003] Currently, the inversion of ice thickness generally involves performing a uniform inversion for all types of ice, using only a single parameter, the backscattering coefficient. Yellow River ice is generally classified into dense ice, subsurface ice, and thermal ice, with significant differences in thickness among these types. Performing a uniform inversion for all types of ice leads to substantial errors. Furthermore, while using a single parameter, the backscattering coefficient, ensures good inversion results for all ice types, combining it with other parameters can improve the accuracy for specific ice types. Therefore, performing a uniform inversion for all ice types using only a single parameter limits the precision of ice thickness inversion.

[0004] Simultaneous inversion of all ice types using only the backscattering coefficient as a single parameter allows for a certain estimation of ice thickness, simplifying the process and reducing complexity. However, in practical applications, different ice types are highly sensitive to different parameters, leading to low accuracy when inverting based on a single parameter. Furthermore, performing inversion on all ice types together ignores the significant differences in thickness between them.

[0005] In summary, existing methods for inverting multiple types of ice based on a single parameter result in low accuracy. In particular, using the same parameter for all types of ice leads to poor inversion precision, making it difficult to quantitatively analyze ice thickness and meet the needs of ice storm disaster monitoring. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for inverting river ice thickness based on multipolarization parameters and texture information, in order to solve the problem of low accuracy caused by using a single parameter to invert multiple types of ice at the same time in the prior art.

[0007] The embodiments of the present invention employ the following technical solution: a river ice thickness inversion method based on multipolarization parameters and texture information, characterized in that it includes:

[0008] Acquire polarization data from synthetic aperture radar and measured information of a preset area, wherein the measured information includes geographical information and river channel information, and the river channel information includes river ice classification information and river ice thickness information corresponding to the river ice classification.

[0009] Based on the polarization data, first feature information and first processed image information are obtained, wherein the first feature information includes polarization parameter information and texture information, and the first processed image information includes the geographic information of the preset region;

[0010] Based on the first processed image information, pseudo-color image information is determined and a correspondence between the pseudo-color image information and the measured information is established.

[0011] The pseudo-color image information is used to determine the classification information of the river ice;

[0012] Based on measured information, the first processed image information, and the first feature information, the average river ice thickness information with the same river ice classification information within the preset area is determined.

[0013] An empirical model for determining river ice thickness is based on the average river ice thickness information and the first feature information.

[0014] Based on the empirical model and the first feature information, the river ice thickness is inverted using the polarization data of the synthetic aperture radar.

[0015] In some embodiments, the determination of first feature information and first processed image information based on the polarization data, wherein the first feature information includes polarization parameter information and texture information, and the first processed image information includes geographic information of the preset region, including:

[0016] Spatial filtering or multi-view processing is performed on the polarization data to obtain first image information;

[0017] The first image information is subjected to distance Doppler terrain correction to determine the second image information, which includes geographic information;

[0018] The first processed image information is determined based on the second image information and the preset region;

[0019] The first feature information is determined based on the first processed image information, and the first feature information includes polarization parameter information and texture information.

[0020] In some embodiments, determining river ice classification information using the pseudo-color image information includes:

[0021] Based on the pseudo-color image information, a first pseudo-color image information with a preset amount of information is determined, wherein the first pseudo-color image information includes river information;

[0022] A training set for river ice classification is determined based on the river ice classification information in the first pseudo-color image information;

[0023] River ice classification information is determined based on the first pseudo-color image information, the training set, and the supervised classification method.

[0024] In some embodiments, the supervised classification method employs Wissaud distribution supervised classification.

[0025] In some embodiments, the use of the Wissaud distribution-supervised classification includes:

[0026] The computational information for each river ice type is determined based on the training set. The computational information includes: the covariance matrix, the average covariance matrix, and the trace of the matrix of the pixels to be classified in the first pseudo-color image.

[0027] In some embodiments, when a pixel to be classified satisfies that its Wissaud distance metric is greater than the Wissaud distance metric of a pixel in the corresponding river ice type, the pixel to be classified is assigned to that river ice type.

[0028] In some embodiments, the measured information includes river information of multiple river sections at locations within a preset area divided according to a preset division method.

[0029] In some embodiments, the texture information is extracted from a gray-level co-occurrence matrix, wherein the gray-level co-occurrence matrix reflects the gray-level spatial correlation in the pseudo-color image by statistically analyzing the conditional probability density function of the pseudo-color image, and the texture information includes the mean value characterizing the regularity of the image texture, the variance characterizing the uniformity of the image gray-level, and the second moment of the angle characterizing the uniformity of the image gray-level distribution and the coarseness of the texture.

[0030] In some embodiments, an empirical model for determining river ice thickness based on the average river ice thickness information and the first feature information includes:

[0031] River ice should be classified into at least two categories: thermal ice and mixed ice.

[0032] For the thermal ice, a linear equation is constructed between the thickness of the thermal ice and the backscattering coefficient and polarization angle in the polarization parameter information:

[0033] For the mixed ice, an empirical model is constructed that relates the thickness of the mixed ice to the mean of the texture information, the second moment of the angular velocity in the texture information, and the polarization angle in the polarization parameter information.

[0034] In some embodiments, the process of retrieving river ice thickness from the polarization data of the synthetic aperture radar based on the empirical model and the first feature information includes:

[0035] The thickness of the water and non-river areas within the preset area is assigned a value of 0.

[0036] For the thermal ice region, the backscattering coefficient and polarization angle corresponding to the first feature information are substituted into the linear equation to solve for the river ice thickness;

[0037] For the mixed ice region, the polarization angle corresponding to the first feature information, the mean value of the texture information, and the second moment of the angle in the texture information are substituted into the empirical model to solve for the river ice thickness.

[0038] This invention also discloses a river ice thickness inversion system, characterized in that it includes:

[0039] The first data acquisition module is configured to acquire polarization data from the synthetic aperture radar and measured information from a preset area.

[0040] The first data processing module is configured to determine first feature information and first processed image information based on the polarization data;

[0041] The second data processing module is configured to determine pseudo-color image information and establish a correspondence between the pseudo-color image information and the measured information based on the first processed image information.

[0042] The third data processing module is configured to use the pseudo-color image information to determine river ice classification information;

[0043] The data processing module is configured to determine the average ice thickness information of the same ice classification information within a preset area based on measured information, the first processed image information and the first feature information.

[0044] The determination module is configured to use an empirical model to determine the ice thickness based on the average ice thickness information and the first feature information.

[0045] The inversion module is configured to perform ice thickness inversion on the polarization data of the synthetic aperture radar based on the empirical model and the first feature information.

[0046] The beneficial effects of the embodiments of the present invention are as follows:

[0047] The polarization data of synthetic aperture radar was processed to determine the first processed image information and the first feature information, including polarization parameter information and texture information. An empirical model of river ice thickness was constructed using the first feature information, the first processed image information and the measured information in the preset area. The river ice thickness was inverted using the empirical model and the first feature information based on the polarization data of synthetic aperture radar. This enabled the inversion of river ice thickness corresponding to different river ice types with the participation of multiple parameters, thus improving the accuracy of the inversion. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments of the present invention or related technologies, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 This is a structural block diagram of the river ice thickness inversion method of the present invention.

[0050] Figure 2 This is a schematic diagram of the river ice thickness inversion results of the present invention.

[0051] Figure 3 This is a schematic diagram illustrating the error in the river ice thickness inversion results of this invention.

[0052] Figure 4 This is a structural block diagram of the inversion system of the present invention. Detailed Implementation

[0053] Various aspects and features of the present invention are described herein with reference to the accompanying drawings.

[0054] It should be understood that various modifications can be made to the embodiments described herein. Therefore, the above description should not be considered as limiting, but merely as an example of embodiments. Other modifications within the scope and spirit of the invention will be apparent to those skilled in the art.

[0055] The accompanying drawings, which are included in and form part of this specification, illustrate embodiments of the invention and, together with the general description of the invention given above and the detailed description of the embodiments given below, serve to explain the principles of the invention.

[0056] These and other features of the invention will become apparent from the following description of preferred forms of embodiments given as non-limiting examples, with reference to the accompanying drawings.

[0057] It should also be understood that although the invention has been described with reference to some specific examples, those skilled in the art can certainly implement many other equivalent forms of the invention, which have the features of the above-described "Summary of the Invention" and are therefore all within the scope of protection defined herein.

[0058] The above and other aspects, features and advantages of the invention will become more apparent when taken in conjunction with the accompanying drawings and in view of the following detailed description.

[0059] Specific embodiments of the invention are described below with reference to the accompanying drawings; however, it should be understood that the claimed embodiments are merely examples of the invention, which can be implemented in various ways. Well-known and / or repeated functions and structures are not described in detail to avoid unnecessary or redundant details that could obscure the invention. Therefore, the specific structural and functional details claimed herein are not intended to be limiting, but merely serve as the basis and representative basis for the foregoing "Summary of the Invention" to teach those skilled in the art to use the invention in various ways with substantially any suitable detailed structure.

[0060] This specification may use the phrases “in one embodiment,” “in another embodiment,” “in yet another embodiment,” or “in other embodiments,” all of which may refer to one or more of the same or different embodiments of the present invention.

[0061] To address the problems in the background art, this invention discloses a method for inverting river ice thickness based on multipolarization parameters and texture information.

[0062] The method includes:

[0063] Step S100: Obtain polarization data from synthetic aperture radar and measured information of a preset area. The measured information includes geographic information and river channel information. The river channel information includes river ice classification information and river ice thickness information corresponding to the river ice classification.

[0064] Synthetic Aperture Radar (SAR) is a high-resolution imaging radar that can obtain high-resolution radar images similar to optical photography under extremely low visibility weather conditions. Fully polarimetric SAR requires the simultaneous transmission of H (horizontal polarization) and V (vertical polarization), resulting in four polarization modes: (HH), (HV), (VV), and (VH).

[0065] The preset area refers to the area corresponding to the actual location that needs to be measured, defined using the Earth's latitude and longitude.

[0066] Measured information refers to information obtained at an actual location through methods not limited to measurement. Measured information includes geographic information and river channel information. Geographic information includes latitude and longitude coordinates, while river channel information includes ice classification information and the corresponding ice thickness information. Ice thickness information can be obtained by drilling into the ice and then measuring the thickness of the ice cross-section.

[0067] Step S101: Based on polarization data, determine first feature information and first processed image information, wherein the first feature information includes polarization parameter information and texture information, and the first processed image information includes geographic information of a preset area.

[0068] This involves preprocessing the polarization data and extracting the first feature information and the first processed image information. Because SAR images inherently possess salt-and-pepper speckle noise, which reduces image quality and makes feature extraction more difficult, speckle is caused by a random combination of dephased coherent echoes scattered by the basic scatterer within each resolution cell. Therefore, the polarization data can be denoised.

[0069] In some embodiments, based on the polarization data, first feature information and first processed image information are determined, wherein the first feature information includes polarization parameter information and texture information, and the first processed image information includes geographic information of the preset region, including:

[0070] Step S101-1: Spatial filtering or multi-view processing is performed on the polarization data to reduce noise and obtain first image information. Specifically, a Refinedlee filter can be used for speckle filtering, and the window size can be 7x7.

[0071] Step S101-2: Perform distance Doppler terrain correction on the first image information to determine the second image information, which includes geographic information.

[0072] Due to factors such as changes in the scene's terrain, the distance information in the first image is distorted, resulting in some image data loss. Terrain correction can effectively reduce distortion and compensate for the distortion, making the ensemble representation of the images as close to the true information as possible.

[0073] Range-Doppler topographic correction implements the range-Doppler orthogonalization method for geocoding images corresponding to the first image information from a single 2D grid Greda geometry. It uses metadata (also known as intermediate data) and reference digital elevation model (DEM) data to derive accurate geographic location information; the default source of the DEM data is SRTM (topographic data).

[0074] Step S101-3: Determine the first processed image information based on the second image information and the preset area.

[0075] Polarization data needs to be combined with measured information for application. Often, the measured information only accounts for a small part of the polarization data, so the polarization data needs to be cropped to retain only the channel portion.

[0076] Step S101-4: Determine the first feature information based on the first processed image information. The first feature information includes polarization parameter information and texture information.

[0077] Polarization parameters can be extracted from the polarization data of the river channel. Polarization scattering entropy H, polarization angle α, and isopolarization ratio Cross-polarization ratio Characteristic parameters, where the subscripts H and V represent horizontal polarization and vertical polarization, for example, the subscript HV indicates that the radar transmitter is vertically polarized and the receiver is horizontally polarized.

[0078] Inversion of mixed ice, consisting of dense thick ice and water-borne ice, requires the use of texture information because the surfaces of these two types of ice are very rough, unlike the smooth surface of thermal ice. The surface of dense thick ice exhibits clear texture information due to compression, stacking, and deformation.

[0079] The gray-level co-occurrence matrix (GLCM) is a texture feature extraction method that reflects the correlation of gray-level spaces in an image by statistically analyzing the conditional probability density function of the image. The texture information used in this invention is as follows:

[0080] 1. Mean:

[0081]

[0082] Where G represents the size of the selected window, P ij This represents the pixel value at this point. The mean value characterizes the regularity of the image texture; when the image texture is uniform, the mean value is relatively large.

[0083] 2. Variance (VAR):

[0084]

[0085] Variance characterizes the uniformity of image gray levels; when the image gray level variation is small, the variance is small.

[0086] 3. Second moment of angle:

[0087]

[0088] The second moment of the angle, also known as energy, characterizes the uniformity of the gray-level distribution in an image and the fineness of the image texture. When the gray-level distribution of an image is relatively uniform, the value of the second moment of the angle is large, and the image exhibits fine texture features.

[0089] Window size is the most important factor affecting texture information effect. Considering factors such as ground feature details, discrimination and computation time, the window size in this invention can be selected as 11, the orientation angle can be selected as 45 degrees, the displacement can be selected as 5, and the gray quantization level can be selected as 64.

[0090] Step S102: Based on the first processed image information, determine the pseudo-color image information and establish the correspondence between the pseudo-color image information and the measured information.

[0091] The first processed image information is combined to form an RGB pseudo-color image, such as a Pauli image. The pseudo-color image can display more detailed information.

[0092] The measured information includes latitude and longitude information from geographic information. Polarization data, after preprocessing, also includes latitude and longitude information. Therefore, the corresponding pixels in the RGB pseudo-color image information are found based on the latitude and longitude information of the measured information. If the latitude and longitude information of the measured information is located at the center of two or four pixels in the RGB pseudo-color image information, then all corresponding pixels must be recorded. During parameter extraction, the polarization parameter information and texture model corresponding to the pixels must be extracted.

[0093] Step S103: Use pseudo-color image information to determine river ice classification information.

[0094] In some embodiments, determining river ice classification information using the pseudo-color image information includes:

[0095] Step S103-1: Determine the first pseudo-color image information with a preset amount of information based on the pseudo-color image information. The first pseudo-color image information includes river channel information.

[0096] Step S103-2: Determine the training set for river ice classification based on the river ice classification information in the first pseudo-color image information.

[0097] Step S103-3: Determine the river ice classification information based on the first pseudo-color image information, the training set, and the supervised classification method.

[0098] The supervised classification method described above employs the Wishart distribution. The advantage of the polarization covariance matrix lies in its Wishart distribution, making it highly suitable for classification applications.

[0099] In some embodiments, the use of the Wissaud distribution-supervised classification includes:

[0100] The computational information for each river ice type is determined based on the training set. The computational information includes: the covariance matrix, the average covariance matrix, and the trace of the matrix of the pixels to be classified in the first pseudo-color image.

[0101] Specifically, the method of classification using the Wishart distance metric is called the Wishart classifier. The Wishart distance metric is:

[0102]

[0103] Where, μ m The representative class is C, which is the covariance matrix of the pixels to be classified. m Let be the average covariance matrix, and Tr be the trace of the matrix. C m The inverse matrix.

[0104] When a pixel to be classified has a Wissaud distance metric value greater than the Wissaud distance metric value of a pixel in the corresponding river ice type, the pixel to be classified is assigned to that river ice type. That is:

[0105] d(C,μ m )<d(C,μ n ),

[0106] This indicates that this pixel (point) is related to μ m The pixel is assigned to the class that is most similar to the other pixels.

[0107] Step S104: Based on the measured information, the first processed image information, and the first feature information, determine the average river ice thickness information of the same river ice classification information within the preset area.

[0108] The measured information includes river information from multiple river sections within a preset area, divided according to a preset division method. During field measurements, the river in the measurement area can be divided into n river sections of equal length. For each river section, a certain number of points must be selected and the ice thickness measured. The selected points must contain information on all ice classifications within that river section. Ice classifications of the same type within the same river section have similar properties, so the thickness of ice classifications of the same type within the same section is averaged.

[0109] Step S105, based on the average river ice thickness information and the first feature information, determines an empirical model for river ice thickness; including:

[0110] River ice should be classified into at least two categories: thermal ice and mixed ice.

[0111] For thermal ice, within a certain thickness range, the thickness of thermal ice exhibits a linear relationship with the backscattering coefficient and also a linear relationship with the polarization angle α. Therefore, the constructed empirical model is a linear equation between the backscattering coefficient and the α angle.

[0112] For mixed ice, due to the formation principle of dense thick ice, dense thick ice that has undergone accumulation, compression and deformation often has a thicker thickness. Therefore, we select the mean of texture information, the second moment of the angle, and the α angle in the polarization parameter to construct an empirical model.

[0113] By utilizing measured information, hypothetical parameter information, and texture information, the relevant parameters of the empirical model are solved to obtain a complete empirical model.

[0114] Step S106 involves retrieving river ice thickness from the synthetic aperture radar polarization data based on an empirical model and the first feature information, including:

[0115] Step S106-1: Assign a thickness of 0 to the water and non-river areas within the preset area.

[0116] Step S106-2: For the thermal ice region, substitute the backscattering coefficient and polarization angle from the corresponding first feature information into the linear equation to solve for the river ice thickness.

[0117] Step S106-3: For the mixed ice region, the polarization angle, the mean value of the texture information, and the second moment of the angle in the texture information are substituted into the empirical model to solve for the river ice thickness.

[0118] Figure 2 The results of river ice thickness inversion are shown (the horizontal axis represents the number, and the vertical axis represents the river ice thickness value in centimeters). Figure 3 The results of the river ice thickness inversion error are shown (the horizontal axis represents the number, and the vertical axis represents the error value, which is the difference between the inversion result and the actual thickness, in centimeters). Figure 2 and Figure 3 As can be seen, the present invention has good inversion accuracy.

[0119] Polarimetric data from synthetic aperture radar (SAR) was processed to determine first processed image information and first feature information, including polarization parameters and texture information. An empirical model of river ice thickness was constructed using the first feature information, the first processed image information, and measured information within a preset area. River ice thickness was then inverted using the empirical model and the first feature information from the SAR polarimetric data. This enabled the inversion of river ice thickness for different river ice types using multiple parameters, including backscattering coefficient, polarization angle, and texture information, thus improving the accuracy of the inversion. This method has positive guiding and practical application value for monitoring ice ridge disasters.

[0120] This invention also discloses a river ice thickness inversion system, such as Figure 4 As shown, the inversion system includes:

[0121] The first data acquisition module is configured to acquire polarization data from the synthetic aperture radar and measured information from a preset area.

[0122] Synthetic Aperture Radar (SAR) is a high-resolution imaging radar that can obtain high-resolution radar images similar to optical photography under extremely low visibility weather conditions. Fully polarimetric SAR requires the simultaneous transmission of H (horizontal polarization) and V (vertical polarization), resulting in four polarization modes: (HH), (HV), (VV), and (VH).

[0123] The preset area refers to the area corresponding to the actual location that needs to be measured, defined using the Earth's latitude and longitude.

[0124] Measured information refers to information obtained at an actual location through methods not limited to measurement. Measured information includes geographic information and river channel information. Geographic information includes latitude and longitude coordinates, while river channel information includes ice classification information and the corresponding ice thickness information. Ice thickness information can be obtained by drilling into the ice and then measuring the thickness of the ice cross-section.

[0125] The first data processing module is configured to determine first feature information and first processed image information based on the polarization data.

[0126] This involves preprocessing the polarization data and extracting the first feature information and the first processed image information. Because SAR images inherently possess salt-and-pepper speckle noise, which reduces image quality and makes feature extraction more difficult, speckle is caused by a random combination of dephased coherent echoes scattered by the basic scatterer within each resolution cell. Therefore, the polarization data can be denoised.

[0127] The second data processing module is configured to determine pseudo-color image information and establish a correspondence between the pseudo-color image information and the measured information based on the first processed image information.

[0128] The first processed image information is combined to form an RGB pseudo-color image, such as a Pauli image. The pseudo-color image can display more detailed information.

[0129] The measured information includes latitude and longitude information from geographic information. Polarization data, after preprocessing, also includes latitude and longitude information. Therefore, the corresponding pixels in the RGB pseudo-color image information are found based on the latitude and longitude information of the measured information. If the latitude and longitude information of the measured information is located at the center of two or four pixels in the RGB pseudo-color image information, then all corresponding pixels must be recorded. During parameter extraction, the polarization parameter information and texture model corresponding to the pixels must be extracted.

[0130] The third data processing module is configured to use the pseudo-color image information to determine river ice classification information.

[0131] Based on the pseudo-color image information, a first pseudo-color image information with a preset amount of information is determined, the first pseudo-color image information including river channel information.

[0132] The training set for river ice classification is determined based on the river ice classification information in the first pseudo-color image information.

[0133] River ice classification information is determined based on the first pseudo-color image information, the training set, and the supervised classification method.

[0134] The supervised classification method employs the Wishart distribution. The advantage of the polarization covariance matrix lies in its Wishart distribution, making it highly suitable for classification applications.

[0135] The method employing the Wissat distribution-supervised classification includes:

[0136] The computational information for each river ice type is determined based on the training set. The computational information includes: the covariance matrix, the average covariance matrix, and the trace of the matrix of the pixels to be classified in the first pseudo-color image.

[0137] Specifically, the method of classification using the Wishart distance metric is called the Wishart classifier. The Wishart distance metric is:

[0138]

[0139] Where, μ m The representative class is C, which is the covariance matrix of the pixels to be classified. m Let be the average covariance matrix, and Tr be the trace of the matrix. C m The inverse matrix.

[0140] When a pixel to be classified has a Wissaud distance metric value greater than the Wissaud distance metric value of a pixel in the corresponding river ice type, the pixel to be classified is assigned to that river ice type. That is:

[0141] d(C,μ m )<d(C,μ n ),

[0142] This indicates that this pixel (point) is related to μ m The pixel is assigned to the class that is most similar to the other pixels.

[0143] The fourth data processing module is configured to determine the average ice thickness information of the same ice classification information within a preset area based on measured information, the first processed image information, and the first feature information.

[0144] The measured information includes river information from multiple river sections within a preset area, divided according to a preset division method. During field measurements, the river in the measurement area can be divided into n river sections of equal length. For each river section, a certain number of points must be selected and the ice thickness measured. The selected points must contain information on all ice classifications within that river section. Ice classifications of the same type within the same river section have similar properties, so the thickness of ice classifications of the same type within the same section is averaged.

[0145] The determination module is configured to determine an empirical model for the thickness of the river ice based on the average river ice thickness information and the first feature information.

[0146] The module is further configured as follows:

[0147] River ice should be classified into at least two categories: thermal ice and mixed ice.

[0148] For thermal ice, within a certain thickness range, the thickness of thermal ice exhibits a linear relationship with the backscattering coefficient and also a linear relationship with the polarization angle α. Therefore, the constructed empirical model is a linear equation between the backscattering coefficient and the α angle.

[0149] For mixed ice, due to the formation principle of dense thick ice, dense thick ice that has undergone accumulation, compression and deformation often has a thicker thickness. Therefore, we select the mean of texture information, the second moment of the angle, and the α angle in the polarization parameter to construct an empirical model.

[0150] By utilizing measured information, hypothetical parameter information, and texture information, the relevant parameters of the empirical model are solved to obtain a complete empirical model.

[0151] The inversion module is configured to perform ice thickness inversion on the polarization data of the synthetic aperture radar based on the empirical model and the first feature information.

[0152] The inversion module is further configured as follows:

[0153] The thickness of the water and non-river areas within the preset area is assigned a value of 0.

[0154] For the thermal ice region, the backscattering coefficient and polarization angle from the corresponding first feature information are substituted into the linear equation to solve for the ice thickness.

[0155] For mixed ice regions, the polarization angle, the mean value of the texture information, and the second moment of the angle in the texture information are substituted into the empirical model to solve for the river ice thickness.

[0156] In some embodiments, the first data processing module is further configured such that the first feature information includes polarization parameter information and texture information, and the first processed image information includes geographic information of a preset region.

[0157] Spatial filtering or multi-view processing is performed on the polarization data to obtain the first image information.

[0158] The first image information is subjected to distance Doppler terrain correction to determine the second image information, which includes geographic information.

[0159] Based on the second image information and the preset area, determine the first processed image information;

[0160] First feature information is determined based on the first processed image information, and the first feature information includes polarization parameter information and texture information.

[0161] In some embodiments, the third data processing module is further configured to:

[0162] Based on the pseudo-color image information, a first pseudo-color image information with a preset amount of information is determined, the first pseudo-color image information including river channel information.

[0163] The training set for river ice classification is determined based on the river ice classification information in the first pseudo-color image information.

[0164] River ice classification information is determined based on the first pseudo-color image information, the training set, and the supervised classification method.

[0165] In some embodiments, the third data processing module is further configured to use a Wissaud distribution-supervised classification method.

[0166] In some embodiments, the third data processing module is further configured to: determine computational information for each type of river ice based on the training set, the computational information including: the covariance matrix, the average covariance matrix, and the trace of the matrix of the pixels to be classified in the first pseudo-color image.

[0167] In some embodiments, the third data processing module is further configured to: when a pixel to be classified satisfies that its Wissaud distance metric value is greater than the Wissaud distance metric value of a pixel in the corresponding river ice type, classify the pixel to be classified into the corresponding river ice type.

[0168] In some embodiments, the first data acquisition module is further configured to include river information of multiple river sections at preset locations divided according to a preset division method within the preset area.

[0169] In some embodiments, the first data processing module is further configured such that: the texture information is extracted from the gray-level co-occurrence matrix, wherein the gray-level co-occurrence matrix reflects the gray-level spatial correlation in the pseudo-color image by statistically analyzing the conditional probability density function of the pseudo-color image, and the texture information includes the mean value characterizing the regularity of the image texture, the variance characterizing the uniformity of the image gray level, and the second moment of the angle characterizing the uniformity of the image gray level distribution and the coarseness of the texture.

[0170] In some embodiments, the determining module is further configured as follows:

[0171] River ice should be classified into at least two categories: thermal ice and mixed ice.

[0172] For the thermal ice, a linear equation is constructed between the thickness of the thermal ice and the backscattering coefficient and polarization angle in the polarization parameter information:

[0173] For the mixed ice, an empirical model is constructed that relates the thickness of the mixed ice to the mean of the texture information, the second moment of the angular velocity in the texture information, and the polarization angle in the polarization parameter information.

[0174] The foregoing has described in detail several embodiments of the present invention, but the present invention is not limited to these specific embodiments. Those skilled in the art can make various variations and modifications based on the concept of the present invention, and all such variations and modifications should fall within the scope of protection claimed by the present invention.

Claims

1. A method for inverting river ice thickness based on multipolarization parameters and texture information, characterized in that, include: Acquire polarization data from synthetic aperture radar and measured information of a preset area. The measured information includes geographical information and river channel information. The river channel information includes river ice classification information and river ice thickness information corresponding to the river ice classification. The river ice is at least divided into thermal ice and mixed ice. The mixed ice is a mixture of submerged ice and dense thick ice. Based on the polarization data, first feature information and first processed image information are determined, wherein the first feature information includes polarization parameter information and texture information extracted based on the gray-level co-occurrence matrix, the texture information being the mean and second moment of the angle, the polarization parameter information including the backscattering coefficient and the polarization angle α, and the first processed image information including the geographical information of the preset region; Based on the first processed image information, pseudo-color image information is determined and a one-to-one correspondence between the pseudo-color image information and the measured information is established. Using the pseudo-color image information, the Waishat distribution supervised classification method is employed to determine the river ice classification information; Based on measured information, the first processed image information, and the first feature information, the average river ice thickness information with the same river ice classification information within the preset area is determined. Based on the average river ice thickness information and the first feature information, specific empirical models for river ice thickness are constructed for thermal ice and mixed ice respectively. Specifically, a linear empirical model for backscattering coefficient and polarization angle α is constructed for thermal ice, and a nonlinear empirical model for polarization angle α, mean value of texture information and second moment of angle is constructed for mixed ice. Based on the proprietary empirical model of river ice thickness and the first feature information, the river ice thickness is inverted using the polarization data of the synthetic aperture radar.

2. The method according to claim 1, characterized in that, Based on the polarization data, first feature information and first processed image information are determined, wherein the first feature information includes polarization parameter information and texture information, and the first processed image information includes geographical information of the preset region, including: Spatial filtering or multi-view processing is performed on the polarization data to obtain first image information; The first image information is subjected to distance Doppler terrain correction to determine the second image information, which includes geographic information; The first processed image information is determined based on the second image information and the preset region; The first feature information is determined based on the first processed image information, and the first feature information includes polarization parameter information and texture information.

3. The method according to claim 1, characterized in that, The process of determining river ice classification information using the pseudo-color image information includes: Based on the pseudo-color image information, a first pseudo-color image information with a preset amount of information is determined, wherein the first pseudo-color image information includes river information; A training set for river ice classification is determined based on the river ice classification information in the first pseudo-color image information; River ice classification information is determined based on the first pseudo-color image information, the training set, and the supervised classification method.

4. The method according to claim 3, characterized in that, The method employing the Wissat distribution-supervised classification includes: The computational information for each river ice type is determined based on the training set. The computational information includes: the covariance matrix, the average covariance matrix, and the trace of the matrix of the pixels to be classified in the first pseudo-color image.

5. The method according to claim 4, characterized in that, When a pixel to be classified satisfies the condition that its Wissaud distance metric is greater than the Wissaud distance metric of the pixel in the corresponding river ice type, the pixel to be classified is assigned to that river ice type.

6. The method according to claim 1, characterized in that, The measured information includes river information of multiple river sections at locations within a preset area, divided according to a preset division method.

7. The method according to claim 2, characterized in that, The gray-level co-occurrence matrix reflects the gray-level spatial correlation in the pseudo-color image by statistically analyzing the conditional probability density function of the pseudo-color image. The texture information includes the mean value representing the regularity of the image texture, the variance representing the uniformity of the image gray-level, and the second moment of the angle representing the uniformity of the image gray-level distribution and the coarseness of the texture.

8. The method according to claim 7, characterized in that, Based on the average river ice thickness information and the first feature information, an empirical model for determining river ice thickness is established, including: For the thermal ice, a linear equation is constructed between the thickness of the thermal ice and the backscattering coefficient and polarization angle in the polarization parameter information: For the mixed ice, an empirical model is constructed that relates the thickness of the mixed ice to the mean of the texture information, the second moment of the angular velocity in the texture information, and the polarization angle in the polarization parameter information.

9. The method according to claim 8, characterized in that, The process of retrieving river ice thickness from the polarization data of the synthetic aperture radar based on the empirical model and the first feature information includes: The thickness of the water and non-river areas within the preset area is assigned a value of 0. For the thermal ice region, the backscattering coefficient and polarization angle corresponding to the first feature information are substituted into the linear equation to solve for the river ice thickness; For the mixed ice region, the polarization angle corresponding to the first feature information, the mean value of the texture information, and the second moment of the angle in the texture information are substituted into the empirical model to solve for the river ice thickness.

10. A river ice thickness inversion system, characterized in that, include: The first data acquisition module is configured to acquire polarization data from synthetic aperture radar and measured information of a preset area; wherein, the measured information includes geographical information and river channel information, the river channel information includes river ice classification information and river ice thickness information corresponding to the river ice classification; the river ice is at least divided into thermal ice and mixed ice, the mixed ice being a mixture of submerged ice and dense thick ice; The first data processing module is configured to determine first feature information and first processed image information based on the polarization data; wherein, the first feature information includes polarization parameter information and texture information extracted based on the gray-level co-occurrence matrix, the texture information being the mean and second moment of the angle, the polarization parameter information including the backscattering coefficient and the polarization angle α, and the first processed image information including the geographical information of the preset region; The second data processing module is configured to determine pseudo-color image information based on the first processed image information and establish a one-to-one correspondence between the pseudo-color image information and the measured information. The third data processing module is configured to use the pseudo-color image information to determine the river ice classification information using the Wissat distribution supervised classification method. The fourth data processing module is configured to determine the average ice thickness information of the same ice classification information within a preset area based on measured information, the first processed image information, and the first feature information. The determination module is configured to construct exclusive empirical models of river ice thickness for thermal ice and mixed ice based on the average river ice thickness information and the first feature information. Specifically, a linear empirical model of backscattering coefficient and polarization angle α is constructed for thermal ice, and a nonlinear empirical model of polarization angle α, mean of texture information and second moment of angle is constructed for mixed ice. The inversion module is configured to perform ice thickness inversion on the polarization data of the synthetic aperture radar based on the dedicated ice thickness empirical model and the first feature information.

Images