Method and device for constructing statistical distribution model of multi-angle synthetic aperture radar image

Abstract

The present invention provides a method and device for constructing a statistical distribution model of a multi-angle synthetic aperture radar image, which is applied to the technical field of image analysis. The method comprises: selecting a plurality of regional images of different angles and different regions from the multi-angle synthetic aperture radar image; preprocessing the plurality of regional images to obtain statistical distribution information corresponding to each regional image; the statistical distribution information represents the distribution of amplitude values ​​or intensity values ​​in the regional image; fitting the statistical distribution of each regional image using a plurality of single-parameter statistical distribution models respectively to obtain a first fitting result of each single-parameter statistical distribution model; obtaining a selection order of the single-parameter statistical distribution model based on the first fitting result; synthesizing the single-parameter statistical distribution model according to the selection order based on the idea of ​​stepwise regression to obtain a target comprehensive model; the target comprehensive model is used to analyze the statistical distribution characteristics of the multi-angle synthetic aperture radar image.

Description

Technical Field

[0001] The present invention relates to the technical field of synthetic aperture radar image analysis, and in particular to a method and device for constructing a multi-angle synthetic aperture radar image statistical distribution model. Background Art

[0002] Synthetic Aperture Radar (SAR) is a high-resolution imaging radar that can provide all-day and all-weather observations. The backscattering information of the target can be obtained through radar illumination. Multi-angle SAR can obtain images of the target at different angles by observing along multiple azimuths. Coherent accumulation of images at different angles can obtain large-aperture SAR images with richer detail information. The statistical distribution characteristics of the target in the large-aperture SAR image contain extremely rich backscattering information. Therefore, it is of great significance to study the statistical distribution characteristics of targets in the scene based on the multi-angle SAR platform.

[0003] The backscatter information of the target presents different statistical distribution characteristics at different aperture angles. However, the existing research on the statistical distribution characteristics of targets in multi-angle SAR images mainly uses a single parameter model to process the image, that is, only a specific mathematical distribution is used to model the statistical distribution characteristics of simple and single ground object types in SAR images. This single parameter model cannot analyze the statistical distribution characteristics of various complex ground object types in high-resolution multi-angle SAR images, and cannot meet the needs of fine modeling of the statistical distribution characteristics of various complex ground object types in high-resolution multi-angle SAR images. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] The present invention provides a method and device for constructing a multi-angle synthetic aperture radar image statistical distribution model, which are used to at least partially solve one of the above technical problems.

[0006] (II) Technical solution

[0007] The embodiment of the present invention provides a method for constructing a statistical distribution model of a multi-angle synthetic aperture radar image, comprising: selecting a plurality of regional images of different angles and different regions from the multi-angle synthetic aperture radar image; preprocessing the plurality of regional images to obtain statistical distribution information corresponding to each regional image; wherein the statistical distribution information is used to represent the distribution of amplitude values ​​or intensity values ​​in the regional image; respectively fitting the statistical distribution of each regional image using a plurality of single-parameter statistical distribution models to obtain a first fitting result of each single-parameter statistical distribution model; obtaining a selection order of the single-parameter statistical distribution model based on the first fitting result; and synthesizing the single-parameter statistical distribution model according to the selection order based on the idea of ​​stepwise regression to obtain a target comprehensive model; wherein the target comprehensive model is used to analyze the statistical distribution characteristics of the multi-angle synthetic aperture radar image.

[0008] Optionally, multiple single-parameter statistical distribution models are used to fit the statistical distribution of each regional image, and the fitting results of each single-parameter statistical distribution model are obtained, including: performing the following operations on each single-parameter statistical distribution model: fitting the statistical distribution of each regional image in turn to obtain the goodness of fit for each region; determining a first fitting result of the single-parameter statistical distribution model based on the goodness of fit of multiple regions; wherein the first fitting result is used to represent the ability of the single-parameter statistical distribution model to characterize the statistical distribution characteristics of multiple different regions.

[0009] Optionally, the goodness of fit is calculated based on difference data and penalty data; the penalty data is used to characterize the influence of the number of model parameters, and is calculated based on the number of data points in the current area and the number of parameters of the model; the difference data is used to characterize the difference between the model and the data, and the difference data is calculated based on the actual value of the amplitude or intensity statistical distribution of the current area, the mean of the actual value of the amplitude or intensity statistical distribution of the current area, and the predicted value of the amplitude or intensity statistical distribution of the current area obtained based on a single parameter statistical distribution model.

[0010] Optionally, based on the goodness of fit of multiple regions, a first fitting result of a single parameter statistical distribution model is determined, including: determining the number of goodness of fits that meet a first preset condition; obtaining the first fitting result of the single parameter statistical distribution model based on the number of goodness of fits; wherein the first preset condition is that the goodness of fit is greater than a first threshold.

[0011] Optionally, a selection order of single-parameter statistical distribution models is determined based on the fitting results, and the single-parameter statistical distribution models are synthesized according to the selection order to obtain a target comprehensive model, including: constructing a comprehensive model container; adding the single-parameter statistical distribution models to the comprehensive model container in sequence to obtain a first comprehensive model; fitting the target area in the multi-angle synthetic aperture radar image based on the first comprehensive model to determine a second fitting result of the first comprehensive model; wherein the target area is determined based on an application scenario of the target comprehensive model; if the second fitting result meets a second preset condition, the first comprehensive model is determined as the target comprehensive model; if the second fitting result does not meet the second preset condition, continue to add the single-parameter statistical distribution model to the comprehensive model container until a first comprehensive model that meets the second preset condition is obtained; wherein the second preset condition is that the goodness of fit is greater than a second threshold, and the second fitting result is used to express the characterization capability of the first comprehensive model for the target area in the multi-angle synthetic aperture radar image.

[0012] Optionally, single parameter statistical distribution models are added to the comprehensive model container in sequence to obtain a first comprehensive model, including: sorting multiple single parameter statistical distribution models based on the fitting results of each single parameter statistical distribution model to obtain a model selection order; adding the single parameter statistical distribution model ranked first in the model selection order to the comprehensive model container, and updating the model in the comprehensive model container; processing the updated model in the comprehensive model container based on a simulated annealing algorithm to obtain the first comprehensive model.

[0013] Optionally, the updated model in the comprehensive model container is processed based on the simulated annealing algorithm to obtain a first comprehensive model, including: performing an initialization operation on the model parameters in the comprehensive model container, using the initialized model parameters as the current parameter solution and the optimal parameter solution, and calculating the objective functions of the current parameter solution and the optimal parameter solution; judging whether the objective function of the optimal parameter solution satisfies a third preset condition; if the objective function of the optimal parameter solution satisfies the third preset condition, outputting the current model as the first comprehensive model; if the objective function does not satisfy the third preset condition, generating a new parameter solution based on an iterative operation until the objective function of the new parameter solution satisfies the third preset condition; wherein the third preset condition is: the objective function is greater than a second threshold or the objective function is the optimal objective function in a finite number of iterative operations.

[0014] Optionally, when the objective function does not satisfy the third preset condition, a new parameter solution is generated based on an iterative operation according to certain rules until the new parameter solution satisfies the third preset condition, including: calculating the objective function of the new parameter solution; determining whether to accept the new parameter solution based on the preset rules and the objective function; when the new parameter solution is not accepted, regenerating a new parameter solution according to the preset rules until the new parameter solution is accepted; when the new parameter solution is accepted, updating the current parameter solution to the new parameter solution; determining whether to accept the current parameter solution as the optimal parameter solution according to corresponding rules based on the updated objective function of the current parameter solution and the objective function of the current optimal parameter solution; when the current parameter solution is not accepted as the optimal parameter solution, regenerating a new parameter solution according to the preset rules and updating the current parameter solution, repeating the above steps until the current parameter solution is accepted as the optimal parameter solution; when the current parameter solution is accepted as the optimal parameter solution, determining whether the objective function of the optimal parameter solution satisfies the third preset condition; when the objective function satisfies the third preset condition, outputting the current model as the first comprehensive model; when the objective function does not satisfy the third preset condition, regenerating a new parameter solution until the new parameter solution is the optimal parameter solution whose objective function satisfies the third preset condition.

[0015] Optionally, the method further includes: after the single parameter statistical distribution model is added to the comprehensive model container, deleting the single parameter statistical distribution model from the model selection order and updating the model selection order.

[0016] The second aspect of the present invention provides a device for constructing a statistical distribution model of a multi-angle synthetic aperture radar image, comprising: a selection module, used to select a plurality of regional images of different angles and different regions from the multi-angle synthetic aperture radar image; a preprocessing module, used to preprocess the plurality of regional images to obtain statistical distribution information corresponding to each regional image; wherein the statistical distribution information is used to represent the distribution of amplitude values ​​or intensity values ​​in the regional image; a fitting module, used to fit the statistical distribution of each regional image using a plurality of single-parameter statistical distribution models respectively to obtain a first fitting result of each single-parameter statistical distribution model; a determination module, used to determine the selection order of the single-parameter statistical distribution model based on the fitting result; and a synthesis module, used to synthesize the single-parameter statistical distribution model according to the selection order based on the stepwise regression idea to obtain a target synthesis model; wherein the target synthesis model is used to analyze the statistical distribution characteristics of the multi-angle synthetic aperture radar image.

[0017] (III) Beneficial effects

[0018] The method for constructing a multi-angle synthetic aperture radar image statistical distribution model provided by the present invention has at least the following beneficial effects:

[0019] The method for constructing a multi-angle synthetic aperture radar image statistical distribution model provided by the present invention utilizes the idea of ​​stepwise regression to adaptively select the type and quantity of single-parameter statistical distribution models participating in the synthesis according to different types of land objects, and obtains a comprehensive model by synthesizing the single-parameter statistical distribution models. The obtained comprehensive model integrates the advantages of multiple single-parameter statistical distribution models, and can effectively improve the accuracy and stability of multi-angle SAR image target area classification and recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The above and other objects, features and advantages of the present invention will become more apparent through the following description of the embodiments of the present invention with reference to the accompanying drawings, in which:

[0021] Figure 1 A flowchart of a method for constructing a multi-angle synthetic aperture radar image statistical distribution model according to an embodiment of the present invention is schematically shown;

[0022] Figure 2 A flowchart schematically shows how to fit the statistical distribution information of each regional image using multiple single-parameter statistical distribution models to obtain fitting results of each single-parameter statistical distribution model according to an embodiment of the present invention;

[0023] Figure 3 Schematically shows a flowchart of determining a selection order of a single parameter statistical distribution model based on fitting results according to an embodiment of the present invention, synthesizing the single parameter statistical distribution model according to the selection order, and obtaining a target comprehensive model;

[0024] Figure 4 Schematically shows a flow chart of sequentially adding single parameter statistical distribution models to a comprehensive model container to obtain a first comprehensive model according to an embodiment of the present invention;

[0025] Figure 5 Schematically shows a flow chart of obtaining a first comprehensive model based on a simulated annealing algorithm according to an embodiment of the present invention;

[0026] Figure 6 A flowchart schematically shows a method of generating a new parameter solution based on an iterative operation until the new parameter solution meets the third preset condition when the objective function does not meet the third preset condition according to an embodiment of the present invention;

[0027] Figure 7 A structural block diagram of a device for constructing a multi-angle synthetic aperture radar image statistical distribution model according to an embodiment of the present invention is schematically shown;

[0028] Figure 8 A block diagram of an electronic device suitable for implementing a method for constructing a multi-angle synthetic aperture radar image statistical distribution according to an embodiment of the present invention is schematically shown. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical scheme and advantages of the present invention more clearly understood, the present invention is further described in detail below in combination with specific embodiments and with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] The terms used herein are only for describing specific embodiments and are not intended to limit the present invention. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0031] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection, an electrical connection, or can communicate with each other; it can be a direct connection, or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0032] In the description of the present invention, it should be understood that the terms "longitudinal", "length", "circumferential", "front", "rear", "left", "right", "top", "bottom", "inside", "outside", etc., indicating the orientation or positional relationship are based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the subsystem or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0033] Throughout the drawings, the same elements are represented by the same or similar reference numerals. Conventional structures or configurations will be omitted when they may cause confusion in the understanding of the present invention. The shapes, sizes, and positional relationships of the components in the drawings do not reflect the actual size, proportion, and actual positional relationship.

[0034] Similarly, in order to simplify the present invention and help understand one or more of the various disclosed aspects, in the above description of exemplary embodiments of the present invention, the various features of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. The description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0035] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0036] Before introducing the specific embodiments of the present invention, the relevant technical background and related terms in the present invention are first introduced.

[0037] Multi-angle SAR image: refers to synthetic aperture radar (SAR) images of different angles obtained by observing in multiple azimuths, and then coherently accumulating SAR images of different angles to obtain large aperture angle SAR images with richer detail information. In multi-angle SAR images, the scattering characteristics of the same ground object or target are different at different observation angles, so the accuracy and robustness of target classification and recognition can be improved by analyzing and studying multi-angle SAR images.

[0038] Target statistical distribution characteristics: refers to the statistical distribution characteristics of backscattering of various targets at different observation angles in multi-angle SAR images, which can be obtained by analyzing the statistical distribution characteristics of amplitude or intensity of various targets in multi-angle SAR images.

[0039] Figure 1 The flowchart of the method for constructing a multi-angle synthetic aperture radar image statistical distribution model according to an embodiment of the present invention is schematically shown.

[0040] like Figure 1 As shown, the target model building method of this embodiment includes operations S110 to S150.

[0041] In operation S110 , a plurality of regional images of different regions at different angles are selected from the multi-angle synthetic aperture radar image.

[0042] In some embodiments, a plurality of area images with the same size are randomly selected from the multi-angle SAR image, and the observation azimuth angles of the area images are different.

[0043] In operation S120, a plurality of region images are preprocessed to obtain statistical distribution information corresponding to each region image.

[0044] In some embodiments, the statistical distribution information is used to represent the distribution of amplitude values ​​or intensity values ​​in the area image.

[0045] In operation S130, a plurality of single-parameter statistical distribution models are used to fit the statistical distribution of each regional image, and a first fitting result of each single-parameter statistical distribution model is obtained.

[0046] In some embodiments, each single parameter statistical distribution model is controlled to fit the statistical distribution of each regional image in turn, and the amplitude or intensity statistical distribution prediction value of the model for each regional image is obtained, and the goodness of fit of the single parameter statistical distribution model on the current regional image is determined based on the amplitude or intensity statistical distribution prediction value output by the model and the actual value of the amplitude or intensity statistical distribution of the regional image. By obtaining the goodness of fit of the single parameter statistical distribution model on each regional image, the fitting result of the single parameter statistical distribution model is obtained.

[0047] The single model may include, for example, a Gamma distribution model, a Lognormal distribution model, a Weibull distribution model, a K distribution model, and a G0 distribution model.

[0048] In operation S140, a selection order of a single parameter statistical distribution model is obtained based on the first fitting result.

[0049] In operation S150, the single parameter statistical distribution model is integrated according to the selection order based on the stepwise regression idea to obtain a target integrated model.

[0050] Among them, the target comprehensive model is used to model and analyze the statistical distribution characteristics of multi-angle SAR images.

[0051] In some embodiments, multiple single-parameter statistical distribution models are sorted based on the fitting results to obtain a model selection order. Based on the idea of ​​stepwise regression, the type and quantity of single-parameter statistical distribution models are adaptively selected for different land feature types according to the model selection order, and the selected single-parameter statistical distribution models are integrated until the target integrated model is obtained.

[0052] Figure 2The flowchart schematically shows how to fit the statistical distribution of each regional image using multiple single-parameter statistical distribution models to obtain the fitting results of each single-parameter statistical distribution model according to an embodiment of the present invention.

[0053] like Figure 2 As shown, this embodiment uses multiple single-parameter statistical distribution models to fit the statistical distribution of each regional image, and obtains the fitting results of each single-parameter statistical distribution model, including operations S121 to S122. Operations S121 to S122 are performed on each single-parameter statistical distribution model.

[0054] In operation S121, a fitting operation is performed on the statistical distribution of each region image in turn, and the goodness of fit of the single parameter statistical distribution model in each region is calculated to obtain the goodness of fit of each region.

[0055] In some embodiments, a single parameter statistical distribution model estimates the parameters of the model for data of an area selected from multi-angle SAR images of different angles and regions to obtain a statistical distribution function that can describe the statistical distribution characteristics of the selected area, and calculates the amplitude or intensity statistical distribution prediction value of the image area based on the statistical distribution function.

[0056] The data is obtained by selecting an area of ​​interest in the image, normalizing the amplitude or intensity value of the selected area and obtaining its statistical distribution. For example, a series of image slices of the same size can be randomly selected from the image or selected according to certain rules, and the amplitude or intensity of the slices can be normalized to obtain their statistical distribution.

[0057] In some embodiments, goodness of fit is also called the corrected determination coefficient, which is used to measure the representation ability and interpretability of the fitted data. It is an absolute indicator for evaluating the goodness of fit and can be used for objective comparison between different models. The closer the goodness of fit value is to 1, the higher the similarity between the data and the model, and the stronger the representation ability and interpretability.

[0058] The goodness of fit can be calculated based on the difference data and the penalty data. The penalty data is used to characterize the influence of the number of model parameters and is calculated based on the number of data points in the current region and the number of model parameters. The difference data is used to characterize the difference between the model and the data and is calculated based on the actual value of the amplitude or intensity statistical distribution of the current region, the mean of the actual value of the amplitude or intensity statistical distribution of the current region, and the predicted value of the amplitude or intensity statistical distribution of the current region obtained based on the single parameter statistical distribution model.

[0059] Specifically, the calculation formula for goodness of fit is as follows:

[0060]

[0061] in, is the goodness of fit of the single parameter statistical distribution model in the current area image, is the difference data, y select_region is the actual value of the amplitude or intensity statistical distribution of the current area image, It is the predicted value of the amplitude or intensity statistical distribution of the current regional image based on a single parameter statistical distribution model. is the mean of the actual value of the amplitude or intensity statistical distribution of the current area image, is the penalty data, m is the number of data points in the current region image, k s is the number of parameters of the one-parameter statistical distribution model.

[0062] In operation S122, a first fitting result of the single parameter statistical distribution model is determined according to the goodness of fit of the multiple regions; wherein the first fitting result is used to indicate the characterization capability of the single parameter statistical distribution model for the statistical distribution characteristics of multiple different regions.

[0063] In some embodiments, the fitting result of the single parameter statistical distribution model is determined based on the number of goodness of fits that meet the first preset condition.

[0064] Specifically, when there is a goodness of fit that meets the first preset condition in the single parameter statistical distribution model, the value of the first fitting result increases by 1 (ie, the value of the first fitting result is equal to the number of goodness of fits that meet the first preset condition).

[0065] Among them, the first preset condition is that the goodness of fit is greater than the first threshold (i.e., Rsquare_model_region>Rsquare_select_threshold), and the first fitting result is used to indicate the characterization ability of the single parameter statistical distribution model for multiple different regional statistical distributions. The higher the value of the first fitting result, the better the characterization ability of the single parameter statistical distribution model for multiple different regional images.

[0066] Figure 3 A flowchart schematically illustrates determining a selection order of single-parameter statistical distribution models based on a first fitting result, synthesizing the single-parameter statistical distribution models according to the selection order, and obtaining a target synthesized model according to an embodiment of the present invention.

[0067] like Figure 3 As shown, this embodiment determines the selection order of the single parameter statistical distribution model based on the first fitting result, synthesizes the single parameter statistical distribution model according to the selection order to obtain the target comprehensive model, including operations S131 to S135.

[0068] In operation S131, a comprehensive model container is constructed.

[0069] In some embodiments, the comprehensive model container is initialized to be empty, that is, there is no model in the comprehensive model container.

[0070] In operation S132, single parameter statistical distribution models are sequentially added to the comprehensive model container to obtain a first comprehensive model.

[0071] In some embodiments, multiple single-parameter statistical distribution models are sorted based on the first fitting results of each single-parameter statistical distribution model to determine the selection order of the multiple single models. The single-parameter statistical distribution models are gradually added to the comprehensive model container based on the selection order. When the comprehensive model container receives a single-parameter statistical distribution model, the comprehensive model container is updated to obtain a first comprehensive model based on the simulated annealing algorithm.

[0072] The comprehensive model container is used to store the single parameter statistical distribution model to be integrated. If the comprehensive model container is empty, the first comprehensive model is the updated single parameter statistical distribution model. If one or more single parameter statistical distribution models already exist in the comprehensive model container, the single parameter statistical distribution model ranked first in the updated model selection order is added to the comprehensive model container with one or more single parameter statistical distribution models, and the comprehensive model is obtained based on the simulated annealing algorithm.

[0073] In operation S133, the target area in the multi-angle SAR image is fitted using the first comprehensive model to determine a second fitting result of the first comprehensive model; wherein the target area is determined based on an application scenario of the target comprehensive model.

[0074] In some embodiments, the image processing operation is performed using the first comprehensive model, for example, the statistical distribution of the amplitude or intensity of the target area of ​​interest in the multi-angle SAR image is fitted by the first comprehensive model to obtain the fitting result of the first comprehensive model.

[0075] Specifically, the calculation formula of the second fitting result Rsquare_temp of the first comprehensive model is as follows:

[0076]

[0077] in, is the actual value of the amplitude or intensity statistical distribution of the target area of ​​interest selected in the multi-angle SAR image, is the predicted value of the statistical distribution of amplitude or intensity in the region obtained by the first comprehensive model, is the mean of the actual value of the statistical distribution of the amplitude or intensity in the region, m is the number of data points in the region, k cis the number of parameters in the first comprehensive model, that is, the sum of the number of parameters of all single-parameter statistical distribution models in the comprehensive model container.

[0078] In operation S134, if the second fitting result satisfies the second preset condition, the first integrated model is determined as a target integrated model.

[0079] In operation S135 , when the second fitting result does not satisfy the second preset condition, single parameter statistical distribution models continue to be added to the comprehensive model container until a first comprehensive model satisfying the second preset condition is obtained.

[0080] In some embodiments, it is determined whether it is necessary to continue adding a single parameter statistical distribution model to the comprehensive model container based on the second fitting result of the current comprehensive model.

[0081] If the second fitting result of the current first comprehensive model meets the second preset condition, the current first comprehensive model is determined as the target comprehensive model.

[0082] If the second fitting result does not satisfy the second preset condition, the above operations S132 to S133 are repeated to continue adding single parameter statistical distribution models to the comprehensive model container until a first comprehensive model satisfying the second preset condition is obtained.

[0083] The second preset condition is that the second fitting result of the current comprehensive model is greater than the second threshold. In the embodiment of the present invention, the second threshold is set to 0.995, and the second threshold can be set according to the actual needs of those skilled in the art, and the present invention is not limited here.

[0084] Figure 4 The flowchart schematically shows a process of sequentially adding single parameter statistical distribution models into a comprehensive model container to obtain a first comprehensive model according to an embodiment of the present invention.

[0085] like Figure 4 As shown, this embodiment sequentially adds single parameter statistical distribution models to the comprehensive model container to obtain a first comprehensive model, including operations S1321 to S1323.

[0086] In operation S1321 , a plurality of single-parameter statistical distribution models are sorted based on the first fitting results of the single-parameter statistical distribution models to obtain a model selection order.

[0087] In operation S1322, the single parameter statistical distribution model ranked first in the model selection order is added to the comprehensive model container, and the models in the comprehensive model container are updated.

[0088] In operation S1323, a first comprehensive model is obtained based on a simulated annealing algorithm.

[0089] In some embodiments, multiple single-parameter statistical distribution models are sorted from large to small according to the fitting results to obtain a model selection order. The single-parameter statistical distribution model ranked first in the model selection order is selected and added to the comprehensive model container, the comprehensive model container is updated, and parameter estimation and model synthesis are performed on one or more single-parameter statistical distribution models in the updated comprehensive model container based on a simulated annealing algorithm to obtain a first comprehensive model.

[0090] In a specific implementation process, after the single parameter statistical distribution model is added to the comprehensive model container, the single parameter statistical distribution model is deleted from the model selection order and the model selection order is updated.

[0091] In the specific implementation process, operation S1323, obtaining a first comprehensive model based on a simulated annealing algorithm, further includes operations S510 to S540.

[0092] Figure 5 The flowchart of obtaining the first comprehensive model based on the simulated annealing algorithm according to an embodiment of the present invention is schematically shown.

[0093] like Figure 5 As shown, the first comprehensive model is obtained based on the simulated annealing algorithm in this embodiment, including operations S510 to S540.

[0094] In operation S510, an initialization operation is performed on the model parameters of the model in the comprehensive model container, the initialized model parameters are used as the current parameter solution and the optimal parameter solution, and the objective functions of the current parameter solution and the optimal parameter solution are calculated.

[0095] In operation S520, it is determined whether the objective function of the optimal parameter solution satisfies a third preset condition.

[0096] In operation S530, in a case where the objective function of the optimal parameter solution satisfies a third preset condition, the current model is output as a first integrated model.

[0097] In operation S540, when the objective function of the optimal parameter solution does not satisfy the third preset condition, a new parameter solution is generated based on an iterative operation until the objective function of the new parameter solution satisfies the third preset condition.

[0098] In some embodiments, the comprehensive model container receives a newly-input single-parameter statistical distribution model and updates the comprehensive model container, and performs an initialization operation on the model parameters of one or more existing single-parameter statistical distribution models in the updated comprehensive model container, and uses them as the current parameter solution and the optimal parameter solution.

[0099] In some embodiments, after the initialization operation is completed, the objective function of the synthesized integrated model is calculated based on the new parameter solution obtained by initialization.

[0100] Specifically, the objective function may be, for example, the fitting result of the comprehensive model to the statistical distribution of the target area of ​​the multi-angle SAR image under the parameter solution. After the initialization operation is completed, the comprehensive model is controlled to perform the multi-angle SAR image processing task based on the new parameter solution, and the fitting result of the comprehensive model to the statistical distribution of the target area of ​​the multi-angle SAR image under the new parameter solution obtained by initialization is calculated.

[0101] The third preset condition is that the objective function is greater than a second threshold or the objective function is the optimal objective function in a finite number of iterative operations.

[0102] When the objective function is greater than the second threshold, the comprehensive model under the parameter solution can be directly output as the first comprehensive model. Otherwise, the iterative operation is continued to generate a new parameter solution until the objective function of the parameter solution is greater than the second threshold, or the number of iterative operations reaches an upper limit. For example, the preset upper limit of the number of iterative operations is 5000. When the 5000th iterative operation is performed, even if the objective function of the parameter solution is not greater than the second threshold, the current model is output as the first comprehensive model.

[0103] Among them, the new parameter solution is usually obtained by making a small random perturbation to the current model parameters. For example, the values ​​of certain parameters can be randomly increased or decreased, or the values ​​of certain parameters can be replaced by random parameters.

[0104] Figure 6 The flowchart schematically shows a method of generating a new parameter solution based on an iterative operation until the new parameter solution meets the third preset condition when the objective function does not meet the third preset condition according to an embodiment of the present invention.

[0105] like Figure 6 As shown, in this embodiment, when the objective function does not satisfy the third preset condition, a new parameter solution is generated based on an iterative operation until the new parameter solution satisfies the third preset condition, including operations S610 to S670.

[0106] In operation S610, an objective function of a new parameter solution is calculated.

[0107] In operation S620, it is determined whether to accept a new parameter solution based on a preset rule and an objective function.

[0108] In operation S630, if the new parameter solution is not accepted, a new parameter solution is regenerated according to a preset rule until the new parameter solution is accepted.

[0109] In operation S640, when the new parameter solution is accepted, the current parameter solution is updated to the new parameter solution.

[0110] In operation S650 , it is determined whether to accept the current parameter solution as the optimal parameter solution based on the updated objective function of the current parameter solution and the objective function of the current optimal parameter solution.

[0111] In operation S660, when the current parameter solution is accepted as the optimal parameter solution, it is determined whether the objective function of the optimal parameter solution satisfies a third preset condition.

[0112] In some embodiments, when the current parameter solution is not accepted as the optimal parameter solution, a new parameter solution is regenerated according to preset rules, and it is determined whether to accept the new parameter solution. Based on the determination result, it is determined whether to update the current parameter solution, and the above steps S610~S650 are repeated until the current parameter solution is accepted as the optimal parameter solution.

[0113] In operation S670, in a case where the objective function of the optimal parameter solution satisfies a third preset condition, the current model is output as a first integrated model.

[0114] In some embodiments, when the objective function of the optimal parameter solution does not satisfy the third preset condition, a new parameter solution is regenerated according to the preset rules, and the above steps S610~S660 are repeated until the new parameter solution is the optimal parameter solution whose objective function satisfies the third preset condition.

[0115] In some embodiments, the objective function of the integrated model is calculated for the new parameter solution generated according to certain rules to determine whether to accept the new parameter solution.

[0116] Specifically, after generating a new parameter solution, the integrated model is controlled to perform the multi-angle SAR image processing task based on the new parameter solution, the fitting result of the integrated model for the statistical distribution of the target area of ​​the multi-angle SAR image under the new parameter solution is calculated, and the objective function of the parameter solution is obtained. According to the objective function, it is determined whether to accept the new parameter solution.

[0117] If the objective function of the new parameter solution is better than the objective function of the current parameter solution, the new parameter solution is taken as the current solution. If the objective function of the new parameter solution is worse than the objective function of the current parameter solution, the new solution is accepted with a certain probability according to certain rules.

[0118] Specifically, the probability of determining whether to accept a new parameter solution includes: calculating the difference between the new parameter solution and the current parameter solution, obtaining the change in the objective function, and generating a random number according to a certain probability distribution. The probability of accepting a new parameter solution is calculated based on the current temperature, the change in the objective function, and a random number, so as to better explore possible solutions in the search space.

[0119] In the case where the new parameter solution is not accepted, a new parameter solution is generated again according to the corresponding rules until the new parameter solution is accepted.

[0120] In the case of accepting a new parameter solution, the current parameter solution is updated to the accepted new parameter solution, and based on the objective function of the updated current parameter solution and the objective function of the optimal parameter solution, it is determined whether to accept the current parameter solution as the optimal parameter solution according to the corresponding rules. If the objective function of the current parameter solution is better than the objective function of the optimal parameter solution, the current parameter solution is accepted. If the objective function of the current parameter solution is worse than the objective function of the optimal parameter solution, the current parameter solution is not accepted as the optimal parameter solution.

[0121] When the current parameter solution is not accepted as the optimal parameter solution, a new parameter solution is generated according to certain rules and the current parameter solution is updated. The above steps are repeated until the current parameter solution is accepted as the optimal parameter solution.

[0122] When the current parameter solution is accepted as the optimal parameter solution, the optimal parameter solution is updated to the current parameter solution, and it is determined whether the objective function of the optimal parameter solution satisfies the third preset condition. If so, it is output as the first comprehensive model to obtain the first comprehensive model. Otherwise, the above steps are repeated to regenerate a new parameter solution according to certain rules and update the current parameter solution and the optimal parameter solution until the objective function of the optimal parameter solution satisfies the third preset condition.

[0123] The method for constructing a multi-angle synthetic aperture radar image statistical distribution model provided by the present invention increases the number of free variables of the model by synthesizing the single parameter statistical distribution model, thereby improving the flexibility of the model and the ability to describe the details of the statistical distribution characteristics of complex scenes in multi-angle SAR images. By adaptively selecting the type and number of single parameter statistical distribution models involved in the synthesis for different types of objects, a more accurate description of the statistical distribution model of the target area of ​​the multi-angle SAR image is achieved using a single parameter model with the least type and number of models, thereby improving the accuracy and universality of the model, so that the model can be applied to SAR images of different angles.

[0124] Based on the above-mentioned method for constructing a statistical distribution model of a multi-angle synthetic aperture radar image, the present invention also provides a device for constructing a statistical distribution model of a multi-angle synthetic aperture radar image. Figure 7 The device is described in detail.

[0125] Figure 7 The structure block diagram of the apparatus for constructing a multi-angle synthetic aperture radar image statistical distribution model according to an embodiment of the present invention is schematically shown.

[0126] like Figure 7As shown, the multi-angle synthetic aperture radar image statistical distribution model construction device 700 of this embodiment includes a selection module 710, a preprocessing module 720, a fitting module 730, a determination module 740 and a synthesis module 750.

[0127] The selection module 710 is used to select a plurality of regional images of different angles and different regions from the multi-angle synthetic aperture radar image. In one embodiment, the selection module 710 can be used to perform the operation S110 described above, which will not be described in detail here.

[0128] The preprocessing module 720 is used to preprocess the multiple regional images to obtain statistical distribution information corresponding to each regional image; wherein the statistical distribution information is used to represent the distribution of amplitude values ​​or intensity values ​​in the regional image. In one embodiment, the preprocessing module 720 can be used to perform the operation S120 described above, which will not be repeated here.

[0129] The fitting module 730 is used to fit the statistical distribution of each regional image using multiple single parameter statistical distribution models to obtain a first fitting result of each single parameter statistical distribution model. In one embodiment, the fitting module 730 can be used to perform the operation S130 described above, which will not be described in detail here.

[0130] The determination module 740 is used to determine the selection order of the single parameter statistical distribution model based on the first fitting result. In one embodiment, the determination module 740 can be used to perform the operation S140 described above, which will not be described in detail here.

[0131] The synthesis module 750 is used to synthesize the single parameter statistical distribution model according to the selection order based on the stepwise regression idea to obtain a target synthesis model; wherein the target synthesis model is used to model and analyze the statistical distribution characteristics of the multi-angle synthetic aperture radar image. In one embodiment, the synthesis module 750 can be used to perform the operation S150 described above, which will not be repeated here.

[0132] According to an embodiment of the present invention, any multiple modules of the selection module 710, the preprocessing module 720, the fitting module 730, the determination module 740 and the synthesis module 750 can be combined into one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present invention, at least one of the selection module 710, the preprocessing module 720, the fitting module 730, the determination module 740 and the synthesis module 750 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware and firmware or in a proper combination of any of them. Alternatively, at least one of the selection module 710, the preprocessing module 720, the fitting module 730, the determination module 740 and the integration module 750 may be at least partially implemented as a computer program module, and when the computer program module is executed, a corresponding function may be performed.

[0133] Figure 8 A block diagram of an electronic device suitable for implementing a method for constructing a multi-angle synthetic aperture radar image statistical distribution model according to an embodiment of the present invention is schematically shown.

[0134] like Figure 8As shown, the electronic device 800 according to the embodiment of the present invention includes a processor 801, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 802 or the program loaded from the storage part 808 to the random access memory (RAM) 803. The processor 801 may include, for example, a general-purpose microprocessor (such as a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (for example, an application-specific integrated circuit (ASIC)), etc. The processor 801 may also include an on-board memory for cache purposes. The processor 801 may include a single processing unit or multiple processing units for performing different actions of the method flow according to the embodiment of the present invention. In the RAM 803, various programs and data required for the operation of the electronic device 800 are stored. The processor 801, ROM802 and RAM 803 are connected to each other through a bus 804. The processor 801 performs various operations of the method flow according to the embodiment of the present invention by executing the program in the ROM 802 and / or RAM 803. It should be noted that the program can also be stored in one or more memories other than ROM802 and RAM 803. The processor 801 may also execute various operations of the method flow according to the embodiment of the present invention by executing programs stored in the one or more memories.

[0135] According to an embodiment of the present invention, the electronic device 800 may further include an input / output (I / O) interface 805, which is also connected to the bus 804. The electronic device 800 may further include one or more of the following components connected to the I / O interface 805: an input portion 806 including a keyboard, a mouse, etc.; an output portion 807 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 808 including a hard disk, etc.; and a communication portion 809 including a network interface card such as a LAN card, a modem, etc. The communication portion 809 performs communication processing via a network such as the Internet. A drive 810 is also connected to the I / O interface 805 as needed. A removable medium 811, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 810 as needed, so that the computer program read therefrom is installed into the storage portion 808 as needed.

[0136] The embodiments of the present invention are described above. However, these embodiments are only for the purpose of illustration, and are not intended to limit the scope of the present invention. Although each embodiment is described above, it does not mean that the measures in each embodiment cannot be used in combination advantageously. Without departing from the scope of the present invention, those skilled in the art may make various substitutions and modifications, which should all fall within the scope of the present invention.

Claims

1. A method for constructing a statistical distribution model of a multi-angle synthetic aperture radar image, characterized in that: include: Selecting a plurality of regional images of different areas at different angles from the multi-angle synthetic aperture radar image; Preprocessing the multiple regional images to obtain statistical distribution information corresponding to each regional image; wherein the statistical distribution information is used to represent the distribution of amplitude values ​​or intensity values ​​in the regional images; Using a plurality of single-parameter statistical distribution models to fit the statistical distribution of each regional image, respectively, to obtain a first fitting result of each single-parameter statistical distribution model; Obtaining a selection order of a single parameter statistical distribution model based on the first fitting result; Based on the stepwise regression idea, the single parameter statistical distribution model is integrated according to the selection order to obtain a target comprehensive model; wherein the target comprehensive model is used to analyze the statistical distribution characteristics of the multi-angle synthetic aperture radar image; The method of fitting the statistical distribution of each regional image using a plurality of single-parameter statistical distribution models to obtain the fitting results of each single-parameter statistical distribution model comprises: performing the following operations on each single-parameter statistical distribution model: fitting the statistical distribution of each regional image in turn, calculating the goodness of fit of the single-parameter statistical distribution model in each region, and obtaining the goodness of fit of each region; determining a first fitting result of the single-parameter statistical distribution model according to the goodness of fit of the plurality of regions; wherein the first fitting result is used to indicate the characterization capability of the single-parameter statistical distribution model for the statistical distribution characteristics of a plurality of different regions; The goodness of fit is calculated based on the difference data and the penalty data; the penalty data is used to characterize the influence of the number of model parameters, and is calculated based on the number of data points in the current area and the number of parameters of the model; the difference data is used to characterize the difference between the model and the data, and the difference data is calculated based on the actual value of the amplitude or intensity statistical distribution of the current area, the mean of the actual value of the amplitude or intensity statistical distribution of the current area, and the predicted value of the amplitude or intensity statistical distribution of the current area obtained based on the single parameter statistical distribution model.

2. The method for constructing a multi-angle synthetic aperture radar image statistical distribution model according to claim 1, characterized in that: Determining a first fitting result of the single parameter statistical distribution model according to the goodness of fit of the multiple regions includes: determining the number of goodness of fits that satisfy a first predetermined condition; Obtaining a first fitting result of the single parameter statistical distribution model based on the number of goodness of fits; Among them, the first preset condition is that the goodness of fit is greater than a first threshold.

3. The method for constructing a multi-angle synthetic aperture radar image statistical distribution model according to claim 1, characterized in that: The step of determining a selection order of single-parameter statistical distribution models based on the fitting results, and synthesizing the single-parameter statistical distribution models according to the selection order to obtain a target comprehensive model includes: Build comprehensive model containers; Adding the single parameter statistical distribution model to the comprehensive model container in sequence to obtain a first comprehensive model; Using the first comprehensive model to fit the target area in the multi-angle synthetic aperture radar image, and determining a second fitting result of the first comprehensive model; wherein the target area is determined based on an application scenario of the target comprehensive model; When the second fitting result satisfies a second preset condition, determining the first comprehensive model as a target comprehensive model; When the second fitting result does not satisfy the second preset condition, continue to add the single parameter statistical distribution model to the comprehensive model container until a first comprehensive model satisfying the second preset condition is obtained; Among them, the second preset condition is that the goodness of fit is greater than a second threshold, and the second fitting result is used to express the characterization ability of the first comprehensive model for the target area in the multi-angle synthetic aperture radar image.

4. The method for constructing a multi-angle synthetic aperture radar image statistical distribution model according to claim 3, characterized in that: The step of sequentially adding the single parameter statistical distribution model to the comprehensive model container to obtain a first comprehensive model includes: sorting the plurality of single-parameter statistical distribution models based on the first fitting results of the single-parameter statistical distribution models to obtain a model selection order; Adding the single parameter statistical distribution model ranked first in the model selection order to the comprehensive model container, and updating the model in the comprehensive model container; The first comprehensive model was obtained based on the simulated annealing algorithm.

5. The method for constructing a multi-angle synthetic aperture radar image statistical distribution model according to claim 4, characterized in that: The method of obtaining a first comprehensive model based on a simulated annealing algorithm comprises: Performing an initialization operation on the model parameters of the first comprehensive model obtained in the comprehensive model container, using the initialized model parameters as the current parameter solution and the optimal parameter solution, and calculating the objective functions of the current parameter solution and the optimal parameter solution; Determine whether the objective function of the optimal parameter solution satisfies a third preset condition; When the objective function of the optimal parameter solution satisfies a third preset condition, outputting the current model as a first comprehensive model; If the objective function does not satisfy the third preset condition, generating a new parameter solution based on an iterative operation until the objective function of the new parameter solution satisfies the third preset condition; The third preset condition is that the objective function is greater than a second threshold or the objective function is the optimal objective function in a finite number of iterative operations.

6. The method for constructing a multi-angle synthetic aperture radar image statistical distribution model according to claim 5, characterized in that: When the objective function does not satisfy the third preset condition, generating a new parameter solution based on an iterative operation until the new parameter solution satisfies the third preset condition includes: Calculating the objective function of the new parameter solution; Determine whether to accept a new parameter solution based on preset rules and the objective function; If the new parameter solution is not accepted, a new parameter solution is generated again according to the preset rules until the new parameter solution is accepted; When a new parameter solution is accepted, the current parameter solution is updated to the new parameter solution; Determine whether to accept the current parameter solution as the optimal parameter solution based on the updated objective function of the current parameter solution and the objective function of the current optimal parameter solution; If the current parameter solution is not accepted as the optimal parameter solution, a new parameter solution is generated according to the preset rules, and the current parameter solution is updated, and the above steps are repeated until the current parameter solution is accepted as the optimal parameter solution; In the case of accepting the current parameter solution as the optimal parameter solution, determining whether the objective function of the optimal parameter solution satisfies a third preset condition; When the objective function satisfies a third preset condition, outputting the current model as a first comprehensive model; When the objective function does not satisfy the third preset condition, a new parameter solution is regenerated until the new parameter solution is an optimal parameter solution in which the objective function satisfies the third preset condition.

7. The method for constructing a multi-angle synthetic aperture radar image statistical distribution model according to claim 4, characterized in that: The method further comprises: After the single parameter statistical distribution model is added to the comprehensive model container, the single parameter statistical distribution model is deleted from the model selection sequence and the model selection sequence is updated.

8. A device for constructing a statistical distribution model of a multi-angle synthetic aperture radar image, characterized in that: The device comprises: A selection module is used to select a plurality of regional images of different angles and regions from the multi-angle synthetic aperture radar image; A preprocessing module, used to preprocess the multiple regional images to obtain statistical distribution information corresponding to each regional image; wherein the statistical distribution information is used to represent the distribution of amplitude values ​​or intensity values ​​in the regional images; A fitting module, used to fit the statistical distribution of each regional image using multiple single-parameter statistical distribution models respectively, to obtain a first fitting result of each single-parameter statistical distribution model; A determination module, used for determining a selection order of a single parameter statistical distribution model based on the fitting result; A synthesis module, used for synthesizing the single parameter statistical distribution model according to the selection order based on the stepwise regression idea to obtain a target synthesis model; wherein the target synthesis model is used to analyze the statistical distribution characteristics of the multi-angle synthetic aperture radar image; The method of fitting the statistical distribution of each regional image using a plurality of single-parameter statistical distribution models to obtain the fitting results of each single-parameter statistical distribution model comprises: performing the following operations on each single-parameter statistical distribution model: fitting the statistical distribution of each regional image in turn, calculating the goodness of fit of the single-parameter statistical distribution model in each region, and obtaining the goodness of fit of each region; determining a first fitting result of the single-parameter statistical distribution model according to the goodness of fit of the plurality of regions; wherein the first fitting result is used to indicate the characterization capability of the single-parameter statistical distribution model for the statistical distribution characteristics of a plurality of different regions; The goodness of fit is calculated based on the difference data and the penalty data; the penalty data is used to characterize the influence of the number of model parameters, and is calculated based on the number of data points in the current area and the number of parameters of the model; the difference data is used to characterize the difference between the model and the data, and the difference data is calculated based on the actual value of the amplitude or intensity statistical distribution of the current area, the mean of the actual value of the amplitude or intensity statistical distribution of the current area, and the predicted value of the amplitude or intensity statistical distribution of the current area obtained based on the single parameter statistical distribution model.

Images