CN116736300A - A fast maximum likelihood estimation method for the equivalent view number map of polarimetric SAR images - Google Patents
A fast maximum likelihood estimation method for the equivalent view number map of polarimetric SAR images Download PDFInfo
- Publication number
- CN116736300A CN116736300A CN202310583567.5A CN202310583567A CN116736300A CN 116736300 A CN116736300 A CN 116736300A CN 202310583567 A CN202310583567 A CN 202310583567A CN 116736300 A CN116736300 A CN 116736300A
- Authority
- CN
- China
- Prior art keywords
- sar image
- polarized sar
- determinant
- graph
- logarithmic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/88—Radar or analogous systems specially adapted for specific applications
- G01S13/89—Radar or analogous systems specially adapted for specific applications for mapping or imaging
- G01S13/90—Radar or analogous systems specially adapted for specific applications for mapping or imaging using synthetic aperture techniques, e.g. synthetic aperture radar [SAR] techniques
- G01S13/9021—SAR image post-processing techniques
Landscapes
- Engineering & Computer Science (AREA)
- Remote Sensing (AREA)
- Radar, Positioning & Navigation (AREA)
- Physics & Mathematics (AREA)
- Electromagnetism (AREA)
- Computer Networks & Wireless Communication (AREA)
- General Physics & Mathematics (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention relates to a quick maximum likelihood estimation method of a polarimetric SAR image equivalent apparent map, and relates to the technical field of imaging radar image processing. Comprising the following steps: obtaining a logarithmic determinant graph of polarized SAR image data by adopting logarithmic operation; calculating a local mean value diagram of the polarized SAR logarithmic determinant diagram; calculating a local mean value graph of data of each channel of the polarized SAR; calculating determinant values of all pixel data of the polarized SAR image after local averaging to form a corresponding determinant graph, and obtaining a logarithmic determinant graph by adopting logarithmic operation; obtaining a local sample statistic map of equivalent apparent number estimation of polarized SAR data by adopting matrix subtraction; and solving a solution by utilizing the analytic approximation of the ML estimation and rapidly estimating an equivalent apparent number diagram of the polarized SAR image based on matrix operation. The method of the invention avoids the problems of iterative numerical operation and initial interval setting of the traditional method, and the efficiency advantage of the fast calculation method based on matrix operation and convolution in estimating the equivalent apparent number diagram of the polarized SAR image is particularly obvious.
Description
Technical Field
The invention relates to the technical field of imaging radar image processing, in particular to a method for quickly realizing maximum likelihood (Maximum Likelihood, ML) estimation of an equivalent apparent number map of a polarized synthetic aperture radar (Synthetic Aperture Radar, SAR) image.
Background
The polarized SAR is provided with a plurality of polarized transceiving combined working modes, has strong information acquisition capability, is not influenced by weather factors such as cloud, rain, fog and the like due to the adoption of an active microwave imaging mode, and can work all day long, so that the polarized SAR image processing technology has important application value in civil and military fields.
Due to a coherent imaging mechanism of SAR, the polarized SAR image contains a large amount of random speckle noise, so that the interpretation difficulty is greatly increased, and a method based on a statistical distribution model becomes an important path for the polarized SAR image interpretation. The complex Wishart distribution is the most widely applied polarized SAR image statistical distribution model at present, wherein the equivalent apparent number is a key parameter, and the accuracy directly influences the reliability of the distribution model. Physically, the equivalent apparent number reflects the degree to which polarized SAR image data is averaged, and can measure the uniformity of the corresponding data. The equivalent view diagram of the polarized SAR image refers to an image formed by equivalent views corresponding to neighborhood data of each pixel, reflects the data uniformity distribution condition of the polarized SAR image in space, and is widely applied to algorithms such as filtering, segmentation and classification of the polarized SAR image. Therefore, the fast and effective estimation of the equivalent apparent number map of the polarized SAR image has great significance for facilitating the interpretation of the polarized SAR image.
In the last decades, many equivalent vision number estimation methods for polarized SAR images have been proposed successively, typically such as a moment estimation method, a fractional order coefficient of variance method, a trace moment estimation method, and an ML estimation method. The variance coefficient method and the fractional moment estimation method are mainly based on each single-channel polarized SAR image data for estimation, and although the method is easy to realize, the method only uses diagonal element information of a polarization matrix (a polarization covariance matrix or a polarization coherence matrix) and has insufficient effectiveness. In contrast, the moment estimation method and the ML estimation method both use all element information of the polarization matrix to estimate, and can obtain more accurate equivalent views. Compared with the moment estimation method, the ML estimation method has smaller deviation and stronger effectiveness, and is an important method for current equivalent vision number estimation. However, the ML estimation method has no analytical formula and needs to be solved by adopting a numerical method, which has the following problems in practice: because a large number of iterative numerical calculations are needed to approach the solution of the equation, the method is time-consuming; in addition, a proper interval of equivalent vision is required to be determined before solving the method, and if the set interval is too narrow to include a real solution, a reliable estimated value is difficult to obtain; if the set interval is large, this will further increase the time-consuming calculation of the numerical value. In particular, when estimating an equivalent view of a polarized SAR image, it is necessary to estimate the equivalent views corresponding to each pixel, and ML estimation is unfavorable for parallel implementation because of no resolution, and therefore, it is very time-consuming to implement and difficult to meet a large number of practical application requirements with high effectiveness requirements.
Disclosure of Invention
The technical problems to be solved by the invention are as follows:
aiming at the problems that ML estimation of the equivalent view diagram of the existing polarized SAR image is free of analytic solution and time-consuming in practical application, an estimation method capable of being rapidly realized on the basis of guaranteeing estimation precision is provided, so that a large number of practical application demands with high practical effectiveness requirements are better met.
In order to solve the technical problems, the invention adopts the following technical scheme:
a method for quickly estimating maximum likelihood of equivalent apparent number map of polarized SAR image is characterized by comprising the following steps:
step 1: computing a logarithmic determinant map of a polarized SAR image
Calculating determinant values of polarization matrixes of all pixels of the polarized SAR image based on matrix operation to form a corresponding determinant graph; obtaining a logarithmic determinant graph of polarized SAR image data by adopting logarithmic operation;
step 2: calculating local mean value graph of logarithmic determinant of polarized SAR image
Calculating a local mean value diagram of a polarized SAR image logarithmic determinant by adopting a convolution method according to the pixel neighborhood window size;
step 3: calculating local mean value graph of data of each channel of polarized SAR image
Calculating a local mean value graph of each channel data of the polarized SAR image by a convolution method according to the pixel neighborhood window size;
step 4: computing a logarithmic determinant graph of a locally averaged polarized SAR image
Based on matrix point multiplication and matrix summation operation, calculating determinant values of polarization matrixes of all pixels of the polarized SAR image after local averaging to form a corresponding determinant graph; obtaining a corresponding logarithmic determinant graph by adopting logarithmic operation;
step 5: computing a local sample statistics map of a polarized SAR image
Calculating a local sample statistic map required by the effective vision number estimation of the polarized SAR image according to the local mean value map of the logarithmic determinant of the polarized SAR image in the step 2 and the logarithmic determinant map of the polarized SAR image after the local average in the step 4;
step 6: estimating an approximately equivalent apparent map of a polarized SAR image
And solving an equivalent apparent graph of the polarized SAR image by utilizing the analytic approximation of the ML estimation.
The invention further adopts the technical scheme that: step 1, obtaining a logarithmic determinant graph of polarized SAR image data by adopting logarithmic operation specifically comprises the following steps of
wherein , wherein Zm M=1, 2, &.9 represents the polarization matrix z, i.e. the m-th element of each pixel of the polarized SAR image, & gt represents the dot multiplication of the matrix, i.e. the multiplication of elements at corresponding positions of two matrices of the same size.
The invention further adopts the technical scheme that: step 2, calculating a local mean value diagram of a polarized SAR image logarithmic determinant by adopting a convolution method, wherein the local mean value diagram specifically comprises the following steps:
wherein ,S1 ×S 2 For the pixel neighborhood window size, conv (·,) represents the convolution operation of the matrix, k is the size S 1 ×S 2 The elements are all 1/(S) 1 ×S 2 ) Is a convolution kernel of (a).
The invention further adopts the technical scheme that: step 3, calculating a local mean value graph of each channel data of the polarized SAR image by adopting a convolution method, wherein the local mean value graph specifically comprises the following steps:
the invention further adopts the technical scheme that: step 4, obtaining a corresponding logarithmic determinant graph by adopting logarithmic operation specifically comprises the following steps:
the invention further adopts the technical scheme that: step 5, calculating a local sample statistic map of polarized SAR data:
the invention further adopts the technical scheme that: step 6, estimating an approximately equivalent view of the polarized SAR image specifically performs the following operation with respect to the local sample statistic G: a is that 0 ＝4G、A 1 ＝6(3-2G)、A 2 ＝8G-37，
Wherein the dividing typeDot division of the matrix A and the matrix B is expressed;
then solving the formula according to the following formula, namely the analytic approximation of ML estimation of equivalent views, so as to obtain an equivalent view diagram of the polarized SAR image:
an error estimation method of a fast maximum likelihood estimation method of a polarimetric SAR image equivalent apparent map is characterized in that a relative estimation error REE is defined:
wherein , andEquivalent visual values of the polarized SAR image obtained by the method of claim 1 and the traditional ML estimation method respectively; REE is in units of percentage, and the value reflects the relative error between the equivalent view number obtained by the method and the equivalent view number obtained by the traditional ML estimation method; the smaller the value of REE, the closer the method of claim 1 is to the result obtained by the conventional ML method.
A computer system, comprising: one or more processors, a computer-readable storage medium storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the methods described above.
A computer readable storage medium, characterized by storing computer executable instructions that when executed are configured to implement the method described above.
The invention has the beneficial effects that:
according to the quick maximum likelihood estimation method for the equivalent view of the polarized SAR image, provided by the invention, the approximate analytic solution of the equivalent view ML estimation of the polarized SAR image is obtained by adopting the progressive of the Digamma function, and the quick calculation of local sample statistics of the equivalent view estimation of the polarized SAR data is realized by adopting matrix operation and convolution operation, so that the realization efficiency of the ML estimation of the equivalent view of the polarized SAR image is effectively improved. The method has important supporting value for high-timeliness interpretation of polarized SAR images.
Drawings
The drawings are only for purposes of illustrating particular embodiments and are not to be construed as limiting the invention, like reference numerals being used to refer to like parts throughout the several views.
FIG. 1 is a flow chart of a method for quickly implementing ML estimation of an equivalent view map of a polarized SAR image of the present invention;
fig. 2 is a Pauli-RGB pseudo-color image of a polarized SAR image used in the present invention, which is constructed with diagonal elements of a polarization coherence matrix as three primary color components of red, green, and blue, respectively.
FIG. 3 is a logarithmic display result of equivalent apparent number graphs of polarized SAR images estimated by the conventional ML method and the method of the present invention under different neighborhood window sizes; (a) The logarithmic display result of the equivalent view of the polarized SAR image obtained by the traditional ML estimation method under the condition of a 3X 3 neighborhood window; (b) The method of the invention obtains the logarithmic display result of equivalent apparent figure of polarized SAR image under the condition of 3X 3 neighborhood window; (c) The logarithmic display result of the equivalent view of the polarized SAR image obtained by the traditional ML estimation method under the condition of 7 multiplied by 7 neighborhood window; (d) The method of the invention obtains the logarithmic display result of equivalent apparent figure of polarized SAR image under the condition of 7X 7 neighborhood window; (e) The logarithmic display result of the equivalent view of the polarized SAR image obtained by the traditional ML estimation method under the condition of 11×11 neighborhood window; (f) The method of the invention displays the result of logarithm of equivalent apparent figure of the polarized SAR image obtained under the condition of 11×11 neighborhood window;
FIG. 4 is a histogram of equivalent apparent plots of the polarized SAR images of FIG. 3; (a) The traditional ML estimation method and the method of the invention obtain the histogram of equivalent apparent number of polarized SAR images under the condition of 3X 3 neighborhood window; (b) The traditional ML estimation method and the method of the invention obtain the histogram of equivalent apparent number of polarized SAR images under the condition of 7×7 neighborhood window; (c) The traditional ML estimation method and the method of the invention obtain the histogram of equivalent view of polarized SAR images under the condition of 11×11 neighborhood windows;
fig. 5 is a histogram of a relative error map between the equivalent apparent map of a polarized SAR image obtained by the conventional ML method and the method of the present invention: (a) Histogram of relative error map between equivalent apparent number map of polarized SAR image obtained by traditional ML estimation method and method of the invention under 3X 3 neighborhood window condition; (b) Histogram of relative error map between equivalent apparent number map of polarized SAR image obtained by traditional ML estimation method and method of the invention under 7X 7 neighborhood window condition; (c) Histogram of relative error map between traditional ML estimation method and equivalent apparent map of polarized SAR image obtained by the method of the invention under 11×11 neighborhood window condition.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention. In addition, technical features of the embodiments of the present invention described below may be combined with each other as long as they do not collide with each other.
According to the quick maximum likelihood estimation method for the equivalent view map of the polarized SAR image, provided by the embodiment of the invention, the Digamma function contained in the ML estimation equation of the equivalent view of the original polarized SAR image is replaced by the Digamma function, and the ML estimation approximation equation with the analytic type is deduced by discarding the higher-order term, so that the iteration numerical operation and the initial interval setting problem of the traditional method are avoided. In addition, a fast calculation method based on matrix operation and convolution is designed in the aspect of sample statistic calculation, and the efficiency advantage is particularly obvious when the equivalent apparent number diagram of the polarized SAR image is estimated.
Referring to fig. 1, in the scheme of the invention, firstly, based on matrix point multiplication and matrix summation operation, all pixel data determinant values of a polarized SAR image are calculated to form a corresponding determinant graph, and then logarithmic operation is adopted to obtain a logarithmic determinant graph of polarized SAR image data. Then, constructing a corresponding convolution kernel by the given pixel neighborhood window size, and calculating a local mean value diagram of the polarized SAR logarithmic determinant diagram by adopting a convolution method. Then, a convolution method is similarly used to calculate a local mean map of the polarized SAR data for each channel. And then, calculating determinant values of all pixel data of the polarized SAR image after local average based on matrix point multiplication and matrix summation operation to form a corresponding determinant graph, and obtaining a logarithmic determinant graph by adopting logarithmic operation. And then, obtaining a local sample statistic map of equivalent apparent number estimation of the polarized SAR data by adopting matrix subtraction. And finally, resolving by utilizing the analytic approximation of the ML estimation and based on matrix operation, so that the equivalent apparent number graph of the polarized SAR image can be rapidly estimated. The method comprises the following steps:
step one: computing a logarithmic determinant map of a polarized SAR image
First, based on matrix operation, calculating determinant values of polarization matrixes of all pixels of the polarized SAR image to form a corresponding determinant graph. And then a logarithmic operation is adopted to obtain a logarithmic determinant graph of polarized SAR image data.
Step two: calculating local mean value graph of logarithmic determinant of polarized SAR image
And (3) calculating a local mean value graph of a polarized SAR image logarithmic determinant by adopting a convolution method according to the pixel neighborhood window size.
Step three: calculating local mean value graph of data of each channel of polarized SAR image
And (3) calculating a local mean value graph of each channel data of the polarized SAR image by adopting a convolution method according to the pixel neighborhood window size.
Step four: computing a logarithmic determinant graph of a locally averaged polarized SAR image
Firstly, based on matrix point multiplication and matrix summation operation, calculating determinant values of polarization matrixes of all pixels of the polarized SAR image after local average to form a corresponding determinant graph. And then obtaining a corresponding logarithmic determinant graph by adopting logarithmic operation.
Step five: computing a local sample statistics map of a polarized SAR image
And (3) calculating a local sample statistic map required by the effective vision number estimation of the polarized SAR image according to the results of the step two and the step four.
Step six: estimating an approximately equivalent apparent map of a polarized SAR image
And solving an equivalent apparent graph of the polarized SAR image by utilizing the analytic approximation of the ML estimation.
The specific implementation steps of the present invention will be described in detail.
Step one: computing a logarithmic determinant map of a polarized SAR image
Each pixel of a multi-view polarized SAR image is typically represented by a 3 x 3 polarization matrix (polarization covariance matrix or polarization coherence matrix) z, denoted as:
wherein Zm M=1, 2,..9 represents the m-th element of the polarization matrix z.
Given a polarized SAR image with the size of MxNx9, the data is recorded as wherein Zm M=1, 2,..9 is data of the M-th channel of the polarized SAR image, i.e., matrix data of m×n size composed of the M-th element of the polarization matrix of each pixel of the polarized SAR image,Representing a concatenation of matrices.
Calculating the values of all pixel data determinant of the polarized SAR image according to the following steps to form a corresponding determinant graph:
wherein +..
Then, a logarithmic operation is adopted to obtain a logarithmic determinant graph of polarized SAR image data
Step two: calculating local mean value graph of logarithmic determinant of polarized SAR image
Given pixel neighborhood window size S 1 ×S 2 The structural size is S 1 ×S 2 The elements are all 1/(S) 1 ×S 2 ) Is the convolution kernel k of (i.e.)
Then, the polarization SAR logarithmic determinant graph Y is extended and filled (Padding) outwards in the row direction and the column directionMirror filling of both ends of the direction (S 1 -1)/2 and (S 2 -1)/2 pixel data resulting in an expanded size (m+s) 1 -1)×(N+S 2 -1) polarization SAR logarithmic determinant graphThe method is characterized by comprising the following steps:
performing convolution operation on the extended polarized SAR logarithmic determinant graph by using a convolution kernel k to obtain a local mean value graph with the size of M multiplied by N:
where Conv (·,) represents the convolution operation of the matrix.
Step three: calculating local mean value graph of data of each channel of polarized SAR image
Each channel data z of the polarized SAR image m M=1, 2,..9 extends the fill outward, mirroring the fill at both ends in the row and column directions (S 1 -1)/2 and (S 2 -1)/2 pixel data resulting in an expanded size (m+s) 1 -1)×(N+S 2 -1) polarized SAR image data:
carrying out convolution operation on the data of each channel of the extended polarized SAR image by using a convolution kernel k to obtain a corresponding local mean value graph with the size of M multiplied by N:
thereby constructing polarized SAR image data after local average
Step four: computing a logarithmic determinant graph of a locally averaged polarized SAR image
Based on matrix point multiplication and matrix summation operation, calculating determinant values of all pixel data of the polarized SAR image after local averaging, and forming a corresponding determinant graph:
then, a corresponding logarithmic determinant graph is obtained by adopting logarithmic operation
Step five: computing a local sample statistics map of a polarized SAR image
Calculating a local sample statistic graph of polarized SAR data by using the results of the second step and the fourth step:
step six: estimating an approximately equivalent apparent map of a polarized SAR image
The following operation is performed with respect to the local sample statistic G: a is that 0 ＝4G、A 1 ＝6(3-2G)、A 2 ＝8G-37，
Wherein the dividing typeThe dot division of matrix a and matrix B (the division of the two matrix corresponding elements) is expressed.
Then solving the formula according to the following formula, namely the analytic approximation of ML estimation of equivalent views, so as to obtain an equivalent view diagram of the polarized SAR image:
the effect of the invention can be further illustrated by the following simulation experiments:
1. simulation experiment content and result analysis
The invention takes an L-band polarized SAR image of san francisco area in the United states, which is acquired by an AIRSAR system shown in FIG. 2, as test data, and the size of the L-band polarized SAR image is 900 multiplied by 1024 pixels. In the test, 3 pixel neighborhood windows with different sizes, namely 3×3, 7×7 and 9×9, are set, and then equivalent vision graphs of polarized SAR images under different neighborhood window conditions are obtained by using a traditional ML estimation method and an inventive method of equivalent vision, wherein the results obtained under the 3×3, 7×7 and 9×9 neighborhood windows by the traditional ML estimation method are respectively shown in fig. 3 (a), (c) and (e), and the results obtained under the 3×3, 7×7 and 9×9 neighborhood windows by the inventive method are respectively shown in fig. 3 (b), (d) and (f). In addition, a comparison of histograms of these equivalent views under 3×3, 7×7, and 9×9 neighborhood window conditions is also presented, as shown in fig. 4 (a) - (c), respectively.
As can be seen from fig. 3, under the same conditions, the equivalent view obtained by the conventional ML estimation method and the method of the present invention look very similar, and the histograms thereof are very close, almost coincident, indicating that the values of the equivalent views estimated by the two methods are very close.
To quantitatively analyze the differences in the results obtained from these two methods, the following Relative Estimation Errors (REEs) were defined in the test:
wherein andEquivalent vision values of polarized SAR images obtained by the method and the traditional ML estimation method are respectively. REE is expressed in percentage units and its value reflects the relative error of the equivalent view number obtained by the method of the present invention and the equivalent view number obtained by the conventional ML estimation method. The smaller the REE value, the closer the results obtained by the method of the present invention to those obtained by the conventional ML method.
The histograms of REEs between equivalent views obtained by the two methods under 3×3, 7×7, and 9×9 neighborhood window conditions are shown in fig. 5 (a) - (c), respectively. It can be seen that these REEs are mainly distributed between 0.1% and 0.29, with all REEs being less than 0.29% and very small, almost negligible. These results further verify that the method of the present invention is able to obtain equivalent vision values of SAR images that are very close to conventional ML methods.
On the other hand, in order to evaluate the efficiency of the method of the present invention, the time consumed by the two methods to estimate the equivalent view of the polarized SAR image was compared under the conditions of 3×3, 7×7 and 9×9 neighborhood windows in the test experiment. For fair comparison, all experiments were run on a notebook computer with 32GB RAM and 3.70GHz Intel Kui 7-8700K CPU, all using MATLAB code. The experiment was repeated 10 times, resulting in two methods of calculating the time consumption of the sample statistics plot (step one through step five) and the ML equation solving (step six), and calculating the total time consumption thereof, as shown in table 1.
TABLE 1
As can be seen from table 1, as the neighborhood window size increases, the time for calculating the sample statistic increases, which is a result of the increase in the amount of sample that needs to be calculated. The traditional method is very time-consuming due to the adoption of a pixel-by-pixel calculation mode: when the neighborhood window sizes are 3×3, 7×7, and 11×11, respectively, it takes 27, 99, and 245 seconds to calculate the sample statistics map, respectively. In contrast, the method of the invention adopts a parallel computing mode, which realizes remarkable improvement of efficiency, takes most time under the condition of 11×11 neighborhood window, and is only less than 0.4 seconds. Furthermore, the conventional or inventive method takes approximately equal time to solve the ML equation at different neighborhood window sizes, about 19 seconds and 0.3 seconds, respectively. It is evident that the process of the present invention is more efficient than the conventional process, and takes only about 1.6% of the time of the conventional process. From a total time consuming perspective, conventional methods are particularly time consuming, especially when the neighborhood window is large, e.g. about 260 seconds is required when using an 11 x 11 window. In contrast, under the same conditions, the process of the present invention takes less than 0.8 seconds, which is less than 1/300 of the time taken by the conventional process.
FIG. 3 is a logarithmic display of equivalent apparent graphs of a polarized SAR image estimated by the conventional ML method and the method of the present invention under different neighborhood window sizes (3×3, 7×7 and 11×11). In contrast, under the same conditions, the equivalent apparent graphs of polarized SAR image data estimated by the two methods look very similar.
Fig. 4 is a histogram of equivalent apparent plots of polarized SAR image data estimated by the two methods of fig. 3. It can be seen that under the same conditions, the histograms of equivalent views obtained by the two methods are very close, with only very subtle differences.
FIG. 5 is a histogram of relative error patterns between equivalent views of polarized SAR images obtained by the conventional ML method and the method of the present invention under different neighborhood window sizes (3×3, 7×7 and 11×11). It can be seen that the relative error between the equivalent apparent number estimated by the method of the present invention and the result estimated by the standard conventional method is very small, the maximum relative error is less than 0.3% (three thousandths), and the error is almost negligible in practical application. By the method, the equivalent apparent map of the polarized SAR image almost consistent with the standard method can be obtained.
In summary, compared with the traditional ML method of the equivalent apparent number map of the polarized SAR image, the method provided by the invention has the advantages that the realization efficiency is obviously improved on the basis of ensuring the estimation precision, and the method has important significance in the practical application of high effectiveness.
While the invention has been described with reference to certain preferred embodiments, it will be understood by those skilled in the art that various changes and substitutions of equivalents may be made without departing from the spirit and scope of the invention.
Claims (10)
1. A method for quickly estimating maximum likelihood of equivalent apparent number map of polarized SAR image is characterized by comprising the following steps:
step 1: computing a logarithmic determinant map of a polarized SAR image
Calculating determinant values of polarization matrixes of all pixels of the polarized SAR image based on matrix operation to form a corresponding determinant graph; obtaining a logarithmic determinant graph of polarized SAR image data by adopting logarithmic operation;
step 2: calculating local mean value graph of logarithmic determinant of polarized SAR image
Calculating a local mean value diagram of a polarized SAR image logarithmic determinant by adopting a convolution method according to the pixel neighborhood window size;
step 3: calculating local mean value graph of data of each channel of polarized SAR image
Calculating a local mean value graph of each channel data of the polarized SAR image by a convolution method according to the pixel neighborhood window size;
step 4: computing a logarithmic determinant graph of a locally averaged polarized SAR image
Based on matrix point multiplication and matrix summation operation, calculating determinant values of polarization matrixes of all pixels of the polarized SAR image after local averaging to form a corresponding determinant graph; obtaining a corresponding logarithmic determinant graph by adopting logarithmic operation;
step 5: computing a local sample statistics map of a polarized SAR image
Calculating a local sample statistic map required by the effective vision number estimation of the polarized SAR image according to the local mean value map of the logarithmic determinant of the polarized SAR image in the step 2 and the logarithmic determinant map of the polarized SAR image after the local average in the step 4;
step 6: estimating an approximately equivalent apparent map of a polarized SAR image
And solving an equivalent apparent graph of the polarized SAR image by utilizing the analytic approximation of the ML estimation.
2. The method for fast maximum likelihood estimation of a polarimetric SAR image equivalent view according to claim 1, wherein the method comprises the following steps: step 1, obtaining a logarithmic determinant graph of polarized SAR image data by adopting logarithmic operation specifically comprises the following steps of
wherein , wherein Zm M=1, 2, …,9 represents the polarization matrix z, i.e. the m-th element of each pixel of the polarized SAR image, and by-indicates the dot multiplication of the matrix, i.e. the multiplication of elements at corresponding positions of two matrices of the same size.
3. The method for fast maximum likelihood estimation of a polarimetric SAR image equivalent view according to claim 2, wherein: step 2, calculating a local mean value diagram of a polarized SAR image logarithmic determinant by adopting a convolution method, wherein the local mean value diagram specifically comprises the following steps:
wherein ,S1 ×S 2 For the pixel neighborhood window size, conv (·,) represents the convolution operation of the matrix, k is the size S 1 ×S 2 The elements are all 1/(S) 1 ×S 2 ) Is a convolution kernel of (a).
4. A method for fast maximum likelihood estimation of a polarimetric SAR image equivalent view according to claim 3, wherein: step 3, calculating a local mean value graph of each channel data of the polarized SAR image by adopting a convolution method, wherein the local mean value graph specifically comprises the following steps:
5. the method for fast maximum likelihood estimation of a polarimetric SAR image equivalent view map of claim 4, wherein the method comprises the steps of: step 4, obtaining a corresponding logarithmic determinant graph by adopting logarithmic operation specifically comprises the following steps:
6. a polarized SAR image equivalent view according to claim 5The rapid maximum likelihood estimation method of the digital graph is characterized in that: step 5, calculating a local sample statistic map of polarized SAR data:
7. the method for fast maximum likelihood estimation of a polarimetric SAR image equivalent view map of claim 6, wherein the method comprises the steps of: step 6 estimating an approximately equivalent apparent map of the polarized SAR image, in particular
The following operation is performed with respect to the local sample statistic G: a is that 0 ＝4G、A 1 ＝6(3-2G)、A 2 ＝8G-37，
Wherein the dividing typeDot division of the matrix A and the matrix B is expressed;
then solving the formula according to the following formula, namely the analytic approximation of ML estimation of equivalent views, so as to obtain an equivalent view diagram of the polarized SAR image:
8. an error estimation method of a fast maximum likelihood estimation method of a polarimetric SAR image equivalent apparent map is characterized in that a relative estimation error REE is defined:
wherein , andEquivalent visual values of the polarized SAR image obtained by the method of claim 1 and the traditional ML estimation method respectively; REE is in units of percentage, and the value reflects the relative error between the equivalent view number obtained by the method and the equivalent view number obtained by the traditional ML estimation method; the smaller the value of REE, the closer the method of claim 1 is to the result obtained by the conventional ML method.
9. A computer system, comprising: one or more processors, a computer-readable storage medium storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method of claim 1.
10. A computer readable storage medium, characterized by storing computer executable instructions that, when executed, are adapted to implement the method of claim 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310583567.5A CN116736300A (en) | 2023-05-23 | 2023-05-23 | A fast maximum likelihood estimation method for the equivalent view number map of polarimetric SAR images |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310583567.5A CN116736300A (en) | 2023-05-23 | 2023-05-23 | A fast maximum likelihood estimation method for the equivalent view number map of polarimetric SAR images |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116736300A true CN116736300A (en) | 2023-09-12 |
Family
ID=87912429
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310583567.5A Pending CN116736300A (en) | 2023-05-23 | 2023-05-23 | A fast maximum likelihood estimation method for the equivalent view number map of polarimetric SAR images |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116736300A (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2146315A1 (en) * | 2008-07-16 | 2010-01-20 | Galileian Plus s.r.l. | Method of filtering SAR images from speckle noise and related device. |
| IL202788A0 (en) * | 2009-12-17 | 2010-11-30 | Elta Systems Ltd | Method and system fro enhancing a sar image |
| WO2011070538A2 (en) * | 2009-12-11 | 2011-06-16 | Eads Singapore Pte. Ltd | Method for despeckling of single-look dual-polarization synthetic aperture radar (sar) data |
| CN102446350A (en) * | 2011-09-16 | 2012-05-09 | 西安电子科技大学 | Anisotropic non-local mean value-based speckle suppression method for polarized SAR (Specific Absorption Rate) data |
| CN103049653A (en) * | 2012-12-17 | 2013-04-17 | 南京航空航天大学 | G0 distribution parameter maximum likelihood estimation method based on EM (Expectation-Maximization) algorithm |
| CN103093432A (en) * | 2013-01-25 | 2013-05-08 | 西安电子科技大学 | Polarized synthetic aperture radar (SAR) image speckle reduction method based on polarization decomposition and image block similarity |
| CN104103046A (en) * | 2014-07-11 | 2014-10-15 | 中国人民解放军信息工程大学 | Quick filtering processing method for polarized SAR (synthetic aperture radar) image |
| CN106093940A (en) * | 2016-06-03 | 2016-11-09 | 中国科学院电子学研究所 | Diameter radar image sequence generating method |
| US20180052224A1 (en) * | 2014-12-15 | 2018-02-22 | Airbus Singapore Private Limited | Automated method for selecting training areas of sea clutter and detecting ship targets in polarimetric synthetic aperture radar imagery |
| CN112698330A (en) * | 2020-12-14 | 2021-04-23 | 大连海事大学 | Polarized SAR image non-local mean value method based on ratio-to-digital product |
| CN115578599A (en) * | 2022-10-27 | 2023-01-06 | 西北工业大学 | A Polarimetric SAR Image Classification Method Based on Superpixel-Hypergraph Feature Enhancement Network |
-
2023
- 2023-05-23 CN CN202310583567.5A patent/CN116736300A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2146315A1 (en) * | 2008-07-16 | 2010-01-20 | Galileian Plus s.r.l. | Method of filtering SAR images from speckle noise and related device. |
| WO2011070538A2 (en) * | 2009-12-11 | 2011-06-16 | Eads Singapore Pte. Ltd | Method for despeckling of single-look dual-polarization synthetic aperture radar (sar) data |
| IL202788A0 (en) * | 2009-12-17 | 2010-11-30 | Elta Systems Ltd | Method and system fro enhancing a sar image |
| CN102446350A (en) * | 2011-09-16 | 2012-05-09 | 西安电子科技大学 | Anisotropic non-local mean value-based speckle suppression method for polarized SAR (Specific Absorption Rate) data |
| CN103049653A (en) * | 2012-12-17 | 2013-04-17 | 南京航空航天大学 | G0 distribution parameter maximum likelihood estimation method based on EM (Expectation-Maximization) algorithm |
| CN103093432A (en) * | 2013-01-25 | 2013-05-08 | 西安电子科技大学 | Polarized synthetic aperture radar (SAR) image speckle reduction method based on polarization decomposition and image block similarity |
| CN104103046A (en) * | 2014-07-11 | 2014-10-15 | 中国人民解放军信息工程大学 | Quick filtering processing method for polarized SAR (synthetic aperture radar) image |
| US20180052224A1 (en) * | 2014-12-15 | 2018-02-22 | Airbus Singapore Private Limited | Automated method for selecting training areas of sea clutter and detecting ship targets in polarimetric synthetic aperture radar imagery |
| CN106093940A (en) * | 2016-06-03 | 2016-11-09 | 中国科学院电子学研究所 | Diameter radar image sequence generating method |
| CN112698330A (en) * | 2020-12-14 | 2021-04-23 | 大连海事大学 | Polarized SAR image non-local mean value method based on ratio-to-digital product |
| CN115578599A (en) * | 2022-10-27 | 2023-01-06 | 西北工业大学 | A Polarimetric SAR Image Classification Method Based on Superpixel-Hypergraph Feature Enhancement Network |
Non-Patent Citations (5)
| Title |
|---|
| BOUHLEL, NIZAR ; MERIC, STEPHANE: "Maximum-Likelihood Parameter Estimation of the Product Model for Multilook Polarimetric SAR Data", IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, vol. 57, no. 3, 31 March 2020 (2020-03-31), pages 1596 - 1611 * |
| QIN, XIANXIANG；ZHANG, YANNING；LI, YING: "Efficient Maximum-Likelihood Estimation of Equivalent Number of Looks for PolSAR Image", IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, vol. 20, 15 September 2023 (2023-09-15) * |
| 杨卓: "基于统计分布和空间信息的极化SAR图像分类", 中国优秀硕士学位论文全文数据库 (信息科技辑), no. 2022, 15 April 2022 (2022-04-15) * |
| 赵明钧；程英蕾；秦先祥: "基于模型微调与AM-Softmax的极化SAR图像分类", 空军工程大学学报, vol. 23, no. 5, 3 November 2022 (2022-11-03), pages 36 - 43 * |
| 韦顺军;张晓玲;: "基于分类统计的PolInSAR植被高度最大似然估计", 现代雷达, vol. 31, no. 11, 30 November 2009 (2009-11-30), pages 60 - 63 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8396285B2 (en) | Estimating vanishing points in images | |
| EP2927864A1 (en) | Image processing device and image processing method | |
| CN104408707A (en) | Rapid digital imaging fuzzy identification and restored image quality assessment method | |
| US20140126808A1 (en) | Recursive conditional means image denoising | |
| Cicconet et al. | A convolutional approach to reflection symmetry | |
| US12541822B2 (en) | Method and apparatus of processing image, computing device, and medium | |
| US10540750B2 (en) | Electronic device with an upscaling processor and associated method | |
| CN107316309B (en) | A saliency target detection method for hyperspectral images based on matrix factorization | |
| CN108537728A (en) | High spectrum image super-resolution forming method and system based on spectrum fidelity | |
| CN114373135A (en) | Ship target detection method based on local saliency feature measurement | |
| CN104734724A (en) | Hyperspectral image compressed sensing method based on heavy weighting laplacian sparse prior | |
| CN102289813B (en) | A No-Reference Image Blur Evaluation Method | |
| US12033315B1 (en) | Machine vision-based techniques for non-contact structural health monitoring | |
| US20240282102A1 (en) | Surface water monitoring method and apparatus, computer device, and storage medium | |
| CN116736300A (en) | A fast maximum likelihood estimation method for the equivalent view number map of polarimetric SAR images | |
| CN105303566A (en) | Target contour clipping-based SAR image target azimuth estimation method | |
| CN112731395B (en) | SAR Imaging Method Based on Non-convex & Total Variation Regularization | |
| CN107845120A (en) | PET image reconstruction method, system, terminal and readable storage medium storing program for executing | |
| CN112817918B (en) | High-resolution three-number data conversion method, terminal equipment and storage medium | |
| CN108932520B (en) | SAR image water body probability mapping method combining prior probability estimation | |
| CN107464246A (en) | A kind of image partition method based on collection of illustrative plates dictionary learning | |
| Longbotham et al. | A family of kernel anomaly change detectors | |
| US12347184B2 (en) | Method and apparatus for suppressing noise in 3D road surface reconstruction | |
| CN106296607B (en) | Method for quickly approximating median filtering of SAR image | |
| CN117745781A (en) | Multi-angle SAR image sequence registration method, device, equipment and medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |