CN104636758B - A kind of SAR image suitability Forecasting Methodology based on support vector regression - Google Patents

Abstract

The invention discloses a radar image adaptability prediction method based on support vector regression. The method includes: in the learning stage, extracting multi-dimensional features of SAR images to form a learning set; after preprocessing the sample features of the learning set, dividing it into learning sets L ₁ and L ₂ , and then using the learning set L ₁ to train a support vector machine, and using _The obtained SVM model classifies the learning set L2, and calculates the adaptation rate of each sample according to the classification accuracy rate, sample characteristics and the distance between the centroids; then uses the learning set features and their corresponding adaptation rates to fit the regression The adaptive prediction function model is obtained; in the prediction stage, the corresponding features of the SAR image to be evaluated are extracted as test sample data, and the data is preprocessed and input into the adaptive prediction function model to calculate the adaptation rate of the image. According to the intensity and texture structure characteristics of the SAR image, the invention establishes the functional relationship between the SAR image adaptation rate and the feature information, and the method is verified by experiments to accurately evaluate the matching performance of the SAR image.

Description

A SAR Image Suitability Prediction Method Based on Support Vector Regression

technical field

The invention belongs to the technical field of machine learning, pattern recognition, and template matching, and in particular relates to a method for predicting the suitability of a Synthetic Aperture Radar (SAR) image based on support vector regression. The method considers SAR imaging characteristics and intensity structure On the basis of the texture features, the prediction of the adaptation performance of the SAR image is realized, and the prediction result of the adaptation performance obtained by this method is accurate and effective.

Background technique

The selection of SAR scene matching sub-areas is the core technology of scene matching, which is mainly to analyze, evaluate and predict the matching positioning performance (ie adaptability) of the SAR scene matching area, so as to determine whether the selected matching sub-area is suitable for matching. So far, there is no mature scheme for the selection of matching regions. Most of the tasks are done manually, which is usually difficult to conduct scientific analysis. It is difficult to manually estimate the matching performance of the selected matching subregions, which is difficult to meet the needs of practical applications. And so far, there is no method for quantitative and probabilistic prediction of matching region selection.

Scholars at home and abroad have done a lot of research on the selection of matching sub-regions, and the main methods proposed include using the similarity of scene matching sub-regions, correlation length, gray variance, cross-correlation peak characteristics, information entropy, texture energy ratio, The feature parameters of image description such as multi-resolution self-similarity measure are used to select scene matching sub-regions. However, these methods only consider the influence of a single factor on the matching performance and fix other indicators in the experiment, and lack of consideration of the correlation of these factors, resulting in poor adaptability and poor anti-interference of scene matching sub-region selection criteria.

In the existing published literature, the method of SAR image adaptation performance prediction has not yet formed a mature solution, and has been applied in engineering practice, and there is no quantitative and probabilistic prediction scheme for SAR image adaptation.

Contents of the invention

The present invention aims at the SAR image adaptation performance evaluation problem in the SAR scene matching system, and proposes a SAR image adaptation performance prediction method based on a support vector regression machine, which specifically includes:

(1) Extract the light and dark object density and structurally significant intensity features of the SAR training image, and the sample information corresponding to each SAR image is composed of the feature set and the given positive and negative category attributes, and the sample information corresponding to all SAR training images constitutes a learning set;

(2) Preprocessing the learning sample feature data, that is, removing the coupling relationship for each two-dimensional feature of the sample set data in the learning set, and normalizing the features after the coupling relationship is removed;

(3) Divide the learning set after data preprocessing into learning set L ₁ and learning set L ₂ , use the samples in learning set L ₁ to train the support vector machine, and obtain the SVM classifier model of positive/negative attribute samples, And the Gaussian distribution characteristics of the positive/negative sample features; use the learning set L ₂ samples to test the performance of the classifier, and count the category attributes of each sample after being classified by the SVM classifier model. According to the given positive and negative category attribute information, calculate The probability P ₊ , P _- of the positive/negative sample centroid feature in the learning set L ₁ belongs to the positive/negative sample;

(4) Using the learning set L ₁ positive/negative sample centroid feature and its corresponding probability of belonging to the positive/negative sample, and the Gaussian distribution characteristics of each dimension feature of the learning set L ₁ positive/negative samples to obtain the learning set Each dimension feature of each sample belongs to the mapping relationship of the probability of positive/negative samples, thus calculating the probability p _j ⁺ , p _j ^- of each dimension feature of each sample in each learning set L ₂ samples belonging to the positive/negative category, and then calculating the learning set L ₂ The adaptation rate p _{j_match} of each dimension feature of the sample;

(5) Through the control variable method and the classification accuracy P _{(j,k) of} the corresponding SVM model classification learning set L2, calculate the sensitivity of each dimension feature of the learning set L2 to the adaptability _;

(6) According to the learning set L obtained in step (4) ₂ each dimension feature adaptation rate and the learning set L obtained in step (5) The sensitivity of each dimension feature of the sample is calculated to obtain the learning set L ₂ sample adaptation rate _; The sample adaptation rate of the set L ₂ and its feature information of each dimension constitute the new sample information of the learning set L ₂ ;

( ₇ ) learning set L2 new sample information obtained by step (5), fitting regression to obtain the image adaptability prediction function model;

(8) For the SAR image to be evaluated, according to the method of steps (1) and (2), extract the feature information of each dimension of the SAR image and perform data preprocessing, and the processed data pass the sample adaptability prediction model of step (7) The adaptation rate of the SAR image to be evaluated is predicted.

Compared with prior art, technical effect of the present invention is reflected in:

In the prior art, the method for evaluating the matching performance of SAR images has not yet formed a mature solution. Most of the tasks are done manually, and it is usually difficult to conduct scientific analysis. It is difficult to manually estimate the matching performance of the selected matching sub-region. meet the needs of practical applications. The present invention obtains the support vector regression machine model by training, establishes the functional relationship between the adaptation rate of the SAR image and the feature information, predicts the adaptation performance of the SAR sub-area, and expands the field to probability prediction, making the result more accurate . It overcomes the defects of artificial subjective evaluation of screening sub-regions, improves the stability, and improves the quality of the screened SAR matching sub-regions.

The invention provides a SAR image matching performance evaluation method based on a support vector regression machine. The method extracts features from the SAR image, trains a support vector machine model, and uses the support vector machine to classify learning samples. According to the classification result and the sample Gaussian distribution characteristics The fitting rate of each sample is obtained, and then the support vector regression machine is trained by using the sample data with the fitting rate to fit the regression prediction function model; finally, the function model is used to evaluate the SAR image to be evaluated to obtain the fitting rate. According to the SAR imaging characteristics and sub-area structural strength and texture features, the present invention comprehensively uses a variety of machine learning and pattern recognition methods, realizes the adaptability prediction of SAR images, forms a set of systematic SAR image adaptability prediction methods, and Extending the field of matching area selection to probability prediction effectively improves the method based on manual screening and improves the accuracy of screening SAR matching sub-areas, which is of great significance to the research and development of SAR matching sub-area selection.

Description of drawings

Fig. 1 is the overall flowchart of the SAR image adaptability method based on SVM of the present invention;

Fig. 2 is a partial SAR image in the embodiment of the present invention;

Fig. 3 is some SAR images to be evaluated in the embodiment of the present invention;

FIG. 4 is a prediction result diagram of SAR image adaptation probability predicted in an embodiment of the present invention;

Fig. 5 is a diagram of the predicted adaptation rate and the corresponding SAR image verification result in the embodiment of the present invention, wherein:

Figure 5(a) shows the SAR image with poor matching performance and the predicted fitting rate (0≤p<0.4);

Figure 5(b) shows the SAR image with moderate matching performance and the predicted matching rate (0.4≤p<0.7);

Figure 5(c) shows the SAR image with strong matching performance and the predicted matching rate (0.7≤p≤1).

detailed description

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not constitute a conflict with each other.

The present invention provides a SAR image adaptability prediction method based on a support vector regression machine, and its overall flow is as shown in Figure 1, and the specific process of the method is:

1 learning stage

1.1 Data preparation stage

Extract the multi-dimensional features of the intensity and texture structure of the SAR image, such as uniformity, divergence, light and dark target density, and significant structural strength, and calibrate the matching results of the SAR image and the real-time image as the category attribute, and the multi-dimensional feature and category attribute as the corresponding SAR image Sample information, the samples corresponding to all SAR images constitute a sample set. The present invention selects the two-dimensional features and category attributes of light and dark object density and structural salience strength as sample information.

1.1.1 Feature Extraction

As shown in FIG. 1 , it is a part of SAR training sample images in the embodiment of the present invention. Extract the multi-dimensional feature information of each SAR image, such as divergence, uniformity, light and dark target density, and significant structural strength;

Uniformity r:

Among them, where μ is the mean value of the gray level, and σ is the standard deviation of the gray level.

divergence div:

Among them, σ ₁ respectively represents the standard deviation of the pixel set whose gray value is less than the mean value μ of the gray value of the SAR image. Correspondingly, _σ2 represents the standard deviation of the pixel set whose gray value is greater than the mean value μ of the gray value of the SAR image.

Light and dark target density: the proportion of bright and dark target pixels. Bright target pixels represent points whose gray value is greater than 2/3 of the full image gray value of the SAR image, and dark target pixels represent points whose gray value is smaller than the full image gray value of the SAR image. Points worth 1/3.

Structural salience strength: the ratio of the total number of marked pixels to the mean value of image width, height, and width after binary edge extraction is performed on the radar image, and the connected domain marking method is used to remove the noise of the connected domain marking with a small number of pixels.

The present invention finally selects the two-dimensional features of light and dark target density and structural conspicuous strength through experiments.

1.1.2 Positive and negative category attributes

Mark the category attribute of each sample according to the matching results of the SAR image and the real-time image. If it is suitable for matching, it will be marked as +1, which is a sample in the adaptation area; if it is not suitable for matching, it will be marked as -1, and it is a sample in the non-adaptation area .

The two-dimensional feature information of bright and dark target density, structural salience intensity and positive and negative category attributes of the SAR image are extracted as the sample information corresponding to each image, and the samples corresponding to all SAR images constitute the sample set.

1.2 Data preprocessing

Perform data preprocessing on the feature data of the sample set, that is, remove the coupling relationship for each two-dimensional feature of the feature data of the sample set, and normalize according to the dimension feature;

1.2.1 Decoupling relationship

Analyzing the relationship distribution between each two-dimensional feature in the sample space, it can be seen that the linear relationship between each two-dimensional feature is not obvious, and it can be considered as a nonlinear correlation. We use Schmidt orthogonalization to remove the coupling relationship between features. Here, the feature matrix Z={Z ₁ , Z ₂ , . . . , Z _m } is used to represent the m-dimensional feature of the sample set.

If the vectors Z ₁ , Z ₂ ,...,Z _m are linearly independent, the reference vector is b ₀ =(1,0,...,0) ^T , let

Then b ₁ , b ₂ ,..., b _m are orthogonal to each other and become an orthogonal vector group, which is unitized to obtain a matrix X, where X={X ₁ , X ₂ ,...,X _m };

Then the process from matrix Z to matrix X is called Schmidt orthogonalization. After the original data is processed based on the above-mentioned Schmidt orthogonalization, the coupling relationship between the characteristics of each dimension of the sample can be removed, and it has the same reference vector, which is convenient for considering the impact of the change of a certain dimension characteristic on the matching accuracy later.

1.2.2 Normalization of features in each dimension

After decoupling, the m-dimensional feature set of n samples is expressed as:

In the above formula, x _ij represents the j-th dimension feature vector value of the i-th sample, the sample is normalized according to the dimension, and the j-th dimension feature x _ij of the sample i is normalized to y _ij , then the mapping of x→y The relationship is as follows:

[y _min ,y _max ] is the set normalization interval, x _max =max(x _1j ,x _2j ,…x _nj ) ^T , x _min =min(x _1j ,x _2j ,…x _nj ) ^T .

Then the normalized feature matrix of the sample set is

In the above formula, y _nm represents the m-th dimension eigenvector value of the n-th sample after normalization. After normalization in the present invention, all data are in the range of [0,1].

1.3 The probability that each dimension feature belongs to a positive/negative sample

1.3.1 Training Support Vector Machine

Due to the difference between samples, the processed sample set data is divided into learning set L ₁ and learning set L ₂ , using the learning set L ₁ samples to train Support Vector Machine (Support Vector Machine, SVM), and the Gaussian kernel in SVM The function c and the penalty coefficient g are optimized for parameters, and the best parameters c and g are selected to train the SVM model. Use the learning set L ₂ samples to test the performance of the classifier, count the category attributes after the classification of the obtained SVM classifier model, and refer to the given positive and negative category attribute information to obtain the positive/negative classification accuracy rate, which is the learning set L ₁ Positive/negative sample centroid features belong to positive/negative sample probabilities P ₊ , P _- .

Suppose the number of samples with positive attributes in the learning set L ₂ is n ₁ , and the number of samples with negative attributes is n ₂ , after being classified by the SVM classifier trained on the learning set L ₁ , there are k ₁ positive attribute samples that are divided into negative samples , there are k ₂ samples with negative attributes are divided into positive samples, and the eigenvectors defining the centroids of the learning set L ₁ positive samples are The characteristics of the learning set L ₁ negative sample centroids are Among them, x _i ⁺ , x _i ^- are the characteristics of the positive and negative samples of the learning set L ₁ respectively, then the probability P ₊ , P _- of the positive/negative sample centroid features of the learning set L ₁ belonging to the positive and negative samples is:

1.3.2 Probability _{p j} ⁺

Let the mean characteristic of the center _of positive attribute samples in the learning set L1 be The corresponding probability of belonging to a positive sample is P ₊ , where x _i ⁺ = (x _i1 , x _i2 ,… _xij ) is a positive sample, and the sample features obey the Gaussian distribution x ₊ ~N(μ ⁺ ,σ ⁺² ). Assuming that the sample characteristics of the learning set and the probability of belonging to the positive sample are in a mapping relationship, according to the Gaussian distribution of the positive sample, the probability p _j ⁺ of the j-th dimension feature of the L ₂ sample belonging to the positive sample is:

in, is the mean value of the j-th dimension feature of the positive sample, is the variance of the j-th dimension feature of the positive sample, and C ₁ and C ₂ are linear coefficients;

When x=μ _j , it corresponds to the probability P ₊ that the positive class centroid belongs to the positive sample;

When x=+∞, the probability of belonging to the positive sample is the highest at this time, and it is considered to be 1 at this time;

When x=-∞, the probability of belonging to the positive sample is the lowest at this time, and it is considered to be 0 at this time.

Substituting this function with x=μ _j into the above formula, we get

1.3.3 _The probability that each dimension feature of the L2 sample belongs to a negative sample

In the same way as in 1.3.2, let the learning set L ₁ negative attribute sample centroid feature be The corresponding probability of belonging to a negative sample is P _- , where x _i ^- = (x _i1 , x _i2 ,…x _ij ) is a negative sample, and the sample features obey the Gaussian distribution x _- ~N(μ ^- ,σ ^-2 ). Assuming that the sample characteristics of the learning set and the probability of belonging to the negative sample are in a mapping relationship, according to the Gaussian distribution of the negative sample, the probability of the j _- th dimension feature of the learning set L2 sample belonging to the negative sample can be obtained for

in, is the centroid of the j-th dimension feature of the negative sample, is the variance of the j-th dimension feature of the negative sample, and C ₁ ', C ₂ ' are linear coefficients;

When x=μ _- , it corresponds to the probability P _- that the negative class centroid belongs to the negative sample;

When x=-∞, the probability of belonging to the negative sample is the highest at this time, and it is considered to be 1 at this time;

When x=+∞, the probability of belonging to the negative sample is the lowest at this time, and it is considered to be 0 at this time.

replace this function with Segmentation, into the above formula

1.3.4 The adaptation rate p _j of each dimension feature of L ₂ samples:

Through this step, the probability P _j ⁺ , P _j ^- of the j-th dimension feature of the L ₂ sample belonging to the positive/negative sample, and the adaptation rate p _{j_match} of each dimension feature of the L ₂ sample are obtained.

1.4 Sensitivity of features in each dimension

Through the control variable method and the classification accuracy P _{(j,k) of} the corresponding SVM model classification learning set L2, calculate the sensitivity of each dimension feature of the learning set L2 sample to the adaptability _;

Considering that different sample features have different effects on matching performance, we introduce the concept of "sensitivity", as follows:

Change the feature vector x _j of the j-th dimension of the sample matrix of the learning set L ₂ , while controlling the remaining feature vectors of the m-1 dimension to remain unchanged. When x _j =(k,k,k,…k) ^T , the sample matrix is

k changes from 0, 0.1, 0.2...1.0 in turn, after preprocessing the sample matrix, use the previous SVM training model to classify X', and the number of correctly classified samples is n _(j,k) , and the total number of samples is n, then the classification accuracy at this time Record the maximum value P _jmax and minimum value P _jmin of its classification accuracy:

P _jmax ＝max{P _(j,k) |k＝0,0.1,...,1.0}

P _jmin ＝min{P _(j,k) |k＝0,0.1,...,1.0}

The normalized value of the m dimensions is used as the sensitivity of the feature adaptation rate of the corresponding dimension, that is, the contribution accuracy or weight of the final adaptation rate.

Among them, W _j is the weight corresponding to the feature vector of the jth dimension.

1.5 Calculate the fitting rate of sample i

According to the probability of each dimension feature vector of the learning set _L2 sample obtained in step 1.3 belonging to positive/negative samples and the sensitivity of each dimension feature vector of the learning set _L2 sample obtained in step 1.4, the _adaptation of each sample of the learning set L2 is calculated Rate, and the adaptation rate of the learning set L ₂ samples and its feature information of each dimension constitute the new learning set L ₂ sample information;

Then the matching accuracy of the samples in the final learning set L ₂ is

a+b=1

Among them, m represents the dimension of the sample feature vector, W _j represents the sensitivity of the sample matching performance to the j-dimensional feature vector, P _j ⁺ represents the probability that the j-dimensional feature vector of the sample belongs to a positive sample, and P _j ^- represents the j-dimensional feature of the sample The probability that the vector belongs to the negative sample, when P _j ⁺ →1, P _j ^- →0; when P _j ⁺ →0, P _j ^- →1; the calculation can be a=0.5, b=0.5, then

1.6 Regression function model

From the new learning set L ₂ sample set information obtained in step 1.5, fit the regression to obtain the image adaptability prediction function model.

For the learning set samples, we use the feature x _i of each sample in the learning set L ₂ and the adaptation rate P _i calculated in step 1.5 to form a new learning set L ₂ sample, and the attribute information of the sample points can be expressed as ( _xi , P _i ). Use the method of step 1.3.1 to optimize the parameters of SVM parameters c and g, and select the parameters c and g that make the classification performance of the SVM classifier the best. Here, the libsvm library function of Professor Lin Zhiren from Taiwan is used to assist in the establishment of the sample adaptation rate prediction model. Using the new learning set L ₂ samples obtained earlier to train the SVM classifier model, the adaptation of the learning set L ₂ samples can be obtained. Sex prediction model F(x).

We used mean squared error (MSE) and squared correlation coefficient (r ² ) to verify the performance of the predictive model.

Among them, y _i represents the input adaptation rate of training sample i, f( _xi ) represents the predicted adaptation rate of training sample i, and l represents the number of training samples.

2 Prediction stage

The regression function model is used to predict the adaptation rate of the SAR sub-image to be evaluated.

As shown in Figure 3, it is part of the SAR images to be evaluated in the embodiment of the present invention. For the SAR images to be evaluated, according to the method of step 1.1 in the learning stage, extract the light and dark target density and structurally significant intensity features of the SAR training image, and then follow the steps The method of 1.2 is used for data preprocessing, and the processed data is used to predict the adaptation rate of the SAR image to be evaluated through the sample adaptation rate regression prediction model in step 1.6, namely:

P _{MatchProbability} = F(x ₁ ,x ₂ ,...,x _m )

Among them, P _{MatchProbability} represents the fitting rate of the sample, and x _m represents the m-th dimension feature of the sample.

As shown in Figure 4, it is the prediction result of the fitting rate of the SAR image to be evaluated in the embodiment of the present invention, wherein samples 1-12 are SAR images with poor matching performance (0≤p<0.4); samples 13-20 are matching Moderate (0.4≤p<0.7) SAR images; samples 21-31 are SAR images with strong matching (0.7≤p≤1). Figure 5 shows the predicted adaptation rate and the corresponding SAR image verification results in the embodiment of the present invention, where Figure 5(a) shows the SAR image with poor matching performance and the predicted adaptation rate (0≤p<0.4), that is, sample 1 The matching rate of ~12 and the corresponding SAR image verification results, Figure 5(b) shows the SAR image with moderate matching performance and the predicted matching rate (0.4≤p<0.7), that is, the matching rate of samples 13-20 and the corresponding SAR image verification results, Figure 5(c) shows the SAR images with strong matching performance and the predicted adaptation rate (0.7≤p≤1), that is, the adaptation rate of samples 21-31 and the corresponding SAR image verification results.

Claims (8)

1. A SAR image adaptability prediction method based on support vector regression is characterized by comprising the following steps:

(1) extracting light and shade target density and structure significant intensity characteristics of SAR training images, forming sample information corresponding to each SAR image by a characteristic set and given positive and negative category attributes, and forming a learning set by the sample information corresponding to all the SAR training images;

(2) preprocessing the characteristic data in the learning set, namely removing the coupling relation of each two-dimensional characteristic of the characteristic data in the learning set, and normalizing the characteristic with the coupling relation removed according to the dimension characteristic;

(3) dividing the learning set after data preprocessing into a learning set L₁And a learning set L₂Using learning set L₁The sample training support vector machine in the method obtains SVM classifier models of positive/negative attribute samples and Gaussian distribution characteristics of positive/negative sample characteristics; using learning sets L₂The sample tests the performance of the classifier, the category attribute of each sample after being classified by the SVM classifier model is counted, and the learning set L is calculated according to the given positive and negative category attribute information₁Probability P that class center feature of middle positive/negative sample belongs to positive/negative sample₊、P_-；

(4) Using learning sets L₁Class-heart feature of positive/negative sample and corresponding probability of belonging to positive/negative sample, and learning set L₁The Gaussian distribution characteristics of each dimension characteristic of the medium positive/negative samples are obtained, the mapping relation of the probability that each dimension characteristic of each sample of the learning set belongs to the positive/negative sample is obtained, and therefore L of each learning set is calculated₂Probability p of each dimension feature of sample belonging to positive/negative category_j ⁺、p_j ^-Then calculates the learning set L₂Adaptation rate p of each dimension characteristic in sample_{j_match}J represents the dimension serial number of the characteristic vector, the value of j is 1 to m, and m represents the dimension of the sample characteristic vector;

(5) classification learning set L by controlling variable method and corresponding SVM model₂Is classified with accuracy P_(j,k)Calculating a learning set L₂Sensitivity of each dimension feature to adaptability, wherein k represents the value of each element in a j-dimension feature vector, and k sequentially changes from 0,0.1,0.2 … 1.0.0;

(6) the learning set L obtained according to the step (4)₂Adaptation rate of each dimension characteristic and learning set L obtained in step (5)₂The sensitivity of each dimension characteristic of the sample is calculated to obtain a learning set L₂The adaptation rate of the sample; learning set L₂The sample adaptation rate and the characteristic information of each dimension form a learning set L₂New sample information;

(7) new learning set L obtained from step (6)₂The information on the samples is used to determine,fitting regression to obtain an image suitability prediction function model;

(8) and (3) for the SAR image to be evaluated, extracting the corresponding characteristics of the SAR image to be evaluated according to the methods in the steps (1) and (2), preprocessing data, and predicting the adaptation rate of the SAR image to be evaluated through the adaptability prediction model in the step (7) by the processed data.

2. The method of claim 1, wherein the step (2) of removing the coupling relation for each two-dimensional feature of the feature data in the learning set is specifically:

let the characteristic matrix Z ═ Z₁,Z₂,...,Z_mDenotes the m-dimensional features of the learning set, with the reference vector b₀＝(1,0,...,0)^TLet us order

<mrow> <msub> <mi>b</mi> <mn>1</mn> </msub> <mo>=</mo> <msub> <mi>Z</mi> <mn>1</mn> </msub> <mo>-</mo> <mfrac> <mrow> <mo>(</mo> <msub> <mi>Z</mi> <mn>1</mn> </msub> <mo>,</mo> <msub> <mi>b</mi> <mn>0</mn> </msub> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msub> <mi>b</mi> <mn>0</mn> </msub> <mo>,</mo> <msub> <mi>b</mi> <mn>0</mn> </msub> <mo>)</mo> </mrow> </mfrac> <msub> <mi>b</mi> <mn>0</mn> </msub> </mrow>

<mrow> <msub> <mi>b</mi> <mn>2</mn> </msub> <mo>=</mo> <msub> <mi>Z</mi> <mn>2</mn> </msub> <mo>-</mo> <mfrac> <mrow> <mo>(</mo> <msub> <mi>Z</mi> <mn>2</mn> </msub> <mo>,</mo> <msub> <mi>b</mi> <mn>1</mn> </msub> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msub> <mi>b</mi> <mn>1</mn> </msub> <mo>,</mo> <msub> <mi>b</mi> <mn>1</mn> </msub> <mo>)</mo> </mrow> </mfrac> <msub> <mi>b</mi> <mn>1</mn> </msub> <mo>-</mo> <mfrac> <mrow> <mo>(</mo> <msub> <mi>Z</mi> <mn>2</mn> </msub> <mo>,</mo> <msub> <mi>b</mi> <mn>0</mn> </msub> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msub> <mi>b</mi> <mn>0</mn> </msub> <mo>,</mo> <msub> <mi>b</mi> <mn>0</mn> </msub> <mo>)</mo> </mrow> </mfrac> <msub> <mi>b</mi> <mn>0</mn> </msub> </mrow>

...

<mrow> <msub> <mi>b</mi> <mi>m</mi> </msub> <mo>=</mo> <msub> <mi>Z</mi> <mi>m</mi> </msub> <mo>-</mo> <mfrac> <mrow> <mo>(</mo> <msub> <mi>Z</mi> <mi>m</mi> </msub> <mo>,</mo> <msub> <mi>b</mi> <mn>0</mn> </msub> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msub> <mi>b</mi> <mn>0</mn> </msub> <mo>,</mo> <msub> <mi>b</mi> <mn>0</mn> </msub> <mo>)</mo> </mrow> </mfrac> <msub> <mi>b</mi> <mn>0</mn> </msub> <mo>-</mo> <mfrac> <mrow> <mo>(</mo> <msub> <mi>Z</mi> <mi>m</mi> </msub> <mo>,</mo> <msub> <mi>b</mi> <mn>1</mn> </msub> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msub> <mi>b</mi> <mn>1</mn> </msub> <mo>,</mo> <msub> <mi>b</mi> <mn>1</mn> </msub> <mo>)</mo> </mrow> </mfrac> <msub> <mi>b</mi> <mn>1</mn> </msub> <mo>-</mo> <mfrac> <mrow> <mo>(</mo> <msub> <mi>Z</mi> <mi>m</mi> </msub> <mo>,</mo> <msub> <mi>b</mi> <mn>2</mn> </msub> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msub> <mi>b</mi> <mn>2</mn> </msub> <mo>,</mo> <msub> <mi>b</mi> <mn>2</mn> </msub> <mo>)</mo> </mrow> </mfrac> <msub> <mi>b</mi> <mn>2</mn> </msub> <mo>-</mo> <mo>...</mo> <mo>-</mo> <mfrac> <mrow> <mo>(</mo> <msub> <mi>Z</mi> <mi>m</mi> </msub> <mo>,</mo> <msub> <mi>b</mi> <mrow> <mi>m</mi> <mo>-</mo> <mn>1</mn> </mrow> </msub> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msub> <mi>b</mi> <mrow> <mi>m</mi> <mo>-</mo> <mn>1</mn> </mrow> </msub> <mo>,</mo> <msub> <mi>b</mi> <mrow> <mi>m</mi> <mo>-</mo> <mn>1</mn> </mrow> </msub> <mo>)</mo> </mrow> </mfrac> <msub> <mi>b</mi> <mrow> <mi>m</mi> <mo>-</mo> <mn>1</mn> </mrow> </msub> </mrow>

Then b is₁，b₂，…，b_mForming orthogonal vector groups by pairwise orthogonality, and unitizing the orthogonal vector groups to obtain a matrix X, wherein X is { X1, X2, … and Xm }, and n is the total number of samples;

<mrow> <mi>X</mi> <mo>=</mo> <mfenced open = "[" close = "]"> <mtable> <mtr> <mtd> <msub> <mi>x</mi> <mn>11</mn> </msub> </mtd> <mtd> <msub> <mi>x</mi> <mn>12</mn> </msub> </mtd> <mtd> <mn>...</mn> </mtd> <mtd> <msub> <mi>x</mi> <mrow> <mn>1</mn> <mi>m</mi> </mrow> </msub> </mtd> </mtr> <mtr> <mtd> <msub> <mi>x</mi> <mn>21</mn> </msub> </mtd> <mtd> <msub> <mi>x</mi> <mn>22</mn> </msub> </mtd> <mtd> <mn>...</mn> </mtd> <mtd> <msub> <mi>x</mi> <mrow> <mn>2</mn> <mi>m</mi> </mrow> </msub> </mtd> </mtr> <mtr> <mtd> <mrow></mrow> </mtd> <mtd> <mrow></mrow> </mtd> <mtd> <mrow></mrow> </mtd> <mtd> <mo>.</mo> </mtd> </mtr> <mtr> <mtd> <mrow></mrow> </mtd> <mtd> <mrow></mrow> </mtd> <mtd> <mrow></mrow> </mtd> <mtd> <mo>.</mo> </mtd> </mtr> <mtr> <mtd> <mrow></mrow> </mtd> <mtd> <mrow></mrow> </mtd> <mtd> <mrow></mrow> </mtd> <mtd> <mo>.</mo> </mtd> </mtr> <mtr> <mtd> <msub> <mi>x</mi> <mrow> <mi>n</mi> <mn>1</mn> </mrow> </msub> </mtd> <mtd> <msub> <mi>x</mi> <mrow> <mi>n</mi> <mn>2</mn> </mrow> </msub> </mtd> <mtd> <mn>...</mn> </mtd> <mtd> <msub> <mi>x</mi> <mrow> <mi>n</mi> <mi>m</mi> </mrow> </msub> </mtd> </mtr> </mtable> </mfenced> </mrow>

<mrow> <mtable> <mtr> <mtd> <mrow> <msub> <mi>X</mi> <mn>1</mn> </msub> <mo>=</mo> <mfrac> <msub> <mi>b</mi> <mn>1</mn> </msub> <mrow> <mo>|</mo> <mo>|</mo> <msub> <mi>b</mi> <mn>1</mn> </msub> <mo>|</mo> <mo>|</mo> </mrow> </mfrac> </mrow> </mtd> <mtd> <mrow> <msub> <mi>X</mi> <mn>2</mn> </msub> <mo>=</mo> <mfrac> <msub> <mi>b</mi> <mn>2</mn> </msub> <mrow> <mo>|</mo> <mo>|</mo> <msub> <mi>b</mi> <mn>2</mn> </msub> <mo>|</mo> <mo>|</mo> </mrow> </mfrac> </mrow> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <mrow> <msub> <mi>X</mi> <mi>m</mi> </msub> <mo>=</mo> <mfrac> <msub> <mi>b</mi> <mi>m</mi> </msub> <mrow> <mo>|</mo> <mo>|</mo> <msub> <mi>b</mi> <mi>m</mi> </msub> <mo>|</mo> <mo>|</mo> </mrow> </mfrac> </mrow> </mtd> </mtr> </mtable> <mo>.</mo> </mrow>。

3. the method according to claim 1 or 2, wherein the normalization of the features after the coupling relationship is removed in the step (2) according to the dimensional features specifically comprises:

let the i j dimension of the sample be the feature x_ijNormalized to y_ijThen the mapping of x → y is as follows:

<mrow> <msub> <mi>y</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> <mo>=</mo> <mrow> <mo>(</mo> <msub> <mi>y</mi> <mrow> <mi>m</mi> <mi>a</mi> <mi>x</mi> </mrow> </msub> <mo>-</mo> <msub> <mi>y</mi> <mi>min</mi> </msub> <mo>)</mo> </mrow> <mfrac> <mrow> <msub> <mi>x</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> <mo>-</mo> <msub> <mi>x</mi> <mi>min</mi> </msub> </mrow> <mrow> <msub> <mi>x</mi> <mrow> <mi>m</mi> <mi>a</mi> <mi>x</mi> </mrow> </msub> <mo>-</mo> <msub> <mi>x</mi> <mi>min</mi> </msub> </mrow> </mfrac> <mo>+</mo> <msub> <mi>y</mi> <mi>min</mi> </msub> </mrow>

[y_min,y_max]for a set normalization interval, x_max＝max(x_1j,x_2j,…x_nj)^T，x_min＝min(x_1j,x_2j,…x_nj)^T；

The normalized sample set feature matrix is

<mrow> <mi>Y</mi> <mo>=</mo> <mfenced open = "[" close = "]"> <mtable> <mtr> <mtd> <msub> <mi>y</mi> <mn>11</mn> </msub> </mtd> <mtd> <msub> <mi>y</mi> <mn>12</mn> </msub> </mtd> <mtd> <mn>...</mn> </mtd> <mtd> <msub> <mi>y</mi> <mrow> <mn>1</mn> <mi>m</mi> </mrow> </msub> </mtd> </mtr> <mtr> <mtd> <msub> <mi>y</mi> <mn>21</mn> </msub> </mtd> <mtd> <msub> <mi>y</mi> <mn>22</mn> </msub> </mtd> <mtd> <mn>...</mn> </mtd> <mtd> <msub> <mi>y</mi> <mrow> <mn>2</mn> <mi>m</mi> </mrow> </msub> </mtd> </mtr> <mtr> <mtd> <mrow></mrow> </mtd> <mtd> <mrow></mrow> </mtd> <mtd> <mrow></mrow> </mtd> <mtd> <mo>.</mo> </mtd> </mtr> <mtr> <mtd> <mrow></mrow> </mtd> <mtd> <mrow></mrow> </mtd> <mtd> <mrow></mrow> </mtd> <mtd> <mo>.</mo> </mtd> </mtr> <mtr> <mtd> <mrow></mrow> </mtd> <mtd> <mrow></mrow> </mtd> <mtd> <mrow></mrow> </mtd> <mtd> <mo>.</mo> </mtd> </mtr> <mtr> <mtd> <msub> <mi>y</mi> <mrow> <mi>n</mi> <mn>1</mn> </mrow> </msub> </mtd> <mtd> <msub> <mi>y</mi> <mrow> <mi>n</mi> <mn>2</mn> </mrow> </msub> </mtd> <mtd> <mn>...</mn> </mtd> <mtd> <msub> <mi>y</mi> <mrow> <mi>n</mi> <mi>m</mi> </mrow> </msub> </mtd> </mtr> </mtable> </mfenced> </mrow>

Wherein, y_nmAnd representing the m-dimension characteristic vector value of the n-th sample after normalization.

4. The method according to claim 1 or 2, wherein the learning set L is calculated in the step (3)₁Probability P that class-center feature of positive/negative sample belongs to positive and negative sample₊、P_-The method specifically comprises the following steps:

set learning set L₂The number of samples with positive attribute is n₁The number of samples with negative attribute is n₂Learning set L₁After the trained SVM classifier is classified, k is₁A positive attribute sample is divided into negative samples with k₂Dividing the negative attribute sample into positive samples, defining a learning set L₁The class center of the positive sample is characterized in thatn represents the total number of samples, learning set L₁The class center of the negative sample is characterized in thatWherein x is_i ⁺、x_i ^-Are respectively a learning set L₁Features of the positive and negative samples, then the learning set L₁The class center feature of the positive/negative sample belongs to the probability P of the positive and negative samples₊、P_-Comprises the following steps:

<mrow> <msub> <mi>P</mi> <mo>+</mo> </msub> <mo>=</mo> <mfrac> <mrow> <msub> <mi>n</mi> <mn>1</mn> </msub> <mo>-</mo> <msub> <mi>k</mi> <mn>1</mn> </msub> </mrow> <msub> <mi>n</mi> <mn>1</mn> </msub> </mfrac> </mrow>

<mrow> <msub> <mi>P</mi> <mo>-</mo> </msub> <mo>=</mo> <mfrac> <mrow> <msub> <mi>n</mi> <mn>2</mn> </msub> <mo>-</mo> <msub> <mi>k</mi> <mn>2</mn> </msub> </mrow> <msub> <mi>n</mi> <mn>2</mn> </msub> </mfrac> <mo>.</mo> </mrow>

5. the method according to claim 1 or 2, wherein the learning set L is obtained in the step (4) as follows₂Adaptation rate of each dimension characteristic of the sample:

(4.1)L₂probability p that each dimension feature of a sample belongs to a positive sample_j ⁺；

Is provided with L₁The mean value of the class centers of the positive attribute samples is characterized asCorresponding to a probability of belonging to a positive sample of P₊Wherein x is_i ⁺＝(x_i1,x_i2,…x_ij) For positive samples, the sample characteristics obey a Gaussian distribution x₊～N(μ⁺,σ⁺2) (ii) a Assuming that the sample characteristics of the learning set and the probability of the learning set belonging to the positive sample are in a mapping relation, and obtaining L according to the Gaussian distribution of the positive sample₂Sample j-th dimension feature belongs to positive sample probability p_j ⁺Comprises the following steps:

<mrow> <msup> <msub> <mi>p</mi> <mi>j</mi> </msub> <mo>+</mo> </msup> <mo>=</mo> <msub> <mi>C</mi> <mn>1</mn> </msub> <mo>*</mo> <mfrac> <mn>1</mn> <mrow> <msqrt> <mrow> <mn>2</mn> <mi>&amp;pi;</mi> </mrow> </msqrt> <msubsup> <mi>&amp;sigma;</mi> <mi>j</mi> <mo>+</mo> </msubsup> </mrow> </mfrac> <msubsup> <mo>&amp;Integral;</mo> <mrow> <mo>-</mo> <mi>&amp;infin;</mi> </mrow> <mi>x</mi> </msubsup> <mi>exp</mi> <mrow> <mo>(</mo> <mfrac> <msup> <mrow> <mo>(</mo> <mi>x</mi> <mo>-</mo> <msubsup> <mi>&amp;mu;</mi> <mi>j</mi> <mo>+</mo> </msubsup> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mrow> <mn>2</mn> <msubsup> <mi>&amp;sigma;</mi> <mi>j</mi> <mrow> <mo>+</mo> <mn>2</mn> </mrow> </msubsup> </mrow> </mfrac> <mo>)</mo> </mrow> <mi>d</mi> <mi>x</mi> <mo>+</mo> <msub> <mi>C</mi> <mn>2</mn> </msub> </mrow>

wherein,is the mean of the j-th dimension features of the positive samples,variance of j-th dimension feature of positive sample, C₁、C₂Is a linear coefficient;

when x is mu_jWhen, the probability P that the positive class center belongs to the positive sample corresponds to₊；

When x is + ∞, the probability of belonging to a positive sample is highest at this time, and is considered to be 1 at this time;

when x is ═ infinity, the probability of belonging to a positive sample is lowest at this time, and it is considered to be 0 at this time;

the function is expressed as x ═ mu_jIs segmented into

<mrow> <msup> <msub> <mi>p</mi> <mi>j</mi> </msub> <mo>+</mo> </msup> <mo>=</mo> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <mn>2</mn> <msub> <mi>P</mi> <mo>+</mo> </msub> <mo>*</mo> <mfrac> <mn>1</mn> <mrow> <msqrt> <mrow> <mn>2</mn> <mi>&amp;pi;</mi> </mrow> </msqrt> <msubsup> <mi>&amp;sigma;</mi> <mi>j</mi> <mo>+</mo> </msubsup> </mrow> </mfrac> <msubsup> <mo>&amp;Integral;</mo> <mrow> <mo>-</mo> <mi>&amp;infin;</mi> </mrow> <mi>x</mi> </msubsup> <mi>exp</mi> <mrow> <mo>(</mo> <mfrac> <msup> <mrow> <mo>(</mo> <mrow> <mi>x</mi> <mo>-</mo> <msubsup> <mi>&amp;mu;</mi> <mi>j</mi> <mo>+</mo> </msubsup> </mrow> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mrow> <mn>2</mn> <msubsup> <mi>&amp;sigma;</mi> <mi>j</mi> <mrow> <mo>+</mo> <mn>2</mn> </mrow> </msubsup> </mrow> </mfrac> <mo>)</mo> </mrow> <mi>d</mi> <mi>x</mi> </mrow> </mtd> <mtd> <mrow> <mi>x</mi> <mo>&amp;le;</mo> <msub> <mi>&amp;mu;</mi> <mi>j</mi> </msub> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <mn>2</mn> <mo>*</mo> <mrow> <mo>(</mo> <mn>1</mn> <mo>-</mo> <msub> <mi>P</mi> <mo>+</mo> </msub> <mo>)</mo> </mrow> <mfrac> <mn>1</mn> <mrow> <msqrt> <mrow> <mn>2</mn> <mi>&amp;pi;</mi> </mrow> </msqrt> <msubsup> <mi>&amp;sigma;</mi> <mi>j</mi> <mo>+</mo> </msubsup> </mrow> </mfrac> <msubsup> <mo>&amp;Integral;</mo> <mrow> <mo>-</mo> <mi>&amp;infin;</mi> </mrow> <mi>x</mi> </msubsup> <mi>exp</mi> <mrow> <mo>(</mo> <mfrac> <msup> <mrow> <mo>(</mo> <mi>x</mi> <mo>-</mo> <msubsup> <mi>&amp;mu;</mi> <mi>j</mi> <mo>+</mo> </msubsup> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mrow> <mn>2</mn> <msubsup> <mi>&amp;sigma;</mi> <mi>j</mi> <mrow> <mo>+</mo> <mn>2</mn> </mrow> </msubsup> </mrow> </mfrac> <mo>)</mo> </mrow> <mi>d</mi> <mi>x</mi> <mo>+</mo> <mn>2</mn> <msub> <mi>P</mi> <mo>+</mo> </msub> <mo>-</mo> <mn>1</mn> </mrow> </mtd> <mtd> <mrow> <mi>x</mi> <mo>&amp;GreaterEqual;</mo> <msub> <mi>&amp;mu;</mi> <mi>j</mi> </msub> </mrow> </mtd> </mtr> </mtable> </mfenced> </mrow>

(4.2)L₂Probability of each dimension feature of sample belonging to negative sample

Is provided with L₁The negative attribute sample class center is characterized byCorresponding to a probability of belonging to a negative example of P_-Wherein x is_i ^-＝(x_i1,x_i2,…x_ij) For negative samples, the sample characteristics obey a Gaussian distribution x_-～N(μ^-,σ ^_-2) (ii) a Assuming that the sample characteristics of the learning set and the probability of the learning set belonging to the negative sample are in a mapping relation, and obtaining L according to the Gaussian distribution of the negative sample₂Probability that j-th dimension feature of sample belongs to negative sampleComprises the following steps:

<mrow> <msubsup> <mi>p</mi> <mi>j</mi> <mo>-</mo> </msubsup> <mo>=</mo> <msup> <msub> <mi>C</mi> <mn>1</mn> </msub> <mo>&amp;prime;</mo> </msup> <mo>*</mo> <mrow> <mo>(</mo> <mn>1</mn> <mo>-</mo> <mfrac> <mn>1</mn> <mrow> <msqrt> <mrow> <mn>2</mn> <mi>&amp;pi;</mi> </mrow> </msqrt> <msubsup> <mi>&amp;sigma;</mi> <mi>j</mi> <mo>-</mo> </msubsup> </mrow> </mfrac> <msubsup> <mo>&amp;Integral;</mo> <mrow> <mo>-</mo> <mi>&amp;infin;</mi> </mrow> <mi>x</mi> </msubsup> <mi>exp</mi> <mo>(</mo> <mfrac> <msup> <mrow> <mo>(</mo> <mi>x</mi> <mo>-</mo> <msubsup> <mi>&amp;mu;</mi> <mi>j</mi> <mo>-</mo> </msubsup> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mrow> <mn>2</mn> <msup> <msubsup> <mi>&amp;sigma;</mi> <mi>j</mi> <mo>-</mo> </msubsup> <mn>2</mn> </msup> </mrow> </mfrac> <mo>)</mo> <mi>d</mi> <mi>x</mi> <mo>)</mo> </mrow> <mo>+</mo> <msup> <msub> <mi>C</mi> <mn>2</mn> </msub> <mo>&amp;prime;</mo> </msup> </mrow>

wherein,is the mean of the j-th dimension features of the negative samples,variance of j-th dimension characteristic of negative sample, C₁'、C₂' is a linear coefficient;

when x is mu_-When, the probability P that the negative class center belongs to the negative sample corresponds to_-；

When x is ═ infinity, the probability of belonging to a negative sample is highest at this time, and it is considered to be 1 at this time;

when x is + ∞, the probability of belonging to the negative sample is the lowest, and the probability is considered to be 0;

by applying the function toSegmented into the above formula

<mrow> <msup> <msub> <mi>p</mi> <mi>j</mi> </msub> <mo>-</mo> </msup> <mo>=</mo> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <mn>2</mn> <mo>*</mo> <mrow> <mo>(</mo> <mrow> <mn>1</mn> <mo>-</mo> <msub> <mi>P</mi> <mo>-</mo> </msub> </mrow> <mo>)</mo> </mrow> <mo>*</mo> <mrow> <mo>(</mo> <mrow> <mn>1</mn> <mo>-</mo> <mfrac> <mn>1</mn> <mrow> <msqrt> <mrow> <mn>2</mn> <mi>&amp;pi;</mi> </mrow> </msqrt> <msubsup> <mi>&amp;sigma;</mi> <mi>j</mi> <mo>-</mo> </msubsup> </mrow> </mfrac> <msubsup> <mo>&amp;Integral;</mo> <mrow> <mo>-</mo> <mi>&amp;infin;</mi> </mrow> <mi>x</mi> </msubsup> <mi>exp</mi> <mrow> <mo>(</mo> <mfrac> <msup> <mrow> <mo>(</mo> <mrow> <mi>x</mi> <mo>-</mo> <msubsup> <mi>&amp;mu;</mi> <mi>j</mi> <mo>-</mo> </msubsup> </mrow> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mrow> <mn>2</mn> <msubsup> <mi>&amp;sigma;</mi> <mi>j</mi> <mrow> <mo>-</mo> <mn>2</mn> </mrow> </msubsup> </mrow> </mfrac> <mo>)</mo> </mrow> <mi>d</mi> <mi>x</mi> </mrow> <mo>)</mo> </mrow> <mo>+</mo> <mn>2</mn> <msub> <mi>P</mi> <mo>-</mo> </msub> <mo>-</mo> <mn>1</mn> </mrow> </mtd> <mtd> <mrow> <mi>x</mi> <mo>&amp;le;</mo> <msubsup> <mi>&amp;mu;</mi> <mi>j</mi> <mo>-</mo> </msubsup> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <mn>2</mn> <msub> <mi>P</mi> <mo>-</mo> </msub> <mo>*</mo> <mrow> <mo>(</mo> <mrow> <mn>1</mn> <mo>-</mo> <mfrac> <mn>1</mn> <mrow> <msqrt> <mrow> <mn>2</mn> <mi>&amp;pi;</mi> </mrow> </msqrt> <msubsup> <mi>&amp;sigma;</mi> <mi>j</mi> <mo>-</mo> </msubsup> </mrow> </mfrac> <msubsup> <mo>&amp;Integral;</mo> <mrow> <mo>-</mo> <mi>&amp;infin;</mi> </mrow> <mi>x</mi> </msubsup> <mi>exp</mi> <mrow> <mo>(</mo> <mfrac> <msup> <mrow> <mo>(</mo> <mrow> <mi>x</mi> <mo>-</mo> <msubsup> <mi>&amp;mu;</mi> <mi>j</mi> <mo>-</mo> </msubsup> </mrow> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mrow> <mn>2</mn> <msubsup> <mi>&amp;sigma;</mi> <mi>j</mi> <mrow> <mo>-</mo> <mn>2</mn> </mrow> </msubsup> </mrow> </mfrac> <mo>)</mo> </mrow> <mi>d</mi> <mi>x</mi> </mrow> <mo>)</mo> </mrow> </mrow> </mtd> <mtd> <mrow> <mi>x</mi> <mo>&amp;GreaterEqual;</mo> <msubsup> <mi>&amp;mu;</mi> <mi>j</mi> <mo>-</mo> </msubsup> </mrow> </mtd> </mtr> </mtable> </mfenced> </mrow>

(4.3)L₂Adaptation rate p of each dimension characteristic of sample_j：

<mrow> <msub> <mi>p</mi> <mrow> <mi>j</mi> <mo>_</mo> <mi>m</mi> <mi>a</mi> <mi>t</mi> <mi>c</mi> <mi>h</mi> </mrow> </msub> <mo>=</mo> <msubsup> <mi>p</mi> <mi>j</mi> <mo>+</mo> </msubsup> <mo>-</mo> <msubsup> <mi>p</mi> <mi>j</mi> <mo>-</mo> </msubsup> </mrow>

L is obtained by this step₂Probability P that j-th dimension feature of sample belongs to positive/negative sample_j ⁺、P_j ^-And L₂Adaptation rate P of each dimension characteristic of sample_{j_match}。

6. The method according to claim 1 or 2, wherein the step (5) is performed by calculating the learning set L as follows₂Sensitivity of each dimensional feature of the sample to suitability:

changing the learning set L₂J-th dimension eigenvector x of sample matrix_jSimultaneously controlling the rest m-1 dimensional feature vectors to be unchanged when x is_j＝(k,k,k,…k)^TK varies sequentially from 0,0.1,0.2 … 1.0.0 with a sample matrix of

<mrow> <msup> <mi>X</mi> <mo>&amp;prime;</mo> </msup> <mo>=</mo> <mfenced open = "[" close = "]"> <mtable> <mtr> <mtd> <msub> <mi>x</mi> <mn>11</mn> </msub> </mtd> <mtd> <msub> <mi>x</mi> <mn>12</mn> </msub> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <mi>k</mi> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <msub> <mi>x</mi> <mrow> <mn>1</mn> <mi>m</mi> </mrow> </msub> </mtd> </mtr> <mtr> <mtd> <msub> <mi>x</mi> <mn>21</mn> </msub> </mtd> <mtd> <msub> <mi>x</mi> <mn>22</mn> </msub> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <mi>k</mi> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <msub> <mi>x</mi> <mrow> <mn>2</mn> <mi>m</mi> </mrow> </msub> </mtd> </mtr> <mtr> <mtd> <mtable> <mtr> <mtd> <mo>.</mo> </mtd> </mtr> <mtr> <mtd> <mo>.</mo> </mtd> </mtr> <mtr> <mtd> <mo>.</mo> </mtd> </mtr> </mtable> </mtd> <mtd> <mtable> <mtr> <mtd> <mo>.</mo> </mtd> </mtr> <mtr> <mtd> <mo>.</mo> </mtd> </mtr> <mtr> <mtd> <mo>.</mo> </mtd> </mtr> </mtable> </mtd> <mtd> <mtable> <mtr> <mtd> <mrow></mrow> </mtd> </mtr> <mtr> <mtd> <mrow></mrow> </mtd> </mtr> <mtr> <mtd> <mrow></mrow> </mtd> </mtr> </mtable> </mtd> <mtd> <mtable> <mtr> <mtd> <mo>.</mo> </mtd> </mtr> <mtr> <mtd> <mo>.</mo> </mtd> </mtr> <mtr> <mtd> <mo>.</mo> </mtd> </mtr> </mtable> </mtd> <mtd> <mtable> <mtr> <mtd> <mrow></mrow> </mtd> </mtr> <mtr> <mtd> <mrow></mrow> </mtd> </mtr> <mtr> <mtd> <mrow></mrow> </mtd> </mtr> </mtable> </mtd> <mtd> <mtable> <mtr> <mtd> <mo>.</mo> </mtd> </mtr> <mtr> <mtd> <mo>.</mo> </mtd> </mtr> <mtr> <mtd> <mo>.</mo> </mtd> </mtr> </mtable> </mtd> </mtr> <mtr> <mtd> <msub> <mi>x</mi> <mrow> <mi>n</mi> <mn>1</mn> </mrow> </msub> </mtd> <mtd> <msub> <mi>x</mi> <mrow> <mi>n</mi> <mn>2</mn> </mrow> </msub> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <mi>k</mi> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <msub> <mi>x</mi> <mrow> <mi>n</mi> <mi>m</mi> </mrow> </msub> </mtd> </mtr> </mtable> </mfenced> </mrow>

After preprocessing the sample matrix, classifying X' by using the previous SVM training model to obtain n samples with correct classification number_(j,k)If the total number of samples is n, the classification accuracy at this time is determinedRecording the maximum value P of the classification accuracy_jmaxAnd a minimum value P_jmin：

P_jmax＝max{P_(j,k)|k＝0,0.1,...,1.0}

P_jmin＝min{P_(j,k)|k＝0,0.1,...,1.0}

The values after the m dimensionalities are normalized are used as the sensitivity of the adaptation rate of the corresponding dimensionality characteristics;

<mrow> <msub> <mi>W</mi> <mi>j</mi> </msub> <mo>=</mo> <mfrac> <mrow> <msub> <mi>P</mi> <mrow> <mi>j</mi> <mi>m</mi> <mi>a</mi> <mi>x</mi> </mrow> </msub> <mo>-</mo> <msub> <mi>P</mi> <mrow> <mi>j</mi> <mi>m</mi> <mi>i</mi> <mi>n</mi> </mrow> </msub> </mrow> <mrow> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>m</mi> </munderover> <mrow> <mo>(</mo> <msub> <mi>P</mi> <mrow> <mi>j</mi> <mi>m</mi> <mi>a</mi> <mi>x</mi> </mrow> </msub> <mo>-</mo> <msub> <mi>P</mi> <mrow> <mi>j</mi> <mi>m</mi> <mi>i</mi> <mi>n</mi> </mrow> </msub> <mo>)</mo> </mrow> </mrow> </mfrac> </mrow>

<mrow> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>m</mi> </munderover> <msub> <mi>W</mi> <mi>j</mi> </msub> <mo>=</mo> <mn>1</mn> </mrow>

wherein, W_jAnd adapting the sensitivity of the rate for the j-th dimension of the feature vector.

7. A method according to claim 1 or 2, characterised in thatThe step (6) obtains the learning set L as follows₂Adaptation rate P of the ith sample_i：

<mrow> <msub> <mi>P</mi> <mi>i</mi> </msub> <mo>=</mo> <mi>a</mi> <munderover> <mi>&amp;Sigma;</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>m</mi> </munderover> <msub> <mi>W</mi> <msup> <mi>j</mi> <mo>*</mo> </msup> </msub> <mrow> <mo>(</mo> <msup> <msub> <mi>P</mi> <mi>j</mi> </msub> <mo>+</mo> </msup> <mo>-</mo> <msup> <msub> <mi>P</mi> <mi>j</mi> </msub> <mo>-</mo> </msup> <mo>)</mo> </mrow> <mo>+</mo> <mi>b</mi> </mrow>

a+b＝1

Where m denotes the dimension of the sample feature, W_jSensitivity, P, representing adaptation rate of j-th dimension feature vector_j ⁺Represents L₂Probability that the j-th feature of a sample belongs to a positive sample, P_j ^-Represents L₂The probability that the j-th dimension feature of a sample belongs to a negative sample,

when P is present_j ⁺→ 1 time, P_j ^-→0，P_i→1；

When P is present_j ⁺Time → 0, P_j ^-→1，P_i→0；

If a is 0.5 and b is 0.5, then

<mrow> <msub> <mi>P</mi> <mi>i</mi> </msub> <mo>=</mo> <mn>0.5</mn> <mo>*</mo> <munderover> <mi>&amp;Sigma;</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>m</mi> </munderover> <msub> <mi>W</mi> <msup> <mi>j</mi> <mo>*</mo> </msup> </msub> <mrow> <mo>(</mo> <msup> <msub> <mi>P</mi> <mi>j</mi> </msub> <mo>+</mo> </msup> <mo>-</mo> <msup> <msub> <mi>P</mi> <mi>j</mi> </msub> <mo>-</mo> </msup> <mo>)</mo> </mrow> <mo>+</mo> <mn>0.5.</mn> </mrow>

8. The method of claim 1 or 2, wherein the features extracted in step (1) are:

bright and dark target density: the proportion of light and dark target pixel points is that the light target pixel points represent points with gray values larger than 2/3 of the SAR image full-image gray value, and the dark target pixel points represent points with gray values smaller than 1/3 of the SAR image full-image gray value;

the structure has obvious strength: and (3) performing binary edge extraction on the radar image, and removing noise of the connected domain mark with less number of pixels by using a connected domain marking method to mark the ratio of the total number of pixels to the image width, height and width mean value.