CN115187830A - SAR image and signal-based fuzzy comprehensive evaluation method for artificial electromagnetic environment construction effect - Google Patents

SAR image and signal-based fuzzy comprehensive evaluation method for artificial electromagnetic environment construction effect Download PDF

Info

Publication number
CN115187830A
CN115187830A CN202210713469.4A CN202210713469A CN115187830A CN 115187830 A CN115187830 A CN 115187830A CN 202210713469 A CN202210713469 A CN 202210713469A CN 115187830 A CN115187830 A CN 115187830A
Authority
CN
China
Prior art keywords
image
fuzzy
indexes
index
signal
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
Application number
CN202210713469.4A
Other languages
Chinese (zh)
Inventor
稂时楠
李贵强
蔡轶珩
杨铭铸
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing University of Technology
Original Assignee
Beijing University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing University of Technology filed Critical Beijing University of Technology
Priority to CN202210713469.4A priority Critical patent/CN115187830A/en
Publication of CN115187830A publication Critical patent/CN115187830A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Human Resources & Organizations (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Economics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • General Engineering & Computer Science (AREA)
  • Medical Informatics (AREA)
  • Computational Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Development Economics (AREA)
  • Educational Administration (AREA)
  • Mathematical Analysis (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Databases & Information Systems (AREA)
  • Strategic Management (AREA)
  • Game Theory and Decision Science (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Marketing (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Multimedia (AREA)
  • Algebra (AREA)
  • Image Analysis (AREA)

Abstract

The invention provides a comprehensive evaluation method for SAR artificial electromagnetic environment construction effect, and belongs to the field of SAR image similarity evaluation. The method comprises two parts of index analysis and selection and fuzzy comprehensive evaluation, image sample amplification, index calculation and discretization processing are carried out on SAR image indexes, so that attribute reduction of the image indexes is carried out, key image indexes are obtained, common SAR signal evaluation indexes are added, the evaluation indexes are comprehensively used, a two-stage fuzzy comprehensive evaluation method is used for comprehensively evaluating SAR artificial electromagnetic environment construction effects, namely, the similarity between an original image and an image constructed in an artificial electromagnetic environment is evaluated, and finally the required evaluation level is obtained.

Description

SAR image and signal-based fuzzy comprehensive evaluation method for artificial electromagnetic environment construction effect
Technical Field
The invention relates to the field of SAR image similarity evaluation, in particular to a fuzzy comprehensive evaluation method suitable for an artificial electromagnetic environment construction effect of SAR images and signals.
Background
The Synthetic Aperture Radar (SAR) can be free from the influence of weather images and can finish high-resolution imaging observation on the terrain, facilities, fixed and low-speed targets in a surveying and mapping area, so that the SAR has important military application value. In order to cover up the important military facilities and battle equipment of the own party and conceal the military operations, the research of the artificial electromagnetic environment construction technology effective to the SAR, namely the technology for interfering the SAR of the enemy and preventing the information such as the facilities and equipment of the own party and the military operations from being acquired by the SAR of the enemy, is urgently needed. Currently, the technology is divided into a press type method and a deception type method. The deception method is that according to the parameters in the spy intercepted SAR signal pulse, the echo signal of a false target is calculated and obtained by a certain method and forwarded to the SAR, so that false target information exists after the SAR signal acquired by an enemy is imaged, and the important target information of the enemy is shielded. Because the false target is coherent in distance and direction, the same processing gain as the echo signal of the real target can be obtained, and the interference power required by the SAR deception artificial electromagnetic environment construction technology is lower, so that the false target becomes a research hotspot in SAR countermeasure.
In order to research the effective deception artificial electromagnetic environment construction technology of the SAR, the deception artificial electromagnetic environment construction effect of the SAR needs to be evaluated, and the development of the artificial electromagnetic environment construction technology is promoted through accurate evaluation.
The existing SAR artificial electromagnetic environment construction effect evaluation mainly has the following two problems: firstly, the SAR image is evaluated only from the characteristics of the SAR image without considering the characteristics of the SAR two-dimensional signal, the peak side lobe ratio, the integral side lobe ratio and the spatial resolution embody the information of SAR signal energy, image definition and the like to a certain extent, the characteristic indexes have important significance in SAR image similarity evaluation, and the accuracy of an evaluation result is influenced by the absence of the characteristic indexes; secondly, no analysis is performed on the selection of indexes: the existing evaluation method mainly comprises the following two modes in the selection of indexes: firstly, one or more indexes are selected subjectively, and the mode can cause that the information which can be expressed by the selected indexes is insufficient, so that the evaluation angle is not comprehensive enough, and the accuracy of the final evaluation result is influenced; and secondly, the number of the selected evaluation indexes is too large, redundant indexes are possibly included, and each index has a corresponding weight, wherein the weight of the redundant index can enable information reflected by the index to be superposed on the weight, so that the accuracy of an evaluation result is reduced.
Disclosure of Invention
In order to overcome the defects of the SAR image comprehensive evaluation method, the invention provides a fuzzy comprehensive evaluation method of an artificial electromagnetic environment construction effect based on an SAR image and a signal, which has the advantages that: firstly, on the basis of considering image characteristics, six important indexes of signal characteristics are added, so that the evaluation angle is richer, and the evaluation result is more accurate; secondly, attribute reduction is carried out on a plurality of common indexes on the indexes of the image characteristics, so that the comprehensiveness of the information expressed by the indexes can be ensured, redundant attributes can be eliminated, evaluation indexes suitable for the group of samples are screened out, the accuracy of the weight of the information expressed by the indexes is ensured, and the accuracy of an evaluation result is further improved. The evaluation method comprises an index selection module and a fuzzy comprehensive evaluation module, and the total flow is shown in figure 1. The index selection module screens indexes of the original images and the simulation images input by each group in the selection of the image indexes, eliminates redundant indexes, realizes the reduction of the indexes, improves the accuracy of the index selection on the basis of ensuring the comprehensiveness of the indexes, and additionally adds six common indexes of SAR signal characteristics. And systematically and comprehensively evaluating the similarity of the original image and signal and the image and signal constructed in the artificial electromagnetic environment by using the reduced indexes and adopting a fuzzy comprehensive evaluation method to obtain a required evaluation grade.
FIG. 2 is a flow chart of the index selection module, including image sample augmentation, image index calculation, data discretization, and attribute reduction. The input of the module is an original image formed by the SAR according to an actual target and a simulated image constructed in an artificial electromagnetic environment.
Image sample augmentation: since attribute reduction requires analysis of each index of a plurality of image samples, and the number of input image samples is insufficient, and subsequent attribute reduction cannot be performed, it is necessary to expand image samples. The traditional image sample augmentation method, such as samples augmented by horizontal or vertical turning, scaling, cutting, translation and other methods, has almost no difference from original samples in various index data of image statistics and geometric characteristics, and cannot analyze differences among sample data. Therefore, it is necessary to use a deep learning method to sample-expand the original image portion and generate sample data close to the original sample, so as to facilitate the subsequent simplification steps. The Image sample augmentation network uses a SinGAN network proposed by Tamar Rott Shaham et al in SinGAN, learning a genetic Model from a Single Natural Image. The rationale for image sample augmentation using SinGAN is: the network can learn the internal distribution of the image from single image data, generate image data close to the input image, and can realize the image sample amplification of a single SAR image. The generator of the network can extract the internal features of the input SAR image, output image data with the features, and judge the authenticity of the generated image data by a discriminator. In the process of 'game' of the generator and the discriminator, the generator is continuously improved to generate more vivid data; meanwhile, the discriminator is continuously improved so as to improve the capability of distinguishing the truth of the image data. And finally, when the discriminator cannot judge whether the output image of the generator is true or false, the output image of the generator is vivid enough, and the image can be used as an image for enlarging an image sample. The method comprises the following specific steps: inputting a simulation image into a SinGAN network, and generating image sample data which is consistent with the style type of the input image sample and is similar to but different from the image content through training. In order to achieve simple calculation and ensure accuracy, the nearest m image samples need to be manually selected from the image samples generated by augmentation.
Calculating an image index: in order to perform the subsequent attribute reduction step, data of image sample statistical characteristics and various indexes of geometric characteristics need to be calculated. The indexes of the statistical characteristics include: the average value, variance, entropy, dynamic range, equivalent vision, radiation resolution and average gradient of the image gray level, and the indexes of the geometric characteristics comprise: area, perimeter, area to perimeter ratio, minimum bounding rectangle perimeter, minimum bounding rectangle area, shape parameters, circularity, aspect ratio, elongation, sphericity, equivalent circle diameter of the image object.
Data discretization: in order to perform the subsequent attribute reduction step, each item of calculated index data needs to be assigned a discrete level by size. Setting the number n of discrete intervals, firstly sorting the same index data of m images in ascending order, calculating the difference value of adjacent data, selecting the end point of the largest n-1 difference values as the break point of the interval, dividing the interval into n discrete intervals, and distributing the discrete levels from 1 to n for each interval according to the size sequence. And respectively carrying out discretization processing on 17 indexes of the image statistical characteristic and the geometric characteristic.
Attribute reduction: a large number of evaluation indexes exist in the evaluation of the artificial electromagnetic environment construction effect of the SAR image, and when the finally adopted evaluation index is selected, redundant indexes in the SAR image evaluation are eliminated by carrying out index reduction on the indexes in the SAR image. According to the concept and the method introduced in the text of 'minimum attribute reduction algorithm based on 0-1 planning' from Zhan Wan and Yuan Hai, the reduction problem is gradually converted into a 0-1 planning problem, and the problem is solved to realize attribute reduction. The basic principle of attribute reduction is as follows: the discretized data is used as an information system in a rough set theory, the reduction of the information system is defined as a minimum attribute set with invariable indiscriminate relation ind (AT) of a non-empty finite set AT of attributes, the attributes in the reduction are taken from non-empty elements in a distinguishing matrix M, repeated elements and empty sets in the distinguishing matrix are deleted, and then the minimum region diversity MS is obtained. To find the minimum attribute reduction R, it is required to satisfy that R has a non-empty intersection with any set S in the minimal distinct set, and the set after the minimum attribute reduction R removes any attribute a in R does not have a non-empty intersection with the set in at least one minimal distinct set. The above problem can be translated into the problem of finding the optimal solution (equation 1):
Figure BDA0003707827130000031
this problem is then transformed into the corresponding 0-1 programming problem (equation 2) according to mathematical theory:
Figure BDA0003707827130000041
where x is the minimum attribute, the reduction R corresponds to a 0-1 column vector, the ith attribute a i The element x of the row in x when selected into the minimum reduction R i Is 1, otherwise is zero. c represents an m-dimensional column vector whose components are all 1. Matrix P = (P) 1 ,p 2 ,…,p l ) T Wherein p is i ＝(p i1 ,p i2 ,…,p im )，
Figure BDA0003707827130000042
i =1,2, \8230;, l, j =1,2, \8230;, m. q is an l-dimensional column vector with all components 1. And solving the 0-1 planning problem, obtaining the minimum attribute reduction according to the result, eliminating redundant attributes, and obtaining the output result of the index selection module, namely the evaluation index in the aspect of the image selected in the fuzzy comprehensive evaluation module. The attribute reduction comprises the following specific steps: firstly, solving a distinguishing matrix M; second, calculate minizone diversity MS = { S = 1 ,S 2 ,···,S l }; thirdly, calculating a matrix P by the minimum region diversity MS; fourthly, solving the 0-1 plan (formula 2) to obtain a minimum attribute reduction, finding out a corresponding attribute, and determining the finally adopted evaluation index.
And the fuzzy comprehensive evaluation module is used for taking the result output by the index selection module, namely the reduced statistical characteristic index and geometric characteristic index of the image part as indexes selected by the image part in the evaluation indexes, adding six indexes of the signal part, and carrying out similarity evaluation on the SAR image and the signals by using a two-stage fuzzy comprehensive evaluation method. The first level is to distribute the weight among three indexes of statistical characteristics, geometric characteristics and signal characteristics, the second level is to distribute the weight of each evaluation index in each characteristic, and the weight of each index can be more accurately regulated and distributed by using the second-level fuzzy comprehensive evaluation, so that the accuracy of the evaluation result is improved. The input part of the fuzzy comprehensive evaluation module is provided with an original image and a reduced image index calculation value of a simulation graph constructed in an artificial electromagnetic environment and an original image which are initially input, and each calculation value of a signal index of the two graphs.
The fuzzy comprehensive evaluation method converts qualitative evaluation into quantitative evaluation according to the membership theory of fuzzy mathematics, namely, fuzzy mathematics is used for making overall evaluation on objects or objects restricted by various factors. The method has the characteristics of clear result and strong systematicness, can better solve the problem of fuzzy and difficult-to-quantify nondeterministic problem, and is suitable for carrying out similarity evaluation on the simulated graph and the original graph constructed in the SAR artificial electromagnetic environment. The basic principle of fuzzy comprehensive evaluation is that from factors influencing the problem, an evaluation set with a plurality of grades from superior to inferior of an evaluated object and the weight of an evaluation index are determined, corresponding fuzzy evaluation is respectively carried out on each index, a membership function is determined, a fuzzy judgment matrix is formed, fuzzy operation is carried out on the fuzzy judgment matrix and the weight matrix, and a quantitative comprehensive evaluation result is obtained. According to the method of Huliping et al in the method of evaluating vehicle target SAR simulation image based on fuzzy comprehensive evaluation and the method of meeting the mark in the method of MATLAB realization of fuzzy comprehensive method, the fuzzy comprehensive evaluation module can be divided into the following steps: determining selected indexes, determining a comment set, calculating a fuzzy relation matrix, determining a weight vector, carrying out fuzzy synthesis and outputting a judgment result. The overall flow chart of the module is shown in figure 3.
Determining selected indexes: the evaluation index is divided into an image index portion and a signal index portion. The image aspect indexes are the results output by the index selection module, namely the reduced statistical characteristic indexes and the geometric characteristic indexes. Indicators of signal characteristics include: azimuth and range peak sidelobe ratios, azimuth and range integral sidelobe ratios, and azimuth and range spatial resolutions. The indexes of the common SAR signal characteristics respectively represent different characteristics, so that the reduction is not carried out.
And (3) determining a comment set: the comment set is a set of final evaluation levels and is used for expressing the similarity degree of the SAR simulation image and the original image. The comment set is defined as five levels, i.e. { good, medium, poor, very poor }.
Fuzzy relation matrix calculation: in order to perform the subsequent fuzzy synthesis, the fuzzy relation matrix of each feature needs to be calculated according to the membership function of each feature. The membership function can be determined by referring to previous experience, calculating the membership by using a decreasing half trapezoid for the statistical characteristics, and calculating the membership by using a triangle and a decreasing half trapezoid for the geometric characteristics and the signal characteristics. According to the method marked in 'fuzzy comprehensive method MATLAB realization' book, the ratio of statistical characteristic index, geometric characteristic index and signal characteristic index of the original image and the simulation image is respectively calculated and is substituted into the corresponding membership function, and the fuzzy relation matrix R of the statistical characteristic is obtained 1 Fuzzy relation matrix R of geometrical characteristics 2 Fuzzy relation matrix R of signal characteristics 3 。
Determining a weight vector: in order to perform the subsequent fuzzy synthesis, it is necessary to calculate a weight vector of each feature from the determination matrix of each feature. The weight vector represents the weight occupied by each index and is also the characteristic vector of the artificially constructed judgment matrix. According to the method of Yan Zhi Qiang et al in the text of 'automated assessment research and practice of 5G cell Performance based on AHP two-stage fuzzy comprehensive evaluation method', the relative importance of each index is determined by pairwise comparison by using 1-9 scale method, and a judgment matrix A is respectively constructed for each index of statistical characteristics, each index of geometric characteristics and each index of signal characteristics of the second layer 1 ,A 2 ,A 3 And a total judgment matrix A of the three characteristics of the statistical characteristic, the geometric characteristic and the signal characteristic of the first layer. Respectively calculating judgment matrixes A 1 ,A 2 ,A 3 Random identity ratio CR of A, performing identity test, if CR<0.1, if the test fails, the judgment matrix needs to be reconstructed and the weight vector needs to be determined again.
After passing consistency check, respectively obtaining characteristic vector W for each judgment matrix 1 ,W 2 ,W 3 And W. The weight vector of each feature is the feature vector of the corresponding judgment matrix, so the weight vector of the second layer is W 1 ,W 2 ,W 3 Corresponding to the statistical, geometric and signal characteristics, respectivelyThe first-layer weight vector is W, which is a weight vector of the total of the three features.
Fuzzy synthesis: in order to obtain the FCE result vector, it is necessary to obtain the result vector by synthesizing the weight vectors of the statistical features, the geometric features, and the signal features with the fuzzy relationship matrix corresponding to each feature. The fuzzy evaluation vector of the statistical characteristics is B 1 ＝W 1 ·R 1 The fuzzy evaluation vector of the geometric features is B 2 ＝W 2 ·R 2 Fuzzy evaluation vector of signal characteristics: b is 3 ＝W 3 ·R 3 The fuzzy relation matrix of the first layer is R = [ B = 1 ,B 2 ,B 3 ]If the first layer weight vector is W, the FCE result vector is: b = W · R = W. [ B ] 1 ,B 2 ,B 3 ](fuzzy operator. Dot product operation).
Outputting a judgment result: and for the FCE result vector after fuzzy synthesis, outputting the evaluation result grade corresponding to the comment set according to the maximum membership principle, namely selecting the grade corresponding to the maximum value in the FCE result vector as the final evaluation result grade.
Advantageous effects
The invention provides a comprehensive evaluation method for the SAR artificial electromagnetic environment construction effect, which adds six important indexes of signal characteristics on the basis of considering image characteristics, so that the evaluation angle is richer, and the evaluation result is more accurate; secondly, attribute reduction is carried out on a plurality of common indexes on the indexes of the image characteristics, so that the comprehensiveness of the information expressed by the indexes can be ensured, redundant attributes can be eliminated, evaluation indexes suitable for the group of samples are screened out, the accuracy of the weight of the information expressed by the indexes is ensured, and the accuracy of an evaluation result is further improved; and the SAR artificial electromagnetic environment construction effect can be systematically and comprehensively evaluated by combining fuzzy comprehensive evaluation to obtain a required evaluation level, so that an effective method is provided for solving the problem of SAR artificial electromagnetic environment construction effect evaluation.
Drawings
FIG. 1 is a general flow chart of the practice of the present invention.
FIG. 2 is a flow chart of index selection.
Fig. 3 is a flowchart of fuzzy comprehensive evaluation.
Fig. 4 is an original image of SAR based on an actual target.
Fig. 5 is a simulation image constructed by the artificial electromagnetic environment.
Fig. 6 (a) - (i) are 9 image samples selected from the image samples generated by augmentation.
FIG. 7 is data of statistical feature indicators for 9 image samples.
Fig. 8 is data of geometric feature indexes of 9 image samples.
Fig. 9 shows data obtained by discretizing the statistical characteristic index.
Fig. 10 shows data obtained by discretizing the geometric feature index.
FIG. 11 is a histogram of the evaluation results of examples.
Detailed Description
The present invention will be described in detail below with reference to specific embodiments shown in the drawings.
Fig. 1 is a general flowchart of a fuzzy comprehensive evaluation method for artificial electromagnetic environment construction effect based on SAR images and signals, which includes an index selection module and a fuzzy comprehensive evaluation module.
FIG. 2 is a flow chart of index selection, including image sample augmentation, image index calculation, data discretization, and attribute reduction.
Image sample augmentation: for subsequent image property reduction, data for multiple image samples is required. The input image is an original image (figure 4) formed by SAR according to an actual target and a simulation image (figure 5) constructed in an artificial electromagnetic environment, and the number of image samples does not meet the number requirement of attribute reduction, so that the image samples need to be expanded. Inputting a simulation image (figure 5) constructed in an artificial electromagnetic environment into a SinGAN network, and generating image sample data which is consistent with the style and type of the input image sample and is very similar to but different from the image content through training. In order to make the calculation simple and ensure the accuracy, the closest 9 image samples are manually selected from the image samples generated by the augmentation, as shown in fig. 6.
Calculating an image index: in order to perform the subsequent attribute reduction step, data of image sample statistical characteristics and various indexes of geometric characteristics need to be calculated. The indexes of the statistical characteristics include: the average value, variance, entropy, dynamic range, equivalent vision, radiation resolution and average gradient of the image gray level, and the indexes of the geometric characteristics comprise: area, perimeter, area to perimeter ratio, minimum bounding rectangle perimeter, minimum bounding rectangle area, shape parameters, circularity, aspect ratio, elongation, sphericity, equivalent circle diameter of the image object. Fig. 7 shows data of the statistical characteristic index and fig. 8 shows data of the geometric characteristic index for 9 image samples.
Data discretization: in order to perform the subsequent attribute reduction step, each item of calculated index data needs to be assigned a discrete level by size. Setting the number of discrete intervals to be 4, firstly sorting the same index data of 9 images in an ascending order, calculating the difference value of adjacent data, selecting the end point of the maximum 3 difference values as the break point of the interval, dividing the interval into 4 discrete intervals, and distributing discrete grades from 1 to 4 to each interval according to the size sequence. Discretization processing is respectively carried out on 17 indexes in total of the image statistical characteristics and the geometric characteristics, the data after statistical characteristic discretization is shown in fig. 9, and the data after geometric characteristic discretization is shown in fig. 10.
Attribute reduction: the discretized data can be used as an information system in a rough set theory, and according to concepts and methods introduced in Zhanlengro and Shanghai in '0-1 planning-based minimum attribute reduction algorithm', the reduction problem is gradually converted into a 0-1 planning problem, and the problem is solved to realize attribute reduction. The method comprises the following specific steps: firstly, solving a distinguishing matrix M; secondly, calculating a minimum region diversity MS; thirdly, calculating a matrix P by the minimum region diversity MS; fourthly, solving the 0-1 plan (formula 2) to obtain a minimum attribute reduction, finding out the corresponding attribute, and determining the finally adopted evaluation index. In this example, the vector of the statistical feature after attribute reduction is [0 0 0 10 10 ], and in the reduction result vector of the statistical feature optimal solution, if the 4 th and 6 th elements of the 7 elements are 1 and the other elements are 0, the indexes after statistical feature reduction are the dynamic range corresponding to the 4 th element and the radiation resolution corresponding to the 6 th element. The vector of the geometric features after attribute reduction is [0 0 0 0 10 0 0], and similarly, the remaining indexes of the geometric features after attribute reduction are the elongation and the minimum circumscribed rectangle area. Therefore, the output result of the index selection module is: the indexes of the statistical characteristics are dynamic range and radiation resolution, and the indexes of the geometric characteristics are elongation and minimum circumscribed rectangle area.
The fuzzy comprehensive evaluation module comprises the steps of determining selected indexes, determining a comment set, calculating a fuzzy relation matrix, determining a weight vector, carrying out fuzzy synthesis and outputting a judgment result.
Determining selected indexes: the evaluation index is divided into an image index portion and a signal index portion. The image aspect indexes are the results output by the index selection module, namely the reduced statistical characteristic indexes and the geometric characteristic indexes. In this example, the indices of the statistical features have a dynamic range and a radiation resolution, and the indices of the geometric features have an elongation and a minimum bounding rectangle area. The indicators of the signal characteristics include: azimuth and range peak sidelobe ratios, azimuth and range integral sidelobe ratios, and azimuth and range spatial resolutions.
And (3) determining a comment set: the comment set is a set of final evaluation levels and is used for expressing the similarity degree of the SAR simulation image and the original image. The comment set is defined as five levels, i.e. { good, medium, poor, very poor }.
Fuzzy relation matrix calculation: in order to perform the subsequent fuzzy synthesis, the fuzzy relation matrix of each feature needs to be calculated according to the membership function of each feature. The membership function can be determined by referring to previous experience, the statistical characteristics adopt a reduced half trapezoid to calculate the membership, and the geometric characteristics and the signal characteristics adopt a triangle and a reduced half trapezoid to calculate the membership. According to the method marked in 'MATLAB implementation by fuzzy synthesis method' book, the ratio of statistical characteristic index, geometric characteristic index and signal characteristic index of original image and simulated image are respectively calculated and are brought into the corresponding membership function to obtain the fuzzy relation matrix of statistical characteristics
Figure BDA0003707827130000091
Fuzzy relation matrix of geometric features
Figure BDA0003707827130000092
Fuzzy relation matrix of signal characteristics
Figure BDA0003707827130000093
Determining a weight vector: in order to perform the subsequent fuzzy synthesis, it is necessary to calculate a weight vector of each feature from the determination matrix of each feature. According to the method and the steps of Yan Zhi Qiang et al in the text of 'automated assessment research and practice of 5G cell performance based on AHP two-stage fuzzy comprehensive evaluation method', a 1-9 scale method is used for determining the relative importance of each index in a pairwise comparison mode to each index of statistical characteristics, each index of geometric characteristics and each index of signal characteristics of a second layer respectively to construct a judgment matrix
Figure BDA0003707827130000094
And a judgment matrix of the three feature totalities of the statistical feature, the geometric feature and the signal feature of the first layer
Figure BDA0003707827130000095
Judgment matrix A 1 ,A 2 ,A 3 And the random consistency ratio CR of A is less than 0.1, and the consistency test is satisfied. Then, the characteristic vector W is obtained for each judgment matrix 1 ＝[0.6 0.4],W 2 ＝[0.6667 0.3333],W 3 ＝[0.1667 0.1667 0.1667 0.1667 0.1667 0.1667]And W = [0.3333 0.3333 0.3333]. The weight vector of each feature is the feature vector of the corresponding judgment matrix, so the weight vector of the second layer is W 1 ,W 2 ,W 3 The first layer weight vector is W, which is the weight vector of the total of the three characteristics.
Fuzzy synthesis: to obtain the FCE result vector, the weight vector of the statistical features, the geometric features, and the signal features and each of them are requiredAnd carrying out synthetic operation on the fuzzy relation matrix corresponding to the characteristics to obtain a result vector. The fuzzy evaluation vector of the statistical characteristics is B 1 ＝W 1 ·R 1 ＝[0.6 0.2402 0 0 0.4]The fuzzy evaluation vector of the geometric features is B 2 ＝W 2 ·R 2 ＝[1 0 0 0 0]Fuzzy evaluation vector of signal characteristics: b is 3 ＝W 3 ·R 3 ＝[0.9453 0.0547 0 0 0]The fuzzy relation matrix of the first layer is:
Figure BDA0003707827130000101
the first layer weight vector is W = [0.3333 0.3333 0.3333]Then the FCE result vector is: b = W · R (= W. [ B) 1 ,B 2 ,B 3 ]＝[0.8484 0.0983 0 0 0.1333](fuzzy operator. Dot product operation).
Outputting a judgment result: and for the FCE result vector after fuzzy synthesis, outputting an evaluation result grade corresponding to the comment set according to a maximum membership principle, namely selecting a grade corresponding to the maximum value in the FCE result vector as a final evaluation result grade. In the FCE result vector B = [0.8484 0.0983 0.1333] in this example, the position where the maximum value 0.8484 is located is the position of the first element, and then the rank "good" corresponding to the maximum value in the FCE result vector is selected as the final evaluation result rank, and the evaluation result histogram is shown in fig. 11.
It is to be understood that while the specification has been described in terms of embodiments, it is not intended that each embodiment includes only a single technical means, but it is intended that the description be construed as illustrative only, and that all technical means within the various embodiments may be combined as suitable, all within the skill of the art, all within the scope of which is to be accorded the full scope of the specification.
The above-listed series of detailed descriptions are merely specific illustrations of possible embodiments of the present invention, and they are not intended to limit the scope of the present invention, and all equivalent embodiments or modifications that do not depart from the technical spirit of the present invention should be included within the scope of the present invention.

Claims (10)

1. A fuzzy comprehensive evaluation method of artificial electromagnetic environment construction effect based on SAR images and signals is characterized by comprising the following steps: particularly realized by an index selection module and a fuzzy comprehensive evaluation module,
the index selection module is used for screening indexes of the input original image and the simulation image to be evaluated and eliminating redundant indexes, and comprises image sample amplification, image index calculation, data discretization and attribute reduction; the image sample is expanded to generate a plurality of image samples which are consistent with the style type of the simulation graph to be evaluated and have quite similar but different image contents; the image index calculation is used for calculating data of each index of the statistical characteristic and the geometric characteristic of the image sample; the data discretization is used for carrying out the subsequent attribute reduction step, and each calculated index data is endowed with a discrete grade according to the size; the attribute reduction uses the data after the discretization processing as an information system in a rough set theory, and utilizes a minimum attribute reduction algorithm to obtain various indexes of the screened statistical characteristics and geometric characteristics to complete index selection;
the fuzzy comprehensive evaluation module respectively constructs a judgment matrix of the statistical characteristics, the geometric characteristics and the signal characteristics and a total judgment matrix of the statistical characteristics, the geometric characteristics and the signal characteristics according to the selected evaluation indexes, obtains a weight vector of each characteristic according to the judgment matrix, then determines a comment set, calculates a fuzzy relation matrix, carries out fuzzy synthesis to obtain a result vector, and obtains a final evaluation grade according to a maximum membership principle; the fuzzy comprehensive evaluation module comprises the steps of determining selected indexes, determining a comment set, calculating a fuzzy relation matrix, determining a weight vector, carrying out fuzzy synthesis and outputting a judgment result.
2. The fuzzy comprehensive evaluation method for the artificial electromagnetic environment construction effect based on the SAR image and the signal as claimed in claim 1, characterized in that: sample augmentation was performed using a SinGAN network.
3. The fuzzy comprehensive evaluation method for the artificial electromagnetic environment construction effect based on the SAR image and the signal as claimed in claim 1, characterized in that: the indexes of the statistical characteristics include: mean value, variance, entropy, dynamic range, equivalent vision, radiation resolution and average gradient of image gray level; the indicators of the geometrical characteristics include: area, perimeter, area to perimeter ratio, minimum bounding rectangle perimeter, minimum bounding rectangle area, shape parameters, circularity, aspect ratio, elongation, sphericity, equivalent circle diameter of the image object.
4. The method for fuzzy comprehensive evaluation of artificial electromagnetic environment construction effect based on SAR image and signal as claimed in claim 1 or 3, characterized in that: determining selected indexes to be used for determining evaluation indexes, wherein the evaluation indexes are divided into an image index part and a signal index part, and the image indexes are results output by an index selection module, namely reduced statistical characteristic indexes and geometric characteristic indexes; indicators of signal characteristics include: azimuth and range peak sidelobe ratios, azimuth and range integral sidelobe ratios, and azimuth and range spatial resolutions.
5. The fuzzy comprehensive evaluation method for the artificial electromagnetic environment construction effect based on the SAR image and the signal as claimed in claim 1, characterized in that: and the comment set is a final evaluation grade set and is used for expressing the similarity degree of the simulated image to be evaluated and the original image.
6. The fuzzy comprehensive evaluation method for the artificial electromagnetic environment construction effect based on the SAR image and the signal as claimed in claim 1, characterized in that: the calculation process of the fuzzy relation matrix is as follows: respectively calculating the ratio of the original image to the simulated image to be evaluated to the statistical characteristic index, the geometric characteristic index and the signal characteristic index, and introducing the ratios into the corresponding membership function to respectively obtain a fuzzy relation matrix R of the statistical characteristics 1 Fuzzy relation matrix R of geometrical characteristics 2 Fuzzy relation matrix R of signal characteristics 3 。
7. The fuzzy comprehensive evaluation method of the artificial electromagnetic environment construction effect based on the SAR image and the signal as claimed in claim 6, characterized in that: and calculating the membership degree by adopting a decreasing half trapezoid for the statistical characteristics, and calculating the membership degree by adopting a triangle and a decreasing half trapezoid for the geometric characteristics and the signal characteristics.
8. The fuzzy comprehensive evaluation method for the artificial electromagnetic environment construction effect based on the SAR image and the signal as claimed in claim 1, characterized in that: determining weight vector means calculating weight vector of each feature according to the judgment matrix of each feature, the fuzzy comprehensive evaluation module adopts a two-stage fuzzy comprehensive evaluation method, the first stage is to distribute weight among three indexes of statistical feature, geometric feature and signal feature, the second stage is to distribute weight to each evaluation index in each feature, specifically, 1-9 scale method is used to determine relative importance of each index in a pairwise comparison mode, and judgment matrix A is respectively constructed for each index of statistical feature, each index of geometric feature and each index of signal feature of the second layer 1 ,A 2 ,A 3 Constructing a total judgment matrix A for the three characteristics of the statistical characteristic, the geometric characteristic and the signal characteristic of the first layer, and respectively calculating the judgment matrix A 1 ,A 2 ,A 3 Random identity ratio CR of A, performing identity test, if CR<0.1, if the consistency check is failed, reconstructing a judgment matrix and re-determining a weight vector until the consistency check is passed;
after passing consistency check, respectively obtaining characteristic vector W for each judgment matrix 1 ,W 2 ,W 3 And W, the weight vector of each feature is the feature vector of the corresponding judgment matrix, so the weight vector of the second layer is W 1 ,W 2 ,W 3 The first layer weight vector is W, which is the weight vector of the total of the three characteristics.
9. The fuzzy comprehensive evaluation method for the artificial electromagnetic environment construction effect based on the SAR image and the signal as claimed in claim 8, characterized in that:
fuzzy synthesis includes separate countingCalculating the fuzzy evaluation vector of the statistical characteristics as B 1 ＝W 1 ·R 1 The fuzzy evaluation vector of the geometric characteristics is B 2 ＝W 2 ·R 2 Fuzzy evaluation vector of signal characteristics: b is 3 ＝W 3 ·R 3 The fuzzy relation matrix of the first layer is R = [ B = [) 1 ,B 2 ,B 3 ]If the first layer weight vector is W, the FCE result vector is: b = W · R = W · [ B · 1 ,B 2 ,B 3 ]。
10. The fuzzy comprehensive evaluation method for the artificial electromagnetic environment construction effect based on the SAR image and the signal according to claim 9, characterized in that:
outputting a judgment result: and for the FCE result vector after fuzzy synthesis, outputting the evaluation result grade corresponding to the comment set according to the maximum membership principle, namely selecting the grade corresponding to the maximum value in the FCE result vector as the final evaluation result grade.
CN202210713469.4A 2022-06-22 2022-06-22 SAR image and signal-based fuzzy comprehensive evaluation method for artificial electromagnetic environment construction effect Pending CN115187830A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210713469.4A CN115187830A (en) 2022-06-22 2022-06-22 SAR image and signal-based fuzzy comprehensive evaluation method for artificial electromagnetic environment construction effect

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210713469.4A CN115187830A (en) 2022-06-22 2022-06-22 SAR image and signal-based fuzzy comprehensive evaluation method for artificial electromagnetic environment construction effect

Publications (1)

Publication Number Publication Date
CN115187830A true CN115187830A (en) 2022-10-14

Family

ID=83515897

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210713469.4A Pending CN115187830A (en) 2022-06-22 2022-06-22 SAR image and signal-based fuzzy comprehensive evaluation method for artificial electromagnetic environment construction effect

Country Status (1)

Country Link
CN (1) CN115187830A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116883414A (en) * 2023-09-08 2023-10-13 国网上海市电力公司 Multi-system data selection method and system suitable for operation and maintenance of power transmission line

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116883414A (en) * 2023-09-08 2023-10-13 国网上海市电力公司 Multi-system data selection method and system suitable for operation and maintenance of power transmission line
CN116883414B (en) * 2023-09-08 2024-01-26 国网上海市电力公司 Multi-system data selection method and system suitable for operation and maintenance of power transmission line

Similar Documents

Publication Publication Date Title
CN110135267B (en) Large-scene SAR image fine target detection method
CN109086700B (en) Radar one-dimensional range profile target identification method based on deep convolutional neural network
Han et al. Convolutional neural network for convective storm nowcasting using 3-D Doppler weather radar data
CN111797717B (en) High-speed high-precision SAR image ship detection method
CN108038445B (en) SAR automatic target identification method based on multi-view deep learning framework
Winiwarter et al. Classification of ALS point clouds using end-to-end deep learning
CN112183432B (en) Building area extraction method and system based on medium-resolution SAR image
CN109284786B (en) SAR image terrain classification method for generating countermeasure network based on distribution and structure matching
CN108983228B (en) RCS near-far field transformation method based on deep neural network
CN113240047B (en) SAR target recognition method based on component analysis multi-scale convolutional neural network
CN105005789B (en) A kind of remote sensing images terrain classification method of view-based access control model vocabulary
CN109086679A (en) A kind of millimetre-wave radar safety check instrument foreign matter detecting method
CN111311702B (en) Image generation and identification module and method based on BlockGAN
CN109977968B (en) SAR change detection method based on deep learning classification comparison
CN113361367B (en) Underground target electromagnetic inversion method and system based on deep learning
CN109446894A (en) The multispectral image change detecting method clustered based on probabilistic segmentation and Gaussian Mixture
CN111160111A (en) Human body key point detection method based on deep learning
CN107330457A (en) A kind of Classification of Polarimetric SAR Image method based on multi-feature fusion
Kamrava et al. Quantifying accuracy of stochastic methods of reconstructing complex materials by deep learning
CN110276746A (en) A kind of robustness method for detecting change of remote sensing image
CN112818849A (en) Crowd density detection algorithm based on context attention convolutional neural network of counterstudy
CN115187830A (en) SAR image and signal-based fuzzy comprehensive evaluation method for artificial electromagnetic environment construction effect
CN114120067A (en) Object identification method, device, equipment and medium
CN115661627A (en) Single-beam underwater target identification method based on GAF-D3Net
CN107766792A (en) A kind of remote sensing images ship seakeeping method

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