CN113960553B - Gaussian weight distribution feature extraction method in one-dimensional image recognition - Google Patents

Gaussian weight distribution feature extraction method in one-dimensional image recognition Download PDF

Info

Publication number
CN113960553B
CN113960553B CN202111197166.3A CN202111197166A CN113960553B CN 113960553 B CN113960553 B CN 113960553B CN 202111197166 A CN202111197166 A CN 202111197166A CN 113960553 B CN113960553 B CN 113960553B
Authority
CN
China
Prior art keywords
matrix
gaussian
dimensional
gaussian weight
weight distribution
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.)
Active
Application number
CN202111197166.3A
Other languages
Chinese (zh)
Other versions
CN113960553A (en
Inventor
周代英
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.)
University of Electronic Science and Technology of China
Original Assignee
University of Electronic Science and Technology of China
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 University of Electronic Science and Technology of China filed Critical University of Electronic Science and Technology of China
Priority to CN202111197166.3A priority Critical patent/CN113960553B/en
Publication of CN113960553A publication Critical patent/CN113960553A/en
Application granted granted Critical
Publication of CN113960553B publication Critical patent/CN113960553B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
    • G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
    • G01S7/411—Identification of targets based on measurements of radar reflectivity
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/02—Systems using reflection of radio waves, e.g. primary radar systems; Analogous systems
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
    • G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
    • G01S7/418—Theoretical aspects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches

Abstract

The invention belongs to the technical field of target identification, and particularly relates to a Gaussian weight distribution feature extraction method in one-dimensional image identification. The invention utilizes the sample distribution structure of the training data set to establish a transformation subspace matrix to extract the Gaussian weight distribution characteristics, and because the Gaussian weight is adopted to describe the connection tightness between samples, the aggregation condition of various samples in a high-dimensional space can be effectively represented, thereby improving the identification performance of the target, and the effectiveness of the method is verified by the simulation experiment of 4 types of targets.

Description

Gaussian weight distribution feature extraction method in one-dimensional image recognition
Technical Field
The invention belongs to the technical field of target identification, and particularly relates to a Gaussian weight distribution feature extraction method in one-dimensional image identification.
Background
The one-dimensional range profile obtained by the broadband radar contains information beneficial to classification, and compared with the traditional narrow-band radar, the one-dimensional range profile has higher classification performance, and compared with the two-dimensional profile of the radar, the one-dimensional range profile is easy to obtain and can be recognized in real time, so that the one-dimensional range profile recognition has very important significance.
The conventional subspace target identification method is a classical pattern identification method, the identification of a target is realized by establishing a subspace under the condition of a given certain criterion to extract target features, and because the connection weight between samples is not considered in the criterion of establishing the subspace, the compact structure of sample distribution cannot be represented to the greatest extent, so that the extracted features are not optimal in classification performance, and therefore, the performance of the conventional subspace target identification method has room for further improvement.
Disclosure of Invention
Aiming at the problems, the invention provides a Gaussian weight distribution feature extraction method, which can effectively represent the distribution compactness of a sample in a high-dimensional space by adopting a Gaussian weight distribution structure to establish a transformation subspace and extracting target classification features, thereby improving the recognition rate of targets.
In order to achieve the purpose, the technical scheme of the invention is as follows:
a Gaussian weight distribution feature extraction method in one-dimensional image recognition comprises the following steps:
s1, setting n-dimensional column vector x ij For the jth training one-dimensional distance image of the ith known target, i is more than or equal to 1 and less than or equal to g, j is more than or equal to 1 and less than or equal to N i ，
Figure BDA0003303496260000011
Wherein g represents the number of classes, N i The number of training samples of the ith type of known target is N, and the total number of the training samples is N;
s2, calculating Euclidean distance between samples:
p ij,rl ＝||x ij -x rl || 2
i＝1,2,…g,j＝1,2,…N i
r＝1,2,…g,l＝1,2,…N r
wherein p is ij,rl Representing the Euclidean distance between the jth one-dimensional range profile sample of the ith type of known target and the ith one-dimensional range profile sample of the r type of known target;
s3, calculating a Gaussian weight w by the Euclidean distance ij,rl ：
Figure BDA0003303496260000021
Wherein σ 2 Is a Gaussian parameter determined by an experimental value and is determined by a Gaussian weight w ij,rl Forming a Gaussian connection weight matrix W between samples:
W＝[w ij,rl ] N×N
calculating the elements and values of each row of the matrix W, and constructing a diagonal matrix D:
D＝diag{d ij }
Figure BDA0003303496260000022
wherein diag {. Denotes a diagonal matrix;
s4, constructing a characteristic equation by the matrixes W and D:
WV＝DVΣ
v is a matrix composed of characteristic vectors, sigma is a diagonal matrix composed of characteristic values, and two sides of a characteristic equation are multiplied by the matrix simultaneously
Figure BDA0003303496260000023
Obtaining:
Figure BDA0003303496260000024
order to
Figure BDA0003303496260000025
To obtain
Figure BDA0003303496260000026
Can know that V 1 The column vectors in the matrix are matrices
Figure BDA0003303496260000031
The feature vector of (2); the feature vector corresponding to the previous M largest feature values->
Figure BDA0003303496260000032
The composition transformation subspace A:
Figure BDA0003303496260000033
using matrix A to one-dimensional range profile x ij And (3) carrying out transformation:
z ij ＝A T x ij
the resulting vector z ij Is a gaussian weight distribution characteristic.
The invention has the beneficial effects that: according to the method, the Gaussian weight is used for describing the connection tightness between the samples, the aggregation condition of various samples in a high-dimensional space can be effectively represented, the target identification performance is improved, and the effectiveness of the method is verified through simulation experiments on 4 types of targets.
Detailed Description
The practical applicability of the present invention will be described below with reference to simulation experiments.
Four point targets were designed: "|" style of calligraphy, "V" style of calligraphy, "dry" style of calligraphy and "little" style of calligraphy target. The bandwidth of radar emission pulse is 150MHZ (the distance resolution is 1m, the radar radial sampling interval is 0.5 m), the target is set to be a uniform scattering point target, the scattering point of the 'I' target is 5, and the scattering points of the other three targets are 9. In the target attitude angle of 0 DEG toIn the one-dimensional distance images of every 1 degree within the range of 70 degrees, the one-dimensional distance images with target attitude angles of 0 degree, 2 degrees, 4 degrees, 6 degrees, once and 70 degrees are taken for training, the one-dimensional distance images of the other attitude angles are taken as test data, and then 35 test samples are marked for each type of targets. In the experiment, the Gaussian parameter σ 2 ＝3.5。
For four targets (i-shaped target, V-shaped target, dry-shaped target and small-shaped target), in the range of the attitude angle of 0-70 degrees, the Gaussian weight distribution feature extraction method is used for extracting the target features, and a minimum distance classifier is used for classifying, so that the average correct recognition rate of 86% is achieved, and the effectiveness of the text method is verified.

Claims (1)

1. A Gaussian weight distribution feature extraction method in one-dimensional image recognition is characterized by comprising the following steps:
s1, setting n-dimensional column vector x ij For the jth training one-dimensional distance image of the ith known target, i is more than or equal to 1 and less than or equal to g, j is more than or equal to 1 and less than or equal to N i ，
Figure FDA0004073471490000011
Wherein g represents the number of classes, N i The number of training samples of the ith type of known target is N, and the total number of the training samples is N;
s2, calculating Euclidean distance between samples:
p ij,rz ＝||x ij -x rz || 2
i＝1,2,…g,j＝1,2,…N i
r＝1,2,…g,z＝1,2,…N r
wherein p is ij,rz Representing the Euclidean distance between the jth one-dimensional range profile sample of the ith type of known target and the zth one-dimensional range profile sample of the ith type of known target;
s3, calculating a Gaussian weight w by the Euclidean distance ij,rz ：
Figure FDA0004073471490000012
Wherein σ 2 Is a Gaussian parameter determined by an experimental value and is determined by a Gaussian weight w ij,rz Forming a Gaussian connection weight matrix W between samples:
W＝[w ij,rz ] N×N
calculating the elements and values of each row of the matrix W, and constructing a diagonal matrix D:
D＝diag{d ij }
Figure FDA0004073471490000013
wherein diag {. Denotes a diagonal matrix;
s4, constructing a characteristic equation by the matrixes W and D:
WV＝DVΣ
v is a matrix composed of characteristic vectors, sigma is a diagonal matrix composed of characteristic values, and two sides of a characteristic equation are multiplied by the matrix simultaneously
Figure FDA0004073471490000021
Obtaining:
Figure FDA0004073471490000022
order to
Figure FDA0004073471490000023
To obtain
Figure FDA0004073471490000024
Can know that V 1 The column vectors in the matrix are matrices
Figure FDA0004073471490000025
Is characterized byA vector; the feature vector corresponding to the previous M largest feature values->
Figure FDA0004073471490000026
The composition transformation subspace A:
Figure FDA0004073471490000027
using matrix A for one-dimensional range profile x ij And (3) carrying out transformation:
z ij ＝A T x ij
the resulting vector z ij Is a gaussian weight distribution characteristic.
CN202111197166.3A 2021-10-14 2021-10-14 Gaussian weight distribution feature extraction method in one-dimensional image recognition Active CN113960553B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111197166.3A CN113960553B (en) 2021-10-14 2021-10-14 Gaussian weight distribution feature extraction method in one-dimensional image recognition

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111197166.3A CN113960553B (en) 2021-10-14 2021-10-14 Gaussian weight distribution feature extraction method in one-dimensional image recognition

Publications (2)

Publication Number Publication Date
CN113960553A CN113960553A (en) 2022-01-21
CN113960553B true CN113960553B (en) 2023-04-18

Family

ID=79464261

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111197166.3A Active CN113960553B (en) 2021-10-14 2021-10-14 Gaussian weight distribution feature extraction method in one-dimensional image recognition

Country Status (1)

Country Link
CN (1) CN113960553B (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107784320A (en) * 2017-09-27 2018-03-09 电子科技大学 Radar range profile's target identification method based on convolution SVMs
CN112149061A (en) * 2020-09-25 2020-12-29 电子科技大学 Multi-class average maximization true and false target feature extraction method

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6571225B1 (en) * 2000-02-11 2003-05-27 International Business Machines Corporation Text categorizers based on regularizing adaptations of the problem of computing linear separators
EP1729145B1 (en) * 2005-06-02 2013-08-07 GMV Aerospace and Defence S.A. Method and system for providing GNSS navigation position solution with guaranteed integrity in non-controlled environments
EP2267480B1 (en) * 2009-06-08 2013-03-13 Honeywell International Inc. Systems and methods for Gaussian decomposition of weather radar data for communication
CN106919980B (en) * 2017-01-24 2020-02-07 南京大学 Incremental target identification system based on ganglion differentiation
CN108761411B (en) * 2018-04-10 2021-11-23 电子科技大学 True and false target one-dimensional range profile feature extraction method
CN110007286B (en) * 2019-04-22 2022-05-24 电子科技大学 Linear discriminant learning true and false target one-dimensional range profile feature extraction method
CN110658507B (en) * 2019-10-12 2022-07-29 电子科技大学 Multi-class average maximization true and false target feature extraction method for radar target identification

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107784320A (en) * 2017-09-27 2018-03-09 电子科技大学 Radar range profile's target identification method based on convolution SVMs
CN112149061A (en) * 2020-09-25 2020-12-29 电子科技大学 Multi-class average maximization true and false target feature extraction method

Also Published As

Publication number Publication date
CN113960553A (en) 2022-01-21

Similar Documents

Publication Publication Date Title
CN109948658B (en) Feature diagram attention mechanism-oriented anti-attack defense method and application
CN107238822B (en) Method for extracting orthogonal nonlinear subspace characteristics of true and false target one-dimensional range profile
CN110007286B (en) Linear discriminant learning true and false target one-dimensional range profile feature extraction method
CN108845302B (en) K-nearest neighbor transformation true and false target feature extraction method
CN115329594B (en) Large-scale missile cluster attack and defense confrontation simulation acceleration method and system
CN111539488B (en) Imaging distortion resisting target classification method for ultra-narrow pulse radar under complex dynamic track
Leung et al. Recognition of handwritten Chinese characters by combining regularization, Fisher's discriminant and distorted sample generation
CN113988119B (en) Block probability distribution feature extraction method in multi-rotor unmanned aerial vehicle recognition
CN110687514B (en) Nonlinear discrimination learning true and false target one-dimensional range profile feature extraction method
CN110221267B (en) Target library attribute discrimination neighborhood preserving learning subspace feature extraction method
CN108845303A (en) A kind of non-linear steady true and false target's feature-extraction method in subspace
CN112163615B (en) Unknown target discrimination method based on threshold optimization dictionary learning
CN108594202B (en) Neighborhood distribution structure nonlinear projection target feature extraction method
CN113960553B (en) Gaussian weight distribution feature extraction method in one-dimensional image recognition
CN113269203A (en) Subspace feature extraction method for multi-rotor unmanned aerial vehicle recognition
CN110068799B (en) Sparse neighborhood center preserving RCS sequence feature extraction method
CN111860356A (en) Polarization SAR image classification method based on nonlinear projection dictionary pair learning
CN110135444B (en) Method for extracting attribute one-dimensional image discrimination optimal eigen-subspace feature of target library
CN107678007B (en) Method for extracting radar true and false target one-dimensional range profile features in exponential domain compact subspace
CN114280566B (en) One-dimensional range profile identification method based on class label association
CN113191447B (en) Method for extracting characteristics of sample distribution structure chart in unknown target discrimination
CN107403136A (en) The SAR target model recognition methods of dictionary learning is kept based on structure
CN112257792A (en) SVM (support vector machine) -based real-time video target dynamic classification method
CN113156416B (en) Unknown target discrimination method based on multi-kernel dictionary learning
CN112668652A (en) Method and system for identifying cluster array and motion trend in unmanned equipment confrontation

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
GR01 Patent grant
GR01 Patent grant