CN113447901B - Sea clutter identification and target detection method independent of probability model - Google Patents
Sea clutter identification and target detection method independent of probability model Download PDFInfo
- Publication number
- CN113447901B CN113447901B CN202110533736.5A CN202110533736A CN113447901B CN 113447901 B CN113447901 B CN 113447901B CN 202110533736 A CN202110533736 A CN 202110533736A CN 113447901 B CN113447901 B CN 113447901B
- Authority
- CN
- China
- Prior art keywords
- target
- matrix
- targets
- echo
- radar
- 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
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 41
- 239000011159 matrix material Substances 0.000 claims abstract description 101
- 238000000034 method Methods 0.000 claims abstract description 29
- 239000013598 vector Substances 0.000 claims abstract description 11
- 238000012544 monitoring process Methods 0.000 claims abstract description 9
- 238000003064 k means clustering Methods 0.000 claims abstract description 6
- 238000002592 echocardiography Methods 0.000 claims description 7
- 230000003595 spectral effect Effects 0.000 claims description 6
- 238000001228 spectrum Methods 0.000 claims description 3
- 238000010276 construction Methods 0.000 abstract 1
- 238000013179 statistical model Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000003672 processing method Methods 0.000 description 2
- 230000001629 suppression Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 239000000779 smoke Substances 0.000 description 1
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
- G01S7/414—Discriminating targets with respect to background clutter
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Abstract
A sea clutter recognition and target detection method independent of probability model is characterized in that echo of the whole monitoring sea surface is scanned by a radar as an object, echo pulse groups are formed into echo matrixes corresponding to all azimuth directions, and the echo matrixes are converted into RD echo matrixes S x The method comprises the steps of carrying out a first treatment on the surface of the S for detecting all azimuth directions of radar x Peak construction target feature matrix F 1 Obtaining a target matrix by using subtractive clustering and C-means clustering; then, the target matrix is processed by using subtractive clustering and K-means clustering, a corresponding category submatrix is obtained, the submatrix with the largest line number is identified as a sea clutter characteristic matrix, and the other submatrices are target characteristic submatrices to be identified, and a target characteristic matrix F thereof is constructed 2 The method comprises the steps of carrying out a first treatment on the surface of the Extracting a submatrix T corresponding to each azimuth in the target matrix gj Let t=f 2 ∩T gj And detecting targets in all directions to obtain target speed and position vectors, eliminating false targets by utilizing track association, and verifying by actually measured sea clutter data sets to prove the advancement and effectiveness of the invention.
Description
Technical Field
The invention belongs to the technical field of target detection, relates to target detection under a sea clutter background, and discloses a sea clutter identification and target detection method independent of a probability model.
Background
The problem of multi-target identification and tracking on the sea is one of the difficulties in the radar signal processing field, the detection performance of small targets on the sea surface is limited due to the complexity and the variability of the marine environment, and the detected targets contain a large number of false targets, so that the difficulty of multi-target tracking processing is greatly increased. Along with miniaturization and stealth of sea surface targets, the slow-speed and floating small sea surface targets become important targets for radar warning. Detection of such small targets has long been a problem in target detection in the context of sea clutter. In general, radar cross sectional area (RCS) of a floating small target is weak and the motion speed is slow, and there is often difficulty in "super clutter detection" in both the time and frequency domains. The traditional target detection method has obvious performance bottleneck for detecting the floating small target. The method is based on a statistical model of a target environment background, the statistical model used for sea clutter is LogNormal, weibull, rayleigh, pareto, K-distribution and the like conventionally, and the probability models have higher accuracy when the correlation detection time of the radar is shorter, but the sea clutter presents obvious non-stationarity along with the increase of the correlation detection time, so that the important assumption precondition of the traditional method is broken, and the detection of weak, small and slow targets is invalid.
In summary, for a sea observation radar as a detection on a weak target on the sea, the echo energy of the target is usually enhanced by increasing the observation time of the sea target area, however, the intensity of sea echo (sea clutter) received in the large coherence processing time is also correspondingly enhanced (see (a) and (b) in fig. 1, which are 3D graphs of a distance×doppler (RD) matrix of the sea observation radar deployed in a shandong smoke-stand horse island and respectively received in an antenna scanning mode and a resident mode, fig. 2 is an azimuth actual scanning angle range when the radar works in the scanning mode), and the weak target detection methods under the traditional sea clutter background all take the sea clutter as being subject to a probability density distribution, so that the difference between probability density distribution characteristics of the target is detected.
The detection of weak and small targets from the background of strong sea clutter is always a hotspot problem in the fields of radar engineering and signal processing, and a great deal of published research results are already available. One of typical solutions is to dynamically track the statistical characteristics of radar echoes, segment the radar echoes according to the principle of similar characteristics, and solve part of problems on an engineering level, but with the improvement of sea conditions, the segmentation of radar echo data is shorter and shorter, and finally, the radar echo data has no essential difference from the short-time observation situation, so that the detection of weak and small targets still has higher false alarm probability and lower discovery probability, therefore, the signal processing method based on the statistical characteristics does not meet the engineering requirements under the high sea conditions, and a sea clutter suppression method independent of a statistical model is required to be searched, so that the problem is solved.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention aims to provide a sea clutter identification and target Detection method which does not depend on a probability model, and by means of a Track-Before-Detection (TBD) method, firstly, the sea monitoring radar (Marine Surveillance Radar) is subjected to echo data centralized processing of scanning a monitoring area for one circle, sea clutter characteristics are identified by a fuzzy clustering method, then, target Detection is carried out on echo data of all directions, sea clutter is removed, target information is extracted, and targets with specific motion rules can be further extracted by means of a Track association algorithm.
In order to achieve the above purpose, the technical scheme adopted by the invention is as follows:
a sea clutter identification and target detection method independent of a probability model comprises the following steps:
s1) taking echo of radar scanning whole monitoring sea surface as an object, and observing the rotation speed omega of a radar antenna according to sea E Beam azimuth width θ of radar A And pulse repetition frequency PRF, and grouping the echo pulses into echo matrixes S corresponding to all azimuth directions r ；
S2) pairs of echo matrices S in azimuth r Performing windowed Fourier transform to convert it into RD echo matrix S x ；
S3) RD echo matrix S for detecting all azimuth orientations of radar x Is used for constructing a target feature matrix F 1 ；
S4) aiming at a target characteristic matrix F 1 Performing subtractive clustering to estimate the number k of target types 1 Will k 1 And F 1 Obtaining centroid matrix C of target cluster as input of fuzzy C-means cluster 2 Sequentially superposing to form a target matrix Targets;
s5) performing subtractive clustering on the target matrix Targets to estimate the number k of target types 2 Will k 2 And Targets is taken as input of K-means cluster to obtain K 2 Centroid matrix C of individual targets t Submatrices { A } of corresponding classes j }，j＝1,2,…,k 2 Identifying the submatrix with the largest line number as a sea clutter feature matrix, and identifying other submatrices as target feature submatrices to be identified;
s6) extracting the submatrix { A } j Target feature submatrix to be identified in the matrix and find out the corresponding centroid matrix C t Is used for constructing a target feature matrix F 2 The method comprises the steps of carrying out a first treatment on the surface of the For each radar beam azimuth direction, according to the effective target number N t Extracting corresponding target submatrices T in target matrixes Targets gj Let t=f 2 ∩T gj Obtaining a target speed and position vector P v ；
S7) detecting targets in all directions, and eliminating false targets by utilizing track association.
In the S1), the number n of radar azimuth pulses is calculated firstly p ：
Then according to n p Dividing echo pulse N of radar scanning whole monitoring sea surface into A zbin Echo of each azimuth unit forms an echo matrix S corresponding to each azimuth r ，S r Is n p ×n t Matrix of A zbin ＝floor(N/n p ) Floor (·) is a rounding function, n t The number of points is sampled for the echo pulses.
In the step S3), a Matlab function for detecting the peak value of the spectrum sequence is constructed as follows:
[pks,locs,w,p]＝findpeaks(Sx(:,i),′MinPeakProminence′,minP)
wherein the input Sx of the function (i) is RD echo matrix S x I=1, 2, …, n t Inputting an option 'MinPeakProminance', wherein minP is the minimum peak significance; the outputs pks, locs, w, p of the function are the height, position, width and significance of the spectral peaks, respectively, that are confirmed to be valid;
by S x The Doppler frequency f is found from the location locs of the height pks of the spectral peak for which each column is validated d Extracting peak value of pks, determining distance unit r according to column number, and constructing S x Target feature matrix F of (2) 1 ＝[pks,w,p,f d ,r]。
In S4), a centroid matrix C of subtractive clustering is calculated by the following formula 1 ：
C 1 ＝subclast(F 1 ,radii),0<radii<1
Wherein radii is the cluster radius, let k 1 ＝length(C 1 1), i.e. k 1 Is C 1 The number of rows of (3); centroid matrix C of target cluster 2 ＝fcm(F 1 ,k 1 ) C is carried out by 2 Target features as primary identification are stored in matrix T g ，T g Number N of lines of (2) t For the effective target number detected by the radar in the corresponding azimuth, repeatedly processing the RD echo matrix S pointed by each azimuth of the radar beam x Matrix S of RD echoes x Matrix T of (2) g Longitudinally arranged as target matrix targets= [ T gj ],j＝1,2,…,A zbin ，T gj The radar beam is directed to the sub-matrix in j corresponding Targets.
In S5), a centroid matrix C is calculated by the following formula t ：
[idx，C t ]＝kmeans(Targets,k 2 )
Wherein idx contains the target type in Targets, and idx is used to extract submatrices { A } from target matrix Targets j And identifying the submatrix with the largest number of rows as sea clutter, and the rest being targets to be identified.
In the step S6), if T is an empty set, the corresponding beam is directed to a non-target; conversely, read column 4 [ f ] of each row T dj ]And into a movement velocity v relative to the radar rj ]Extracting distance unit value [ r ] corresponding to 5 th column of each row of T kj ]Output target speed and position vector P v ＝[v rj ,r kj ]J=1, 2, … k, k is the number of lines of T, i.e. the target number.
In the S7), use is made ofTrack association method, speed and position vector P of adjacent azimuth units v And carrying out preliminary track association, if the target track almost penetrates all radar beam azimuth pointing angles, judging as a false target caused by sea clutter, taking a short track or an isolated point as a current target detection result, and verifying the accuracy of the next radar scanning data to be the target of detecting a ship or a low-altitude flight from a plurality of radar scanning periods.
Compared with the prior art, the method provided by the invention realizes the suppression of the sea clutter without depending on a sea clutter statistical model, can effectively eliminate the sea clutter and detect targets which cannot be found by the traditional method under the given sea clutter background, and has obvious improvement on detection precision.
Drawings
Fig. 1 is a 3D diagram of a distance×doppler (RD) matrix of sea clutter received by a sea observation radar in a mountain-east smoke-stand-by island, where (a) is sea clutter received by an antenna scanning mode and (b) is sea clutter received by an antenna residence mode.
Fig. 2 shows the actual azimuth scanning angle range of the radar.
Fig. 3 is a flow chart of the present invention.
Fig. 4 is a schematic diagram of peak saliency.
Fig. 5 is a schematic diagram of target detection.
FIG. 6 is a schematic diagram of target detection in gaze mode;
in fig. 6, "x" is the position of 5 beacons:
Detailed Description
Embodiments of the present invention will be described in detail below with reference to the accompanying drawings and examples.
The invention relates to a sea clutter identification and target detection method independent of a probability model, which comprises the steps of firstly calculating the number of echo pulses received when radar beams reside in a certain azimuth according to the scanning speed of a radar antenna, the width of a main beam of the antenna and the pulse repetition period, segmenting radar echo data in the azimuth direction, then carrying out windowing Fourier transform on the segmented echoes along the azimuth direction to form a distance X Doppler (RD) echo matrix, carrying out peak detection on the RD echo matrix, clustering the detected peaks by using subtractive clustering and fuzzy C-means clustering to form a characteristic matrix of sea clutter and targets in each azimuth direction, classifying the characteristic matrix by using subtractive clustering and K-means clustering again, and regarding the maximum classification of the gauge mode as sea clutter, wherein the processing process can be regarded as self-adaptive filtering independent of an echo statistical model, and the sea clutter intensity in the processed radar echo data is effectively suppressed, thereby being beneficial to improving the detection probability of subsequent targets.
The specific steps of the invention are shown in fig. 3, comprising:
s1) taking echo of radar scanning whole monitoring sea surface as an object, and observing the rotation speed omega of a radar antenna according to sea E Beam azimuth width θ of radar A And pulse repetition frequency PRF, calculating radar azimuth pulse number n p ：
Taking one of the radar echo data files 20191014020821_01_scanning. Mat as an example, namely, the first sea echo data experimentally collected in the No. 1 of the smoke counter and horse island at the time of 2 minutes and 21 seconds of 10 months in 2019 illustrates the main performance index of the radar. The radar being an X-band radar (f c =9.3 to 9.5 GHz), main beam azimuth width θ A =1.2°, vertical beam width θ E =20°. Antenna rotation speed omega E =24r/min. The radar transmit pulse is "mode 2", i.e. 1 normal pulse (t1=0.04 μs) and 1 LFM pulse (t2=3 μs), distance to the sampling frequency f s The pulse repetition frequency prf=3 kHz, according to the above parameters, calculates the number of pulses transmitted by the radar at one azimuth angle as: n is n p =25. If the radar antenna is operated in the stay mode, i.e. the rotation speed of the antenna is approximately zero, let n p ＝100。
S2) according ton p Dividing echo pulse N of radar scanning whole monitoring sea surface into A zbin Echo of azimuth unit, A zbin ＝floor(N/n p ) Floor (·) is a rounding function, forming n corresponding to each azimuth p ×n t Echo matrix S r ，n p Is S r Number of lines, n t Is S r I.e. the number of echo pulse samples.
The radar echo data file 20191014020821_01_scanning. Mat contains: 7362×1320 (normal pulse) and 7362×5250 (LFM pulse) 2 echo data matrices, the signal bandwidth is the same (25 MHz). The matrix row corresponds to the azimuth direction, the column corresponds to the distance direction, the normal pulse start distance gate is 77.5m, and the lfm pulse start distance is 435m. According to n p The value divides the echo equally into: a is that zbin ＝floor(7362/n p ) =163 azimuth unit echoes, and thus the whole echo data file is divided into 163 25×1320 (normal pulse) and 25×5250 (LFM pulse) echo matrices S r 。
The radar echo data file 20210106160919_01_station of the stay mode contains: 10000×2224 (normal pulse) and 10000×4346 (LFM pulse) 2 echo data matrices, according to n p Dividing 100 into 100 echo matrices S of 25 x 2224 (normal pulse) and 25 x 4346 (LFM pulse) r 。
S3) pair S along azimuth direction r Each column is subjected to a windowed Fourier transform, which is converted into an RD echo matrix S x 。
As shown in fig. 1 (a), a conventional echo pulse is shown in radar beam direction a for echo data file 20191014020821_01_scanning z RD echo matrix S when= 235.4944 ° x The method comprises the steps of carrying out a first treatment on the surface of the In fig. 1 (b), the echo data file 20210106160919_01_station is the RD echo matrix S pointed by the fixed beam x 。
S4) RD echo matrix S for detecting all azimuth orientations of radar x By setting a "minimum peak significance" during peak detection, the spurious signals in the sea clutter are limited to a low degree.
[pks,locs,w,p]＝findpeaks(Sx(:,i),′MinPeakProminence′,minP)(2)
Equation (2) is Matlab function for detecting spectrum sequence peak value, the input Sx of the function (i) is RD echo matrix S x Column i (i=1, 2, …, n t ) The input option 'minpeakpominance', minP is "minimum peak saliency".
As shown in FIG. 4, only the peak significance is higher than the RD echo matrix S x The minimum value of the column mean, i.e., minp=min (mean (S x ) Is identified as a valid peak. The output of the function [ pks, locs, w, p]The height, position, width and significance of the spectral peaks, respectively, were confirmed to be valid.
S5) to S x The Doppler frequency f is found from the location locs of the height pks of the spectral peak for which each column is validated d Extracting peak value of pks, determining distance unit r according to column number, and constructing S x Target feature matrix F of (2) 1 ＝[pks，w,p,f d ,r]。F 1 The structure of the system does not depend on any statistical model, can adaptively record the characteristics of sea clutter and targets in time, frequency and space domain, and is favorable for distinguishing the sea clutter and the targets.
S6) estimating the target feature matrix F by using the subtractive clustering (subtractive clustering) 1 Is a target type number of (1):
C 1 ＝subclast(F 1 ,radii),0<radii<1
C 1 for the centroid matrix of subtractive clustering, radii is the cluster radius, let k 1 ＝length(C 1 1), i.e. k 1 Is C 1 To number of lines of k 1 And F 1 Obtaining centroid matrix C of target cluster as input of fuzzy C-means cluster 2 ＝fcm(F 1 ，k 1 ) And stored in matrix T as the target feature of the preliminary recognition g ，T g Number N of lines of (2) t For the number of valid targets detected by the azimuth radar.
The step utilizes subtractive clustering and fuzzy C-means clustering to effectively extract the feature matrix F 1 The target characteristics, thereby ensuring the accuracy of the target matrix Targets.
S7) repeating S3) to S6), and processing all parties of radar beamsBit-directed RD echo matrix S x . Matrix S of RD echoes x Matrix T of (2) g Longitudinally arranged as target matrix targets= [ T gj ]，j＝1，2，…，A zbin ，T gj The radar beam is directed to the sub-matrix in j corresponding Targets.
S8) estimating the number k of target types of the target matrix Targets by using the subtractive clustering method of S6) again 2 Will k 2 And the Targets are used as the input of K-means clustering, and the number and the mass centers of various Targets in the target matrix Targets are calculated by using the K-means clustering:
[idx,C t ]＝kmeans(Targets,k 2 )
wherein idx comprises the target type in Targets, C t Is k 2 Centroid matrix of individual targets. C (C) t Corresponds to a centroid of a category, e.g. C t Line 1 of (c) corresponds to category 1. Extracting submatrices { A from target matrix Targets using idx j }，j＝1,2,…,k 2 And identifying the submatrix with the largest line number as a sea clutter feature matrix, and identifying other submatrices as target feature submatrices to be identified.
The step utilizes subtractive clustering and k-means clustering to effectively separate sea clutter and Targets in a target matrix Targets.
S9) extracting the submatrix { A } j Target feature submatrix to be identified in the matrix and find out the corresponding centroid matrix C t Is used for constructing a target feature matrix F 2 The method comprises the steps of carrying out a first treatment on the surface of the For each radar beam azimuth pointing, according to the effective target number N in S6) t Extracting corresponding target submatrices T in target matrixes Targets gj 。T gj Is a certain RD echo matrix S obtained by fuzzy C-means clustering x Is { A } j The } is formed by each T through K-means cluster pairs gj Results after clustering of Targets arranged longitudinally have different meanings.
Let t=f 2 ∩T gj If T is an empty set, the corresponding beam is directed to a non-target; conversely, read column 4 [ f ] of each row T dj ]And into a movement velocity v relative to the radar rj ]Extracting distance unit value [ r ] corresponding to 5 th column of each row of T kj ]Output target speed and position vector P v ＝[v rj ,r kj ]. j=1, 2, … k, k being the number of lines (target number) of T. By the method, false tracks caused by sea clutter are effectively suppressed.
S10) applying a track association method to the velocity and position vectors P of adjacent azimuth units v And carrying out preliminary track association, if the target track almost penetrates all radar beam azimuth pointing angles, judging as a false target caused by sea clutter, taking a short track or an isolated point as a current target detection result, and verifying the accuracy of the next radar scanning data to be the target of detecting a ship or a low-altitude flight from a plurality of radar scanning periods.
The sea clutter data 2019101110708_01_scanning. Ma is processed as follows:
s5): the target number of the target matrix Targets is detected: 891
S6) in: subtracting and clustering the target matrix Targets, and detecting the number of the target types: 3, clustering by k-means and extracting corresponding submatrices A from target matrixes Targets 1 (389×5)，A 2 (497X 5) and A 3 (5X 5); wherein A is 2 Is determined as sea clutter, A 1 And A 3 Is an effective target feature submatrix;
s7): from A 1 And A 3 Center of mass matrix C t The nearest row vector of (a) forms the target feature matrix F 2 The targets for each azimuth are as follows:
beam azimuth angle: 235.5197 find 1 target speed: 37.7626m/s distance 4258.1354m;
beam azimuth angle: 19.6462 find 1 target speed: 1649.4463m/s distance 3545.584m;
20191014020821_07_scanning. Mat was processed in the same procedure as follows:
beam azimuth angle: 236.7094 find 1 target speed: 42.4656m/s distance 4244.6885m
Beam azimuth angle: 29.2903 find 1 target speed: 1620.0313m/s distance 3511.4924m
From the above detection results, it can be seen that there are 2 moving targets driven off/toward the radar at different times in the same sea area, the sea clutter is completely eliminated, t=f in S9) 2 ∩T gi In target detection, no false track occurs in which the target track extends almost throughout all radar beam azimuth angles (see (a) and (b) in fig. 5, which are target position maps detected from 2 data files, respectively). It should be noted here that if the target feature matrix F is not constructed 2 And directly using the target feature submatrix { A } j If target detection is performed, a certain number of false tracks will be generated, and the reason for this is that: in order to eliminate the diversity caused by the non-stable sea clutter to the maximum extent, a target matrix target is generated in S6) by adopting fuzzy clustering, thereby obtaining a target characteristic submatrix { A j The characteristic vector of the sea clutter still remained partially is adopted to construct a target characteristic matrix F 2 Effectively eliminating the influence of sea clutter and accurately detecting 2 moving targets. The speed of motion of the target should include Doppler frequency caused by rotation of the radar antenna.
Processing results of 20210106160919_01_staring. Mat gaze mode sea clutter:
beam azimuth angle: 17.9092 find 1 target speed: 175.4581m/s distance 7076.3234m
Beam azimuth angle: 17.9092 find 1 target speed: 31.3129m/s distance 6390.8806m
Beam azimuth angle: 17.9219 find 1 target speed: 159.025m/s distance 6422.6343m
Fig. 6 is a schematic diagram of the positions of the above targets, and it can be seen that the sea clutter pointed by the radar beam is effectively suppressed, and 3 targets are clearly visible.
The innovation of the method provided by the invention is as follows: the method takes echo data of the whole scanning period of the sea observation radar as a processing object, utilizes a fuzzy clustering method to identify sea clutter in the sea observation radar, realizes the detection and positioning of targets on the basis, and is a signal processing method independent of a sea clutter probability distribution model.
Claims (7)
1. The sea clutter identification and target detection method independent of the probability model is characterized by comprising the following steps of:
s1) taking echo of radar scanning whole monitoring sea surface as an object, and observing the rotation speed omega of a radar antenna according to sea E Beam azimuth width θ of radar A And pulse repetition frequency PRF, and grouping the echo pulses into echo matrixes S corresponding to all azimuth directions r ；
S2) pairs of echo matrices S in azimuth r Performing windowed Fourier transform to convert it into RD echo matrix S x ；
S3) RD echo matrix S for detecting all azimuth orientations of radar x Is used for constructing a target feature matrix F 1 ；
S4) aiming at a target characteristic matrix F 1 Performing subtractive clustering to estimate the number k of target types 1 Will k 1 And F 1 Obtaining centroid matrix C of target cluster as input of fuzzy C-means cluster 2 Sequentially superposing to form a target matrix Targets;
s5) performing subtractive clustering on the target matrix Targets to estimate the number k of target types 2 Will k 2 And Targets as inputs to K-means clustering to obtain K 2 Centroid matrix C of individual targets t Submatrices { A } of corresponding classes j }，j＝1,2,…,k 2 Identifying the submatrix with the largest line number as a sea clutter feature matrix, and identifying other submatrices as target feature submatrices to be identified;
s6) extracting the submatrix { A } j Target feature submatrix to be identified in the matrix and find out the corresponding centroid matrix C t Is used for constructing a target feature matrix F 2 The method comprises the steps of carrying out a first treatment on the surface of the For each radar beam azimuth direction, according to the effective target number N t Extracting corresponding target submatrices T in target matrixes Targets gj Let t=f 2 ∩T gj Obtaining a target speed and position vector P v ；
S7) detecting targets in all directions, and eliminating false targets by utilizing track association.
2. According toThe method for sea clutter recognition and target detection independent of probability model as claimed in claim 1, wherein in S1), the number n of radar azimuth pulses is calculated first p ：
Then according to n p Dividing echo pulse N of radar scanning whole monitoring sea surface into A zbin Echo of each azimuth unit forms an echo matrix S corresponding to each azimuth r ，S r Is n p ×n t Matrix of A zbin ＝floor(N/n p ) Floor (·) is a rounding function, n t The number of points is sampled for the echo pulses.
3. The method for identifying sea clutter and detecting targets independent of probability models according to claim 2, wherein in S3), a Matlab function for detecting peaks of a spectrum sequence is constructed as follows:
[pks，locs,w,p]＝findpeaks(Sx(:，i),′MinPeakProminence′,minP)
wherein the input Sx of the function (i) is RD echo matrix S x I=1, 2, …, n t Inputting an option 'MinPeakProminance', wherein minP is the minimum peak significance; the outputs pks, locs, w, p of the function are the height, position, width and significance of the spectral peaks, respectively, that are confirmed to be valid;
by S x The Doppler frequency f is found from the location locs of the height pks of the spectral peak for which each column is validated d Extracting peak value of pks, determining distance unit r according to column number, and constructing S x Target feature matrix F of (2) 1 ＝[pks,w，p,f d ，r]。
4. The method for sea clutter recognition and target detection independent of probability model according to claim 1, wherein in S4), the centroid matrix C of subtractive clustering is calculated by the following formula 1 ：
C 1 ＝subclast(F 1 ,radii),0<radii<1
Wherein radii is the cluster radius, let k 1 ＝length(C 1 1), i.e. k 1 Is C 1 The number of rows of (3); centroid matrix C of target cluster 2 ＝fcm(F 1 ,k 1 ) C is carried out by 2 Target features as primary identification are stored in matrix T g ，T g Number N of lines of (2) t For the effective target number detected by the radar in the corresponding azimuth, repeatedly processing the RD echo matrix S pointed by each azimuth of the radar beam x Matrix S of RD echoes x Matrix T of (2) g Longitudinally arranged as target matrix targets= [ T gj ]，j＝1，2，…，A zbin ，T gj The radar beam is directed to the sub-matrix in j corresponding Targets.
5. The method for sea clutter recognition and target detection independent of probability model according to claim 4, wherein in S5), the centroid matrix C is calculated by the following formula t ：
[idx,C t ]＝kmeans(Targets,k 2 )
Wherein idx contains the target type in Targets, and idx is used to extract submatrices { A } from target matrix Targets j }。
6. The method for identifying sea clutter and detecting targets independent of probability models according to claim 5, wherein in S6), if T is an empty set, the corresponding beam is directed to a no-target; conversely, read column 4 [ f ] of each row T dj ]And into a movement velocity v relative to the radar rj ]Extracting distance unit value [ r ] corresponding to 5 th column of each row of T kj ]Output target speed and position vector P v ＝[v rj ,r kj ]J=1, 2, … k, k is the number of lines of T, i.e. the target number.
7. The method for sea clutter identification and target detection independent of probability model according to claim 6, wherein in S7), a track correlation method is applied to adjacent azimuth unitsVelocity and position vector P v And carrying out preliminary track association, if the target track almost penetrates all radar beam azimuth pointing angles, judging as a false target caused by sea clutter, taking a short track or an isolated point as a current target detection result, and verifying the accuracy of the next radar scanning data to be the target of detecting a ship or a low-altitude flight from a plurality of radar scanning periods.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110533736.5A CN113447901B (en) | 2021-05-17 | 2021-05-17 | Sea clutter identification and target detection method independent of probability model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110533736.5A CN113447901B (en) | 2021-05-17 | 2021-05-17 | Sea clutter identification and target detection method independent of probability model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113447901A CN113447901A (en) | 2021-09-28 |
| CN113447901B true CN113447901B (en) | 2023-11-17 |
Family
ID=77809801
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110533736.5A Active CN113447901B (en) | 2021-05-17 | 2021-05-17 | Sea clutter identification and target detection method independent of probability model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113447901B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116503746B (en) * | 2023-06-29 | 2023-09-12 | 南京信息工程大学 | Infrared small target detection method based on multilayer nested non-full-mapping U-shaped network |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH1031067A (en) * | 1996-03-28 | 1998-02-03 | He Holdings Inc Dba Hughes Electron | Radar target tracking system in background clutter |
| CN106970386A (en) * | 2017-03-31 | 2017-07-21 | 西安电子科技大学 | A kind of optimization method of RADOP beam sharpening |
| CN109100696A (en) * | 2018-07-06 | 2018-12-28 | 西安电子科技大学 | Microinching target removing method based on point clutter figure |
| CN112162269A (en) * | 2020-09-28 | 2021-01-01 | 西安大衡天成信息科技有限公司 | Sea clutter suppression and target detection method based on singular spectrum decomposition |
-
2021
- 2021-05-17 CN CN202110533736.5A patent/CN113447901B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH1031067A (en) * | 1996-03-28 | 1998-02-03 | He Holdings Inc Dba Hughes Electron | Radar target tracking system in background clutter |
| CN106970386A (en) * | 2017-03-31 | 2017-07-21 | 西安电子科技大学 | A kind of optimization method of RADOP beam sharpening |
| CN109100696A (en) * | 2018-07-06 | 2018-12-28 | 西安电子科技大学 | Microinching target removing method based on point clutter figure |
| CN112162269A (en) * | 2020-09-28 | 2021-01-01 | 西安大衡天成信息科技有限公司 | Sea clutter suppression and target detection method based on singular spectrum decomposition |
Non-Patent Citations (1)
| Title |
|---|
| 相控阵雷达对海面目标的检测跟踪方法;吕江涛;肖靖;徐振来;;现代雷达(第04期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113447901A (en) | 2021-09-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11906616B2 (en) | Method of target feature extraction based on millimeter-wave radar echo | |
| CN104215951B (en) | System and method for detecting low-speed small target under sea cluster background | |
| CN109655822A (en) | A kind of improved track initiation method | |
| CN109633598B (en) | Array radar target detection method based on characteristic analysis | |
| CN109100696B (en) | Slow-speed moving target elimination method based on dot clutter map | |
| CN105785340B (en) | The angle measurement and recognition methods that monopulse radar interferes target in main lobe and bait | |
| CN107576959B (en) | High repetition frequency radar target tracking method before detection based on area mapping deblurring | |
| JP2018205175A (en) | Radar device and radar signal processing method thereof | |
| CN113447901B (en) | Sea clutter identification and target detection method independent of probability model | |
| CN115061113A (en) | Target detection model training method and device for radar and storage medium | |
| Sim et al. | Road environment recognition for automotive FMCW radar systems through convolutional neural network | |
| CN110988867B (en) | Elliptical cross target positioning method for one-transmitting and double-receiving through-wall radar | |
| CN109581366B (en) | Discrete sidelobe clutter identification method based on target steering vector mismatch | |
| CN113093174B (en) | PHD filter radar fluctuation weak multi-target-based pre-detection tracking method | |
| CN115657005A (en) | Unmanned aerial vehicle identification method based on Doppler radar multi-information fusion | |
| CN113625266A (en) | Method, device, storage medium and equipment for detecting low-speed target by using radar | |
| CN109934840A (en) | Circumference SAR motion target tracking method based on GMPHD filter | |
| Stadelmayer et al. | Out-of-distribution detection for radar-based gesture recognition using metric-learning | |
| CN116879863B (en) | Multi-target measuring method and system for continuous wave 4D millimeter wave radar | |
| Wang et al. | Suppress false alarms by exploiting ambiguity features of targets with machine learning | |
| Zhang et al. | An efficient track-before-detect algorithm based on complex likelihood ratio in radar systems | |
| CN116008946B (en) | Automatic judging method and system for plasma sheath of near space high dynamic aircraft | |
| CN112731399B (en) | Broadband radar aircraft target frequency identification method based on two-dimensional sparse structure | |
| CN112965038B (en) | Ground activity target classification method based on trace point data and frequency spectrum characteristics | |
| Xu et al. | Sea-Surface Floating Small Target Detection Based on SVM and Multidimensional Features |
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 |