CN108254727A - A kind of radar plot condensing method based on Contour extraction - Google Patents
A kind of radar plot condensing method based on Contour extraction Download PDFInfo
- Publication number
- CN108254727A CN108254727A CN201711294042.0A CN201711294042A CN108254727A CN 108254727 A CN108254727 A CN 108254727A CN 201711294042 A CN201711294042 A CN 201711294042A CN 108254727 A CN108254727 A CN 108254727A
- Authority
- CN
- China
- Prior art keywords
- radar
- target
- data
- orientation
- contour extraction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- 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
Abstract
The invention discloses a kind of radar plot condensing methods based on Contour extraction, and it is high to solve the problems, such as that progressive scan data cause to take.Its realization includes, and inputs radar information；It determines that radar bearing samples span, span azimuth sample is pressed to inputting video data；Traversal finds the data cell for having detection bit flag across the sampled point in orientation；Contour extraction is carried out to the data cell for having detection bit flag determined；All data cells in search profile are obtained with region-growing method；Plot coherence processing is carried out, Targets Dots in radar video data is completed and agglomerates.The present invention changes progressive scan operating mode, a kind of Plot coherence method of quick lookup target is formed across azimuth sample combination Contour extraction, region-growing method, only need very small part in ergodic data, find out interesting target, save a large amount of calculating, it is theoretical to prove there is no target missing inspection with emulation, meet high-resolution radar data processing real-time.Data processing is agglomerated for radar plot.
Description
Technical field
The invention belongs to Radar Technology fields, further relate to the detection of radar data processing target, specifically a kind of base
In the radar plot condensing method of Contour extraction.The present invention can be used for fast by the contour feature of target during Radar Targets'Detection
Speed detects target, and the target to detecting carries out Plot coherence.It is quickly detected for the target of High Accuracy Radar.
Background technology
Plot coherence is exactly to judge whether certain point mark block is target and estimates in the detection of radar data processing target
Position of the barycenter of this mark block as target.Target detection refers to that the data after the quantization obtained by radar return are found
Go out the process of target.Radar plot cohesion is always the pith in Radar Technology field, is the key content of data processing
One of.
Zhongzhi L etc. are in paper " High resolution radar data fusion based on
Clustering algorithm " (in November, 2010 27-28 meeting IEEE 2nd International Workshop
On Database Technology and Applications) in propose it is a kind of for high-resolution radar point mark sub-clustering
Fast algorithm, it is more efficient than other several existing algorithms that the algorithm is demonstrated in paper.Simultaneously in a mark clustering process
In, since destination number is unknown and time-varying, classical cluster-dividing method K-means methods can not work here.At these
All data will be called twice in method, and since data volume is larger, institute is time-consuming more.
Luo Liyan etc. is in paper " Improved centroid extraction algorithm for autonomous
star sensor”(《IET Image Processing》9th phase in 2015, page 901) in propose it is a kind of efficiently for star
The extraction algorithm of planet barycenter in figure.In that article, Arbabmir MV etc. be compared in " Improving night sky
star image processing algorithm for star sensors”。(《Journal of the Optical
Society of America A》The third phase in 2014, page 794) in the algorithm of target centroid extraction that proposes, experiment proves
Its method has obtained the more high barycenter of precision efficient while, and all data of the method call of Luo Li swallows etc. are primary herein,
All data of method call of Arbabmir MV etc. are twice.
Merely by Shu-Yen W. etc. " Symmetric region growing " (《IEEE Transactions
on Image Processing》12 phases page 1007 in 2003) region-growing method and the thought of sampling calculate the matter of target
The heart, only partial data are called, therefore greatly reduce calculation amount, and higher effect has been obtained at no large area clutter area
Rate, if but have large area in radar search coverage for clutter area, the data in clutter area can be called repeatedly, occupy largely
Calculating, so that requirement of real-time can not be met.
It can be seen that all there is the characteristics of repeatedly traversing all echo point marks in Plot coherence algorithm described above, high-precision
In the case of spending radar, these methods will repeatedly traverse all data, and calculation amount is huge, it is impossible to meet radar real-time
Demand.
Invention content
The present invention in view of the deficiency of the prior art, proposes a kind of small radar based on Contour extraction of calculation amount
Plot coherence method.
A kind of radar plot condensing method based on Contour extraction, which is characterized in that include the following steps：
(1) radar information is inputted,
The parameter of (1a) input radar system requirement
According to the system requirements of radar, the running parameter of radar is inputted, input parameter includes the beam angle of radar, radiation
Range, azimuth dimension required by radar target and apart from dimension size information；
(1b) inputs radar video data
It radar video data will be inputted after Radar Signal Processing；Radar video data packet contain target orientation, away from
From, amplitude information and check bit flag information；
(2) it determines that radar bearing samples span, azimuth sample is carried out by span to inputting video data
The span of azimuth sample is determined according to across azimuth sample principle, from Initial Azimuth, it is entire to complete radar by orientation span
Azimuth sample on inputting video data obtains the data cell in each sampling orientation after the completion of sampling；
Across azimuth sample principle：Span across orientation is necessarily less than the orientation extension number of minimum extension target；Foundation
The system requirements of radar determines extension size of the minimum target in orientation；
(3) traversal finds the data cell for having detection bit flag across the sampled point in orientation
All data cells in present sample orientation are traversed, judge whether the sampling unit has detection bit flag, if
Data cell has detection bit flag, performs step (4), carries out Contour extraction；If data cell does not detect bit flag, continue
Traverse all remaining sampling units；If present orientation it is all data cell traversal complete, it is next sampling orientation on after
It is continuous to perform step (3)；It is completed until the sampled data in all orientation traverses, radar plot agglomeration step is completed；
(4) Contour extraction is carried out to the data cell for having detection bit flag that traversal sampling determines
Using current detection bit flag unit as Contour extraction start element, recycle and search out successively according to Contour extraction criterion
Next profile unit, for the first time back to Contour extraction start element during finding, then Contour extraction completion, preservation is searched
Rope outline data obtains the search profile corresponding to current detection bit flag unit, obtains a mark block, this mark block is by taking turns
The data cell with detection bit flag that the obtained search profile of exterior feature tracking and search contoured interior contain；If this point
The azimuth dimension of mark block and the condition for meeting radar system target simultaneously apart from dimension, are judged as radar target, then perform step (5),
Otherwise, step (3) is performed to continue to traverse remaining data cell；
(5) all data cells in search profile are obtained with region-growing method
For the radar target for judging to obtain in step (4), obtained in radar target search profile using region-growing method
All data cells with detection bit flag in portion simultaneously preserve the data, and all point marks of the radar target are included search takes turns
The data cell that wide and growth method traverses, referred to as radar target point mark give Plot coherence module and carry out Plot coherence, i.e.,
Perform step (6)；
(6) agglomeration process of mark is put
The radar target point mark of acquisition is condensed processing according to the method for amplitude weighting；Radar mesh after being agglomerated
Cursor position information, after save location information data, return to step (3) continues to traverse remaining data cell.
Compared with the prior art, the present invention has the following advantages：
Relative to traditional radar plot agglomerative algorithm, the present invention selects the method across azimuth sample only to need time traversed
Small part in wave number video data, so as to save calculation amount.
First, the present invention is to traverse video data based on the method across azimuth sample, since the present invention is according to radar target
Data characteristic using radar target is extracted across azimuth sample combination Contour extraction, repeatedly traverses all videos with conventional method
It is compared under real-time when data, a large amount of cycle calculations cause the whole system to run, the present invention reduces the data volume of processing
To between several times to tens times, the data of processing greatly reduce, improve the performance of system.
Second, the present invention sets the sampling span across orientation, traversal is across side after azimuth sample according to radar design parameter
The data cell of position sampling can quickly find radar target profile according to Contour extraction method, improve Objective extraction efficiency, together
When Objective extraction precision do not reduce.Meet in modernized war to the required precision of system.From theoretical and two sides of emulation
Face proves the situation of no target missing inspection.
Third, the present invention introduce algorithm of region growing on the basis of Contour tracing, are compared and sent out by experiment simulation
In the case that present elapsed time is basically unchanged, whether in the case of the low sea condition or in the case of high sea condition, association area
Algorithm after the growth method of domain obtains more accurate target position information.
4th, since the present invention is by the way that across azimuth sample, Plot coherence, this method are carried out with reference to contour tracing method
The quantity of ergodic data is reduced, therefore this algorithm saves the time that Plot coherence consumes in radar data processing, the algorithm
The calculating time meet the requirement of High Accuracy Radar real-time.
Description of the drawings
Fig. 1 is to extra large radar illumination object delineation；
Fig. 2 particular flow sheets of the present invention；
Fig. 3 positions wave beam is inswept object module schematic diagram；
Fig. 4 objective contour detects schematic diagrams；
Fig. 5 Contour extraction method total algorithm schematic diagrames；
Fig. 6 Contour extraction method result schematic diagrams；
Fig. 7 radar real video data (one) schematic diagram；
Fig. 8 radar real video data (two) schematic diagram.
Specific embodiment
The step of the present invention is described in detail below in conjunction with the accompanying drawings.
Embodiment 1
Using conventional point mark condensing method, region growing method, planet centroid algorithm and is chased after SDDC algorithms based on profile
The object extraction algorithm processing high-precision of track surveys radar video data to maritime patrol.All there are one for Plot coherence algorithm described above
It is secondary or the characteristics of repeatedly traverse all video datas, it is primary or repeatedly traverse entire video in the case of High Accuracy Radar
Data, calculation amount is huge, and time-consuming, it is impossible to meet requirement of the modern radar system to real-time.The present invention proposes a kind of thus
Radar plot condensing method based on Contour extraction, referring to Fig. 1, Fig. 1 is to extra large radar illumination object delineation, when the wave of radar
During target that beam is inswept, if antenna bearingtWith initial azimuthTo termination azimuthInitial Azimuth meetsWhen,
The amplitude for the cell being considered as on this antenna direction receives the influence of the target.
Radar signal data processing system generally comprises signal processing module, Plot coherence module, track association module and
Target tracking module.
Referring to Fig. 2, Fig. 2 is particular flow sheet of the present invention；The present invention is based on the radar plot condensing method packets of Contour extraction
It includes and has the following steps：Input radar system requirement parameter and radar video data, by being sampled across bearing data, Contour extraction obtains
Contouring, region-growing method obtain target internal point mark information, carry out Plot coherence to target.
(1) radar information is inputted：
The parameter of (1a) input radar system requirement
According to the system requirements of radar, the running parameter of radar is inputted, input parameter includes the beam angle of radar, radiation
Range, radar rotating speed, pulse number, azimuth dimension required by radar target and apart from dimension size information；Radar illumination target mistake
Journey：When the inswept target of the wave beam of radar, if antenna bearingtWith initial azimuthTo termination azimuthInitial Azimuth is expired
FootWhen, the amplitude for the cell being considered as on this antenna direction receives the influence of the target, initial azimuthIt arrives
Terminate azimuthOrientation coverage area be calculated by the distance between radar beam width and target to radar.
In order to which judging unit lattice are with the presence or absence of target, an object function I is definedm(), the detection width of radar CFAR detection
Angle value (xm,ym) and video data range value (x (i, j), y (i, j)), wherein (i, j) represents, in the position of x-y plane, to pass through mesh
Scalar functions Im() calculates, if object function Im() value is less than 1, then it is assumed that target is present in cell (i, j), input
Radar video data cell detection flag information be 1, that is, have detection bit flag, otherwise it is assumed that target is not present in cell
In (i, j).
(1b) inputs radar video data
The radar video data of quantization after Radar Signal Processing are input to radar plot agglomeration process module.Thunder
Each data cell has orientation, distance, amplitude information and check bit flag information, check bit flag information up in video data
It is usually represented with 0 or 1, wherein 1 indicates detection bit flag, 0 represents without detection bit flag.
(2) it determines that radar bearing samples span, inputting video data is carried out by span across azimuth sample：
The span of azimuth sample is determined according to across azimuth sample principle, azimuth sample is from Initial Azimuth, by orientation span
The azimuth sample on entire inputting video data is completed, orientation up-sampling is data sheets all in the video data orientation
Member obtains all data cells in the orientation；The data cell in each sampling orientation is obtained after the completion of sampling.In data cell
Include check bit flag information, that is to say, that the data cell having has detection bit flag, and some data cells do not have check bit
Mark.
Across the azimuth sample principle of the present invention is：Span across orientation is necessarily less than the orientation extension of minimum radar target
Number；System requirements according to radar determines extension size of the minimum target in orientation.
(3) traversal finds the data cell for having detection bit flag across the sampled point in orientation：
All data cells in present sample orientation are traversed, judge whether the sampling unit has detection bit flag, if
Data cell has detection bit flag, performs step (4), carries out Contour extraction；If data cell does not detect bit flag, continue
Traverse all remaining sampling units.
If all data cell traversals of present orientation are completed, step (3) is continued to execute in next sampling orientation；
It is completed until the sampled data unit in all orientation traverses, radar plot agglomeration step is completed.
(4) Contour extraction is carried out to the data cell for having detection bit flag that traversal sampling determines：
Using current detection bit flag unit as Contour extraction start element, recycle and search out successively according to Contour extraction criterion
Next profile unit, for the first time back to Contour extraction start element during finding, then Contour extraction completion, preservation is searched
Rope outline data obtains the search profile corresponding to current detection bit flag unit, obtains a mark block, this mark block is by taking turns
The data cell with detection bit flag that the obtained search profile of exterior feature tracking and search contoured interior contain.If this point
The azimuth dimension of mark block and the condition for meeting radar system target simultaneously apart from dimension, are judged as radar target, then perform step (5),
Otherwise, step (3) is performed to continue to traverse remaining data cell.
(5) all data cells in search profile are obtained with region-growing method：
For the radar target for judging to obtain in step (4), obtained in radar target search profile using region-growing method
All data cells with detection bit flag in portion simultaneously preserve the data, and all point marks of the radar target are included search takes turns
The data cell that wide and growth method traverses, referred to as radar target point mark give Plot coherence module and carry out Plot coherence, i.e.,
Perform step (6).
(6) agglomeration process of mark is put
The radar target point mark of acquisition is condensed processing according to the method for amplitude weighting；Radar mesh after being agglomerated
Cursor position information, after save location information data, return to step (3) continues to traverse remaining data cell.
Since High Accuracy Radar detection range is remote, precision is high, and the data volume of generation is very huge, in practical engineer application
In, hardware system is generally difficult to handle these data in real time, generation system time delay or even causes system crash.In view of war
It strives under environment, if data processing is not in time, by the influence of bringing on a disaster property, changes war pattern.Due to radar video data
Amount is very big, and High Accuracy Radar, which often scans a circle, can generate the data volume of several G or more than ten of G, and one or many traversals are whole
The time that a data expend is long.The present invention expands this research, using across azimuth sample traversal using data, with Contour extraction
It is combined with the method for region growing, successfully solves the data processing amount reduced, improve the processing speed of system.
Embodiment 2
Radar plot condensing method based on Contour extraction is with embodiment 1, and radar is across the span of azimuth sample in step (2)
It determines to be as follows：
2.1 determine sampling span：By the radar system parameters information of input, the maximum span i.e. A of target bearing is calculated
Minimum value, that is, radar, across the maximum span of azimuth sample, ordinary circumstance down-sampling span should be less than the minimum of the A values
Value sets across the orientation size of radar by the orientation minimum value being calculated, in a practical situation in order to improve system speed, drops
Low time delay, can be by increasing the numerical value of parameter, and increase sampling span value is realized.
After 2.2 sampling spans determine, carried out from the radar video data of input across azimuth sample, each orientation span choosing
A scan line is taken to be detected.The size of orientation span is calculated by the first step.Sampling is completed to obtain each sampling orientation
On data cell.
The present invention is that video data is sampled by span, then traverses the video data after resampling again, and is passed
The method of system is directly one or many entire video datas of traversal, and calculation amount is huge, and time-consuming.It, can in radar system
It is quick to find target, it is the important indicator of radar system, the present invention can guarantee the minimum radar for enough detecting radar system requirement
Target, does not reduce the precision of system processing while ensure that and reduce data processing complexity, scanning search target is more fast
Speed takes shorter.
Embodiment 3
Radar plot condensing method based on Contour extraction is with embodiment 1-2, with the check bit data of sampling in step (4)
Contour extraction is carried out, is as follows：
4.1 from present orientation data cell scan line traverse data cell in the orientation of sampling one by one from down to up,
Until the data cell for finding to have detection bit flag.
4.2 when finding to have the data cell of detection bit flag, is risen using the data cell for having check bit as Contour extraction
Beginning unit sequentially finds next profile unit according to Contour extraction criterion.
Contour extraction criterion：On the basis of active cell, traversed successively since a direction adjacent with the unit and base
The adjacent all data cells of quasi- unit, if finding the unit for having check bit, preserve after data using the data cell that finds as
Benchmark continues to begin stepping through the adjacent data cell of current data unit, there is the data cell of detection bit flag until finding.
It is 8 neighborhood units referring to 8 adjacent data cells of the data cell surrounding of Fig. 4 (a) black dot expression, with this
Data southeast corner data cell starts the number consecutively from 0 to 7.If the data cell is profile start element since 0 successively
Traversal counterclockwise.If not start element, active cell is the grey in 0 or No. 7 cell position such as figure of previous search unit
Unit from No. 6 of 8 neighborhood of current data unit when then searching for, is shown in oblique line dot unit search spread counterclockwise；If active cell
It is 1 or No. 2 cell position in previous search unit, then is searched counterclockwise from No. 0 unit of 8 neighborhood of current data unit when searching for
Rope traverses；If active cell is 3 or No. 4 cell positions in previous search unit, from 8 neighborhood of current data unit when searching for
No. 2 unit search spreads counterclockwise；If active cell is 5 or No. 6 cell positions in previous search unit, when searching for from
No. 4 unit search spreads counterclockwise of 8 neighborhood of current data unit.
The point mark of looked for profile is preserved and then detection bit flag is set to 0, until the profile unit found rises for profile
The cell of beginning, Contour searching terminate.
4.3 judge the search profile size looked for azimuth dimension and apart from dimension whether meet step (4) simultaneously in formula require
Constraints is radar target if constraints is met, is not otherwise.Step (5) is carried out if meeting, is otherwise continued
Perform step (3).
After 4.4 have searched range cell all in present sample orientation, continue in next orientation, next orientation number
According to repetition step (3).
The prior art looks for radar video data of the radar target profile by one or many progressive scans, by finding mesh
Profile is marked, the present invention need to only be traversed obtains radar video data after azimuth sample, only traverses very little in entire video data
A part is with regard to that can find interested radar target profile.Radar target profile is found out by the method for the present invention, compared to others
Method faster reduces the time delay of system.The situation of no target missing inspection is proved in terms of theoretical and emulation two.
Embodiment 4
Radar plot condensing method based on Contour extraction with embodiment 1-3, the agglomeration process of step (6) point mark it is specific
Step is as follows：
The 6.1 point marks for obtaining region-growing method are the data and Contour tracing that have check bit label searched in profile
The point mark of acquisition is searched for outline data and is merged, and forms the point mark of entire radar target point mark block；
6.2 using point marks orientation, distance and amplitude information, according to Plot coherence formula by the point of entire Targets Dots block
Mark carries out the agglomeration process of a mark in the way of amplitude weighting, obtains the congealing point of a target；
6.3 give target congealing point to subsequent radar track relating module, continue radar data processing.
Embodiment 5
Radar plot condensing method based on Contour extraction with embodiment 1-4, calculating target bearing in step (2.1)
The minimum value A of maximum span, that is, Amin, specific formula for calculation is：
Wherein NAFor localizer unit lattice number.θ0Represent beam angle, symbolRepresent downward rounding.The formula calculates
What is obtained is the size that minimum radar target is tieed up in direction.If the A being calculated is less than AminThis value is it is assumed that make an uproar
Sound, small clutter or the interference signal for thering is noise and small clutter to collectively form.
The present invention is by the target that provides in the maximum span calculation formula of azimuth dimension, minimum span target bearing AminValue
Compare, while can ensureing the precision of radar, filter out the information such as clutter, false-alarm.
Embodiment 6
With embodiment 1-5, m-th mark block described in step (3) exists radar plot condensing method based on Contour extraction
Tie up A in directionmValue and apart from dimension RmValue and radar target actual size between restriction relation, be shown below.
Its constraints is
Wherein CR represents the powerful radius of radar, symbolRepresent downward rounding, initial azimuthTo termination azimuthIt is calculated by orientation coverage area by the distance between radar beam main lobe width and target to radar, φmRepresent target
The angle of long axis and x coordinate axis；NAFor localizer unit lattice number, NRRange cell lattice number, lmTarget long axis length, smTarget
Minor axis length, lmaxAnd smaxRepresent that system detects single radar target maximum length and target maximum width, θ respectively0It represents
Beam main lobe width, φmRepresent target long axis and the angle of plane coordinates axle x-axis, αmRepresent hanging down for target long axis and radar antenna
The angle of line.It is the inswept object module schematic diagram of position wave beam to see Fig. 3, Fig. 3.
Know that the interested target of radar should meet in video data such as co-relation by above formula, in other words, be discontented with
Foot illustrates such as co-relation if, illustrates that this mark block is not generated by target, it may be possible to caused by sea clutter or background object.
The Plot coherence of High Accuracy Radar is handled, a kind of Objective extraction based on Contour tracing proposed by the invention
Method, the complexity of calculating is linear, and only the point mark of some is scanned.And in processing procedure, compare
Big clutter block, which will directly be abandoned, not to be handled.Therefore, the algorithm can also good work in complex environment.Institute
With the present invention has very high practical value in real work.
Being given below one, more completely the present invention is further described with detailed example.
Embodiment 7
Radar plot condensing method based on Contour extraction with embodiment 1-6,
Step 1：Input radar information
Input radar parameter information
Wherein according to the parameter information of radar, referring to Fig. 1, Fig. 1 is extra large radar illumination object delineation, when the wave beam of radar
During inswept target, once there is partial target to be irradiated to by radar beam, be considered as the amplitude of cell on this antenna direction by
The influence of the target is arrived.In order to which judging unit lattice are with the presence or absence of target, the profile information of target can be obtained.Define one
Function Im() is defined as follows
Wherein：
If (xm,ym) and (x (i, j), y (i, j)) poor rotation alphamAgain divided by N, if quotient is less than 1, then it is assumed that target is deposited
It is this element lattice (i, j), otherwise there is no αmRepresent the angle of the vertical line of target long axis and radar antenna.(x(i,j),y(i,
J) it) represents in the position of x-y plane lower unit lattice (i, j).L and s is its systematic parameter, and referring to Fig. 3, Fig. 3 is that a wave beam is inswept
Object module schematic diagram, schematic diagram during displaying inswept target of wave beam of the schematic diagram upper left side, upper right side is antenna targeted scans
To kth apart from path partially enlarged diagram, lower section is the partial enlarged view to upper right side.
Such as upper right portion in Fig. 3, target is inswept by wave beam, if antenna bearingtWith initial azimuthTo termination orientation
AngleInitial Azimuth meetsWhen, the cell measured value in this direction is by object effects in figure.It arrivesRange can be by
Following formula calculates：
In formula, θ0Represent beam angle, RmRepresent that target arrives the distance between radar, lmTarget long axis length.
Input radar video data
Radar video data after the quantization of Radar Signal Processing are input to Plot coherence processing module.The radar
Video data includes orientation, distance, amplitude information and the detection bit flag for whether having target of target.
Step 2：Across the bearing data sampling of radar
Size across the orientation of azimuth sample is determined according to the parameter information of radar, the sampling in each orientation is all the party
All data cells on position.Principle across orientation：Size across orientation is necessarily less than the orientation extension of minimum radar target
Number.
According to the following formula, the maximum span i.e. minimum value of A of target bearing, that is, radar are calculated across the maximum of azimuth sample
The span of span ordinary circumstance down-sampling should be less than the minimum value of the value, i.e. the minimum value A of Amin, specific formula for calculation is：
Sampled point is chosen from incoming radar video data after signal processing, each orientation span is chosen one and swept
Line is retouched to be detected.
Step 3：Traverse sampled data unit
The data cell on present orientation is traversed one by one, judges whether the data cell has detection to indicate, if there is detection
Bit flag performs step 4, if traversal is completed to perform the data cell that step 3 traverses next orientation.
Step 4：To there is the starting of check bit flag data unit to carry out Contour extraction
Next profile unit is sequentially found according to Contour extraction direction on the basis of having the data cell of detection bit flag,
Until the cell that the profile unit found is profile starting, then illustrate that Contour extraction finishes, have found all profile units.It is false
Size such as the profile unit of tracking meets following formula,
And constraints
It is that a radar target performs step 5 then to think it, otherwise performs step 3.Institute in present sample orientation is searched
Some data cells, next bearing data repeat step 3.
The present invention sequentially finds next profile unit according to Contour extraction direction, until the profile unit found is profile
The cell of starting then illustrates that Contour extraction finishes, and has found all profile units.The tracking direction of definition is as shown in figure 4, figure
Dark circles are the profile unit that last iteration has been found in 4 (a), and grey dot is the profile unit this time found, wheel
Wide unit is the unit for having detection bit flag；Grey dot relative black dot has eight kinds of directions, referred to as Contour extraction direction.
For each Contour extraction direction, the cell that search is first had in current iteration is different, shown in such as Fig. 4 (a) of direction,
Arrow starting point in figure is the data cell that this finds, and the data cell that arrow is directed toward is the direction begun stepping through.In order to
It is better described, an example of contour following algorithm is illustrated in Fig. 4 (b).Black color dots are the profile for having found and having preserved
Unit, target are to sequentially find all black color dots, and a and b are the two profile cells obtained at first, and direction is a to b, then for
That find next next search of profile cell is b0, followed by b1, subsequently b2, i.e., according to counterclockwise in eight neighborhood
Next profile point is looked in cell, when c cells are found, tracking direction is updated to b to c, and following first search is just
For c2Cell.Next profile cell is found successively, until returning to starting profile searching data unit.
The complete example of contour following algorithm is illustrated in Figure 5, and grey reality dot was sampled according to span in Fig. 5
Unit, black reality dot is target data unit.
There are two targets in Figure 5：One target accounts for multiple across localizer unit lattice, with the method across azimuth sample, every
One orientation span takes a scan line to be detected, white in this master drawing with regard to progress Contour extraction if there is object exists
Cell does not just have to be searched, that is, no longer need to progressively scan, so as to greatly reduce calculation amount.If the point mark of tracking
Block meets radar target constraints, it is believed that it is a target, then carries out region-growing method to obtain the radar target institute
Some cells.
Step 5：With all the points mark in region-growing method growth profile
Carry out region-growing method and obtain all data cells of the target, by all point mark include Contour searching and
The point mark of region-growing method traversal gives Plot coherence module and carries out Plot coherence, that is, performs step 6.
Step 6：The agglomeration process of point mark
The point mark block of acquisition is condensed processing according to the method for amplitude weighting.
The first step merges the point mark of region growing and the point mark of Contour tracing, forms the point of entire Targets Dots block
Mark；
Second step, using the orientation of mark, distance and amplitude information press the point mark of entire object block according to equation below
The agglomeration process of a mark is carried out according to the mode of amplitude weighting, obtains the congealing point of a target；
z(i,j)Represent the amplitude measurements for being located at (i, j) data cell in radar video data.Zd m(i, j)=1 indicates
Detect the data cell of bit flag.(iopt,jopt) represent cohesion after point where position.
Referring to Fig. 6, Fig. 6 is Contour extraction method result schematic diagram, and quadrangle asterisk will is exactly the data sheet after cohesion in Fig. 6
Member.
Third walks, and gives the target point after cohesion to subsequent radar track relating module, continues at radar data
Reason.
The present invention improves the efficiency of target detection in radar data processing, in traditional Plot coherence method, needs
One or many entire data of traversal find target, and with the raising of radar resolution, the data transfer rate of radar greatly increases, difficult
To meet the requirement handled in real time.The present invention, by the method across azimuth sample, only needs to traverse according to radar target data characteristic
Entire data small part, finds out interested target, so as to save a large amount of calculating, while is proved from theoretical with emulation
There is no the situation appearance of target missing inspection.
The technique effect of the present invention is explained again below by emulation
Embodiment 8
Radar plot condensing method based on Contour extraction with embodiment 1-7,
Emulation experiment is further described the effect of the present invention below.
Simulated conditions：
The present invention emulation be Intel (R) Core (TM) 2Duo of dominant frequency 2.4GHZ, the hardware environment of memory 4GB and
It is carried out under the software environment of MATLAB R2014a.
By two different true radar return video datas, as shown in Figure 7 and Figure 8, radar plot cohesion is carried out.Figure
7 be radar real video data (one) schematic diagram, and Fig. 8 is radar real video data (two) schematic diagram.
Handle radar video data content：
The radar video data of the respectively 8192*8192 data unit size of two radar video data.
Treatment effect is analyzed：
Using the algorithm of five kinds of Plot coherences, conventional point mark condensing method, region growing method, planet centroid method,
SDDC methods and object extraction algorithm the present invention is based on Contour tracing handle high-precision and survey radar data to maritime patrol respectively.Fig. 7
It is the original video data of the radar return under two kinds of varying environments with Fig. 8.The monitoring data of radar, are divided into 8192*
8192 data cells.In video data shown in Fig. 7, monitoring region is divided into inshore and off-lying sea bank.Video shown in Fig. 8
In data, the region of radar monitoring is divided into seashore, inshore and off-lying sea bank.Following table be to the data in above-mentioned two groups of videos into
A statistical disposition is gone, as a result, it has been found that there are the barriers such as seashore island and mountain in coastal region, so existing here a large amount of
False-alarm.And the target and false-alarm at off-lying sea are all smaller.The time that table 1 is consumed for each algorithm when handling data.
Table 1 gives the side using SDDC methods, planet centroid method, region growing method, conventional method and the present invention
Method handles the radar return video data under two groups of different scenes respectively.
1 each method aggregation efficiency contrast table of table
As seen from Table 1, if there is bigger false-alarm, SDDC algorithms just need to consume the more time.Planet barycenter
No matter the time that algorithm video data shown in Fig. 7 or video data shown in Fig. 8 are consumed all is far longer than other algorithms.
So planet centroid algorithm does not apply to radar data processing.The data of different two scenes of traditional Plot coherence algorithm process
The time phase difference of consumption is little.But the treatment effeciency of the algorithm cannot meet radar real-time and must ask.Algorithm of region growing
It is that the time more efficiently consumed is also smaller in video data shown in Fig. 7, but in the higher data of processing false-alarm
When time for consuming can sharply increase.From the point of view of the time of data processing consumption, the elapsed time of algorithm of region growing
Number depending on mark block.As can be seen from the above table the present invention is based on the algorithm of Contour extraction, than any of the above algorithm
The time of consumption is all few.It is 7.6 seconds that all point mark elapsed times are scanned in video data shown in Fig. 7.Video counts shown in Fig. 8
It it is 17.6 seconds according to all point mark elapsed times of middle scanning.It can be seen that the present invention can quickly find radar target, this hair
Bright method completes sampling, and scanning, the time of cohesion all processes consumption is with all data cell consumption of other methods single pass
Time approach.The method that the present invention is carried is to compare to reduce several orders of magnitude on processing time with other methods.It is more suitable
For engineer application.
In brief, a kind of radar plot condensing method based on Contour extraction disclosed by the invention, solves and sweeps line by line
The high problem of complexity caused by retouching data.Its realization includes, and inputs radar information；Determine that radar bearing samples span, to input
Video data presses span azimuth sample；Traversal finds the data cell for having detection bit flag across the sampled point in orientation；To traversal
What sampling determined has the data cell progress Contour extraction of detection bit flag；It is obtained with region-growing method all in search profile
Data cell；Plot coherence processing is carried out, Targets Dots in radar video data is completed and agglomerates.The present invention changes progressive scan
Operating mode, using across azimuth sample, with reference to Contour extraction, region-growing method constitute it is a kind of it is quick search target, carry out
The method of Plot coherence only need to traverse entire data small part, find out interested mark block, save big gauge
It calculates, it is theoretical to also turn out that the situation of no target missing inspection occurs with emulation, it saves Plot coherence and is consumed in radar data processing
Time, meet the requirement of real-time of high-resolution radar data processing.Data processing is agglomerated for radar plot.
Claims (5)
1. a kind of radar plot condensing method based on Contour extraction, which is characterized in that include the following steps：
(1) radar information is inputted
The parameter of (1a) input radar system requirement
According to the system requirements of radar, the running parameter of radar is inputted, input parameter includes the beam angle of radar, radiates model
It encloses, azimuth dimension required by radar target and apart from dimension size information；
(1b) inputs radar video data
It radar video data will be inputted after Radar Signal Processing；Radar video data packet contains the orientation, distance, width of target
Spend information and check bit flag information；
(2) it determines that radar bearing samples span, azimuth sample is carried out by span to inputting video data
The span of azimuth sample is determined according to across azimuth sample principle, from Initial Azimuth, it is entirely defeated to complete radar by orientation span
Enter the azimuth sample on video data, the data cell in each sampling orientation is obtained after the completion of sampling；
Across azimuth sample principle：Span across orientation is necessarily less than the orientation extension number of minimum extension target；According to radar
System requirements determine extension size of the minimum target in orientation；
(3) traversal finds the data cell for having detection bit flag across the sampled point in orientation
All data cells in present sample orientation are traversed, judge whether the sampling unit has detection bit flag, if data
Unit has detection bit flag, performs step (4), carries out Contour extraction；If data cell does not detect bit flag, continue to traverse
All residue sampling units；If all data cell traversals of present orientation are completed, continue to hold in next sampling orientation
Row step (3)；It is completed until the sampled data in all orientation traverses, radar plot agglomeration step is completed；
(4) Contour extraction is carried out to the data cell for having detection bit flag that traversal sampling determines
Using current detection bit flag unit as Contour extraction start element, recycled successively according to Contour extraction criterion search out it is next
A profile unit, for the first time back to Contour extraction start element during finding, then Contour extraction completion preserves search wheel
Wide data, obtain the search profile corresponding to current detection bit flag unit, obtain a mark block, this mark block by profile with
The data cell with detection bit flag that the obtained search profile of track and search contoured interior contain；If this mark block
Azimuth dimension and meet the condition of radar system target simultaneously apart from dimension, be judged as radar target, then perform step (5), otherwise,
Step (3) is performed to continue to traverse remaining data cell；
(5) all data cells in search profile are obtained with region-growing method
For the radar target for judging to obtain in step (4), radar target search contoured interior institute is obtained using region-growing method
Some with detection bit flag data cell simultaneously preserve the data, by all point marks of the radar target include search profile and
The data cell that growth method traverses, referred to as radar target point mark give Plot coherence module and carry out Plot coherence, that is, performs
Step (6)；
(6) agglomeration process of mark is put
The radar target point mark of acquisition is condensed processing according to the method for amplitude weighting；Radar target position after being agglomerated
Confidence ceases, and after save location information data, return to step (3) continues to traverse remaining data cell.
2. a kind of radar plot condensing method based on Contour extraction according to claim 1, it is characterised in that：Step
(2) radar determines that azimuth sample span is as follows：
2.1 determine sampling span：By the radar system parameters information of input, the maximum span i.e. A of target bearing is calculated most
Small value, that is, radar, across the maximum span of azimuth sample, sampling span should be less than the minimum value of A；If the minimum value A of Amin,
Specific formula for calculation is：
Wherein NAFor localizer unit lattice number, θ0Represent beam angle, symbolRepresent downward rounding.
3. a kind of radar plot condensing method based on Contour extraction according to claim 1, it is characterised in that：Step
(4) there are detection flag bit data to carry out Contour extraction to be as follows：
4.1 from present orientation data cell ergodic data unit one by one from down to up, until the number for finding to have detection bit flag
According to unit；
4.2 when finding to have the data cell of detection bit flag, using the data cell as Contour extraction start element, according to profile
Tracking criterion sequentially finds next profile unit, until the cell that the profile unit found is profile starting, Contour searching
Terminate；
4.3 judge whether the search profile size looked for meets in step (4) simultaneously in azimuth dimension and apart from the constraint item for tieing up formula
Part judges whether it is radar target.Step (5) is carried out if meeting, otherwise continues to execute step (3).
4. a kind of radar plot condensing method based on Contour extraction according to claim 1, it is characterised in that：Step
(6) agglomeration process of point mark is as follows：
The point mark that the 6.1 point marks for obtaining region-growing method and Contour tracing obtain merges, and forms entire radar target point
Mark block；
6.2 using point marks orientation, distance and amplitude information, according to Plot coherence formula by the point of entire radar target point mark block
Mark carries out the agglomeration process of a mark in the way of amplitude weighting, obtains a target congealing point；
6.3 give target congealing point to subsequent radar track relating module, continue radar data processing.
5. a kind of radar plot condensing method based on Contour extraction according to claim 3, it is characterised in that：Target institute
The Unit Cell number of influence is respectively A on azran is off-axismAnd Rm
M represents m-th of target that certain azimuthal channel measures, and wherein CR represents the powerful radius of radar, symbolRepresent downward
Rounding, initial azimuthTo termination azimuthBy orientation coverage area by radar beam main lobe width and target to radar it
Between distance be calculated, φmRepresent the angle of target long axis and x coordinate axis；NAFor localizer unit lattice number, NRRange cell
Lattice number, smTarget minor axis length, lmaxAnd smaxRepresent that system detects radar target maximum length and target is maximum wide respectively
Degree, lmTarget long axis length, smTarget minor axis length, θ0It represents beam main lobe width, the interested radar of radar is known by above formula
Target should meet in video data such as co-relation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711294042.0A CN108254727A (en) | 2017-12-08 | 2017-12-08 | A kind of radar plot condensing method based on Contour extraction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711294042.0A CN108254727A (en) | 2017-12-08 | 2017-12-08 | A kind of radar plot condensing method based on Contour extraction |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108254727A true CN108254727A (en) | 2018-07-06 |
Family
ID=62721037
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711294042.0A Pending CN108254727A (en) | 2017-12-08 | 2017-12-08 | A kind of radar plot condensing method based on Contour extraction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108254727A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109343021A (en) * | 2018-12-03 | 2019-02-15 | 北京遥感设备研究所 | A kind of Plot coherence method |
| CN109522114A (en) * | 2018-09-30 | 2019-03-26 | 华东计算技术研究所（中国电子科技集团公司第三十二研究所） | Radar data high-speed communication processing module of virtualization framework |
| CN109856628A (en) * | 2019-01-11 | 2019-06-07 | 中国船舶重工集团公司第七二四研究所 | A kind of target three-dimensional acceleration motion model calculation method based on scanning radar |
| CN111047622A (en) * | 2019-11-20 | 2020-04-21 | 腾讯科技（深圳）有限公司 | Method and device for matching objects in video, storage medium and electronic device |
| CN111208501A (en) * | 2019-12-30 | 2020-05-29 | 广州市番禺奥莱照明电器有限公司 | Multi-target aggregation method, device, equipment and storage medium based on radar detection |
| CN111610521A (en) * | 2020-05-27 | 2020-09-01 | 西安长远电子工程有限责任公司 | Radar terrain data processing method |
| CN112034432A (en) * | 2019-06-03 | 2020-12-04 | 华为技术有限公司 | Radar target clustering method and related device |
| CN112835001A (en) * | 2020-12-29 | 2021-05-25 | 西安电子科技大学 | Sea surface target radar trace condensation method, electronic equipment and storage medium |
| CN113589246A (en) * | 2021-05-24 | 2021-11-02 | 中国南方电网有限责任公司超高压输电公司广州局 | Target detection method and device based on radar and computer equipment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101872484A (en) * | 2010-06-30 | 2010-10-27 | 西安电子科技大学 | Method for detecting self-adaptive growth of a plurality of dim target tracks in image domain |
| CN106918807A (en) * | 2017-02-28 | 2017-07-04 | 西安电子科技大学 | A kind of Targets Dots condensing method of radar return data |
-
2017
- 2017-12-08 CN CN201711294042.0A patent/CN108254727A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101872484A (en) * | 2010-06-30 | 2010-10-27 | 西安电子科技大学 | Method for detecting self-adaptive growth of a plurality of dim target tracks in image domain |
| CN106918807A (en) * | 2017-02-28 | 2017-07-04 | 西安电子科技大学 | A kind of Targets Dots condensing method of radar return data |
Non-Patent Citations (2)
| Title |
|---|
| BO YAN ET AL.: "Improved plot fusion method for dynamic programming based track before detect algorithm", 《INTERNATIONAL JOURNAL OF ELECTRONICS AND COMMUNICATIONS (AEÜ)》 * |
| BO. YAN ET AL.: "An Efficient Plot Fusion Method for High Resolution Radar Based on Contour Tracking Algorithm", 《INTERNATIONAL JOURNAL OF ANTENNAS AND PROPAGATION》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109522114A (en) * | 2018-09-30 | 2019-03-26 | 华东计算技术研究所（中国电子科技集团公司第三十二研究所） | Radar data high-speed communication processing module of virtualization framework |
| CN109343021A (en) * | 2018-12-03 | 2019-02-15 | 北京遥感设备研究所 | A kind of Plot coherence method |
| CN109856628A (en) * | 2019-01-11 | 2019-06-07 | 中国船舶重工集团公司第七二四研究所 | A kind of target three-dimensional acceleration motion model calculation method based on scanning radar |
| CN112034432A (en) * | 2019-06-03 | 2020-12-04 | 华为技术有限公司 | Radar target clustering method and related device |
| CN111047622A (en) * | 2019-11-20 | 2020-04-21 | 腾讯科技（深圳）有限公司 | Method and device for matching objects in video, storage medium and electronic device |
| CN111047622B (en) * | 2019-11-20 | 2023-05-30 | 腾讯科技（深圳）有限公司 | Method and device for matching objects in video, storage medium and electronic device |
| CN111208501A (en) * | 2019-12-30 | 2020-05-29 | 广州市番禺奥莱照明电器有限公司 | Multi-target aggregation method, device, equipment and storage medium based on radar detection |
| CN111208501B (en) * | 2019-12-30 | 2022-06-03 | 广州市番禺奥莱照明电器有限公司 | Multi-target aggregation method, device, equipment and storage medium based on radar detection |
| CN111610521A (en) * | 2020-05-27 | 2020-09-01 | 西安长远电子工程有限责任公司 | Radar terrain data processing method |
| CN112835001A (en) * | 2020-12-29 | 2021-05-25 | 西安电子科技大学 | Sea surface target radar trace condensation method, electronic equipment and storage medium |
| CN112835001B (en) * | 2020-12-29 | 2023-08-08 | 西安电子科技大学 | Sea surface target radar trace condensation method, electronic equipment and storage medium |
| CN113589246A (en) * | 2021-05-24 | 2021-11-02 | 中国南方电网有限责任公司超高压输电公司广州局 | Target detection method and device based on radar and computer equipment |
| CN113589246B (en) * | 2021-05-24 | 2023-09-01 | 中国南方电网有限责任公司超高压输电公司广州局海口分局 | Target detection method and device based on radar and computer equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108254727A (en) | A kind of radar plot condensing method based on Contour extraction | |
| CN103424112B (en) | A kind of motion carrier vision navigation method auxiliary based on laser plane | |
| CN111368769B (en) | Ship multi-target detection method based on improved anchor point frame generation model | |
| CN112395987B (en) | SAR image target detection method based on unsupervised domain adaptive CNN | |
| CN103353988B (en) | Allos SAR scene Feature Correspondence Algorithm performance estimating method | |
| CN109871823B (en) | Satellite image ship detection method combining rotating frame and context information | |
| CN103295232B (en) | Based on the SAR image registration method in straight line and region | |
| CN107424177A (en) | Positioning amendment long-range track algorithm based on serial correlation wave filter | |
| CN110132284B (en) | Global positioning method based on depth information | |
| CN113643284B (en) | Polarized synthetic aperture radar image ship detection method based on convolutional neural network | |
| CN103839257B (en) | Method for detecting changes of SAR images of generalized Gaussian K&I | |
| CN111476159A (en) | Method and device for training and detecting detection model based on double-angle regression | |
| CN106156758B (en) | A kind of tidal saltmarsh method in SAR seashore image | |
| CN110516560B (en) | Optical remote sensing image target detection method based on FPGA heterogeneous deep learning | |
| CN110674674A (en) | Rotary target detection method based on YOLO V3 | |
| CN102360503B (en) | SAR (Specific Absorption Rate) image change detection method based on space approach degree and pixel similarity | |
| CN104331886A (en) | Port region ship and warship detection method based on high resolution SAR image | |
| CN109308713A (en) | A kind of improvement core correlation filtering Method for Underwater Target Tracking based on Forward-looking Sonar | |
| CN112487912A (en) | Arbitrary-direction ship detection method based on improved YOLOv3 | |
| CN105303566B (en) | A kind of SAR image azimuth of target method of estimation cut based on objective contour | |
| CN106407975A (en) | Multi-dimensional layered object detection method based on space-spectrum constraint | |
| CN116310837B (en) | SAR ship target rotation detection method and system | |
| CN107170004A (en) | To the image matching method of matching matrix in a kind of unmanned vehicle monocular vision positioning | |
| CN107657272A (en) | A kind of remote sensing images method for detecting targets at sea | |
| CN104537690A (en) | Motor point target detection method based on maximum-time index union |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180706 |