CN103885053A - Radar data moving-target detection processing method based on tracking filter - Google Patents

Abstract

The invention discloses a radar data moving-target detection processing method based on a tracking filter. The method comprises the following steps: (1) conducting signal preprocessing on a pulse pressure signal; (2) filtering the signal obtained in the step (1), wherein filtration is divided into a detection state and a tracking state, in the detection state, M MTD filters are adopted to output target signals of M Doppler frequency channels, and in the tracking state, a tracking filter is adopted for filtering specified Doppler frequency signals; (3) conducting clutter map processing on output data in the step (2); (4) conducting constant false-alarm processing on a result of the step (3); (5) analyzing and calculating the clutter property, classifying clusters, adopting detecting thresholds in various corresponding clutter areas, and conducting target detection; (6) synchronizing detection signals of the same distance element of various channels to obtain a target detection result of a distance unit.

Description

A kind of radar data moving-target Check processing method based on tracking filter

Technical field

The present invention relates to a kind of radar data disposal route, particularly a kind of radar data moving-target based on tracking filter detects (MTD, Moving Target Detect, MTD) disposal route.

Background technology

It is the important technology that in Radar Signal Processing, clutter suppresses that moving-target detects (Moving Target Detect, MTD).Clutter in Radar Signal Processing comprises ground clutter, sea clutter, meteorological clutter etc.MTD process be that radar echo signal is carried out to coherent accumulation matched filtering, be characterized in good inhibition move clutter ability, raising target signal to noise ratio (S/N ratio).MTD processes the scope that adopts arrowband Doppler filter group to cover whole radar repetition frequency, because the Doppler frequency of clutter and target is different, they appear in different narrow band filters, thereby filtering clutter, stick signal easily, thereby the ability of raising detections of radar echo signal.So, the analysis to MTD technology and research, not only significant in theory, in practicality, also there is active demand.

Moving-target detects (MTD) radar and has used various wave filters, and filtering fixed clutter and take out the echo of motion clutter can improve the ability that detects moving target under clutter background, the antijamming capability of raising radar.By by the bank of filters of multiple bandpass filter composition, radar return being carried out to filtering processing, then output detects to find target to bank of filters.

How to improve MTD performance, always be the hot issue of radar data processing.By to different distance sectionally weighting, and to the in addition a series of optimal control of different weights coefficient, can reduce the leap loss between filter channel, improve the Weak target detectability of radar under strong clutter background.

There is following shortcoming in existing MTD technology: (1) is because radar data is subject to the constraint of CPI, in a wave beam, only have a limited number of CPI data, limit the filter channel number of MTD, limited filter channel, can not cover all frequency domains completely, between two adjacent filters, certainly exist and cross over and intersect.(2) traditional MTD processes, and the center of bank of filters is fixed, and the Doppler frequency of moving target may drop between two adjacent filters Anywhere, and therefore the loss of signal can occur.MTD performance is directly subject to the impact of crossing over loss and intersecting loss, the inefficacy that can cause clutter to suppress in the situation that loss is serious.

Summary of the invention

Goal of the invention: technical matters to be solved by this invention is for the deficiencies in the prior art, provides a kind of radar data moving-target Check processing based on tracking filter.

The invention discloses a kind of radar data moving-target Check processing method based on tracking filter, comprise the following steps:

(1) to signal process Signal Pretreatment after pulse pressure, remove interference and clutter in signal.

(2) signal step 1 being obtained carries out filtering, be divided into and detect and follow the trail of two states: detected state adopts the echo signal of MTD wave filter output M road, M road Doppler frequency passage, and tracking state adopts tracking filter to carry out filtering to specific Doppler frequency signal;

(3) the output data to step (2), carry out clutter map processing;

(4) result to step (3), carries out CFAR processing;

(5) distribute according to clutter, noise performance is carried out to analytical calculation, to clutter region class, corresponding various clutters district adopts corresponding detection threshold, carries out target detection; Target detection data detect through sliding window again, according to area type, sliding window detection threshold are set respectively;

(6) detection signal of each passage same distance unit is synthesized, obtain the target detection result of range unit.

In step of the present invention (2), tracking filter carries out filtering to specific Doppler frequency signal and comprises, by the signal that aims at the mark the maximum zero point of tracking filter, and the Doppler frequency center of estimating target signal.

In the present invention: if tracking filter in tracking state, utilizes the velocity variations estimating Doppler frequency center of target detection when at the Doppler frequency center of tracking filter estimating target; If tracking filter in detected state, utilizes classical spectrum estimate algorithm period map method, or centroid algorithm estimating Doppler frequency center.

In the present invention, in step (2), under tracking state, adopt following formula estimating Doppler frequency center:

$f_c=\frac{{2V}_r}{\mathrm{\λ}},$

Wherein, f _cdoppler frequency, V _rbe target radial speed, λ is the operation wavelength of radar.

In step (2), under detected state, centroid algorithm estimating Doppler frequency center comprises the following steps:

(a) the Signal for Pulse p obtaining for step (1) _m(n):

p _m(n),n＝0,1,…,N-1,m＝m ₁,m ₁+1,…,m ₂，

M participates in the range unit that power spectrum signal is estimated, m ₁apart from start element, m ₂that distance stops unit, the length that N is Signal for Pulse, n represents corresponding Signal for Pulse;

(b) rated output spectrum S _m(n):

S _m(n)＝|FFT[s _m(n)]| ²；

FFT represents signal to carry out Fourier transform (Fast Fourier Transform)

(c) calculate the power spectrum mean value S (n) of all range units:

$S\left(n\right)=\frac{1}{m_2-{\mathrm{<figure-callout\; id="1"\; label="m"\; filenames="HDA0000483379010000012.png,HDA0000483379010000021.png"\; state="\{\{state\}\}">m</figure-callout>}}_1+1}\underset{m_1}{\overset{m_2}{\mathrm{\&Sigma;}}}{S}_m\left(n\right);$

(d) calculate phase corresponding to barycenter vector ₀:

${\mathrm{\&phi;}}_0=\mathrm{phase}\left[\underset{m_1}{\overset{m_2}{\mathrm{\&Sigma;}}}\left(S_m\left(n\right)\exp (j\&CenterDot;2\mathrm{\&pi;n}/N)\right)\right],$

φ ₀be the phase place that barycenter vector is corresponding, N is pulse pressure data length, and Phase asks phase place to complex signal, and j is imaginary part unit, and exp represents to ask exponent arithmetic;

(e) estimated frequency center f _c:

$f_c={\mathrm{\&phi;}}_0\&CenterDot;\frac{f_p}{2\mathrm{\&pi;}},$

Wherein, f _pbeing radar pulse repetition frequency, is the radar system parameter of collection signal.

In step (2), under detected state, utilize the period map method estimating Doppler frequency center of classical spectrum estimate: by signal is carried out to FFT processing, the amplitude of frequency spectrum is carried out square to estimated signal frequency center, i.e. Doppler frequency.

Signal is first weighted to auto-correlation and carries out again FFT processing.

The present invention adopts the MTD of 4 road fixed filters passages and 1 road tracking filter passage to process.During radar MTD signal based on tracking filter is processed, tracking filter adopts data processing and signal processing to combine, and by the center of accurate control wave filter, carries out optimum matching filtering, eliminates doppler filtering mismatch, obtains maximum coherence processing gain.The design of tracking filter, comprehensively adopts the Doppler frequency center of various ways estimating target: the velocity variations information of utilizing the target detection of sensor under tracking state; Under detected state, utilize the estimating Doppler frequency centers such as classical spectrum estimate algorithm, period map method, centroid algorithm; In the situation that appointed condition is allowed, expand filter channel, make the wave filter Doppler center that aims at the mark, reach optimal filtering effect.

Advantage of the present invention is embodied in: (1) for existing and cross over loss between wave filter, designed can adaptive tracing target Doppler frequency center tracking filter, fill up the blank between different passages.(2) carrying out CFAR detection and sliding window detects the crossover loss existing for each filter channel data, reducing the impact of false alarm rate, designed can adaptive tracing target Doppler frequency center tracking filter.(3) can be used in design and the optimization of the bank of filters of any point number, and design of filter mode is flexible: can adopt data processing and signal processing to combine, utilize radar data to process the target velocity of feedback, calculate Doppler frequency center, by the center of accurate control wave filter, carry out optimum matching filtering, eliminate doppler filtering mismatch, obtain maximum coherence processing gain.

Brief description of the drawings

Below in conjunction with the drawings and specific embodiments, the present invention is done further and illustrated, above-mentioned advantage of the present invention will become apparent.

Fig. 1 is the process flow diagram that the present invention is based on the MTD radar data processing of tracking filter.

Fig. 2 is wave filter amplitude-frequency response schematic diagram.

Fig. 3 is the overlapping schematic diagram of wave filter.

Fig. 4 follows the trail of optimum results schematic diagram.

Embodiment

As shown in Figure 1, the concrete implementation step of the present invention is as follows:

1. after pair pulse pressure, signal, through Signal Pretreatment, is removed interference and clutter in signal.Comprise the narrow pulse interference in time domain, strong signal secondary lobe under secondary lobe compacting disturbs, monocycle asynchronous interference etc., respectively according to interference characteristic, carry out relevant counter-jamming processing, disposal route referring to anti interference pertinent literature (for example: Wan Xianrong etc., certain intelligence radar interference asynchronism in the same frequency inhibition method research, electronic letters, vol, 2012 (3)).

2. adopt multichannel bank of filters to carry out MTD processing to data, filtering is divided into detection and follows the trail of two states: detected state adopts the echo signal of common MTD bank of filters output M road, M road Doppler frequency passage, and tracking state only adopts tracking filter to carry out filtering to specific Doppler frequency signal.

General filter design: wave filter amplitude-frequency response schematic diagram as shown in Figure 2, it is respectively the wave filter amplitude-frequency response of even weighted sum Chebychev weighting, when main lobe, the clutter that falls into filter channel at signal falls into filter channel secondary lobe, can obtain good clutter inhibition.In actual environment, the speed of echo signal is non-constant, and signal Doppler shift also often changes, and by regulating adaptively and changing wave filter main lobe position, can improve MTD filtering performance.This method falls into by making wave filter form zero in the frequency processing of clutter adaptively, clutter reduction effectively.Because the conventional adaptive M TD wave filter obtaining by the design of this method has higher minor level, interacting between the easy like this target that causes different wave filters, can bring false-alarm; In addition, the clutter that the clutter that enters wave filter secondary lobe also can reduce (comprising residue and the meteorological clutter of land clutter) wave filter improves performance.So, need to be weighted to reduce to conventional adaptive M TD wave filter the minor level of each wave filter.

Tracking filter design: traditional MTD processes, and the center of bank of filters is fixed, the Doppler frequency of moving target may drop between two adjacent filters Anywhere, and therefore the loss of signal can occur.In order to reduce this loss, the present invention has designed the tracking filter of position changeable.Tracking filter carries out filtering to specific Doppler frequency signal, obtains optimum matching filtering performance, can be used in design and the optimization of the bank of filters of any point number.During real data is processed, the signal that aims at the mark the maximum zero point of tracking filter, to obtain maximum filtering performance.The problem that signal is aimed at, is summed up as the doppler centroid that solves signal, or the speed of target.Tracking filter design is flexible, under actual conditions, can adopt flexibly diverse ways, by the center of accurate control wave filter, carries out optimum matching filtering, eliminates doppler filtering mismatch, obtains maximum coherence processing gain.

Detect in (MTD) at moving-target, due to the motion of Texas tower, the reason such as the skew of beam position and pulse repetition rate are limited, clutter power spectrum may be distributed in any field between spectrum region.For effective clutter reduction, need to effectively estimate clutter spectrum parameter (clutter spectrum center and clutter spectrum width).Frequency center method of estimation:

(1) can adopt data processing and signal processing to combine, utilize radar data to process the target velocity of feedback, calculate Doppler frequency center.The poor f of Doppler frequency _cwith radial velocity V _rbetween relation:

$f_c=\frac{{2V}_r}{\mathrm{\&lambda;}}$

Wherein, λ is the operation wavelength of radar.

(2) classical spectrum estimate method---periodogram analysis: by signal is carried out to FFT processing, the amplitude of frequency spectrum is carried out square to estimated signal frequency center, namely Doppler frequency; Or signal is first weighted to auto-correlation, and then FFT processes.

Autocorrelation function is:

${\mathrm{\&phi;}}_{\mathrm{xx}}\left(m\right)=\frac{1}{N}\underset{n=-\&infin;}{\overset{\&infin;}{\mathrm{\&Sigma;}}}y\left(n\right)y(n+m)=y\left(m\right)*y(-m),$

The Fourier transform of y (m) is:

$\underset{m=-\&infin;}{\overset{\&infin;}{\mathrm{\&Sigma;}}}y\left(m\right)e^{-\mathrm{jwm}}=\underset{m=-\&infin;}{\overset{\&infin;}{\mathrm{\&Sigma;}}}d\left(m\right)x\left(m\right)e^{\mathrm{jwm}}=X_N\left(e^{\mathrm{jw}}\right),$

X _n(e ^jw) be the Fourier transform of finite length sequence x (n).In the time that x (n) is real sequence, y (n) is also real sequence, and (Fourier transform m) is y power Spectral Estimation be related function Fourier transform:

$P_{\mathrm{xx}}\left(w\right)=\frac{1}{N}{X}_N\left(e^{\mathrm{jw}}\right)X_N^{*}\left(e^{\mathrm{jw}}\right)=\frac{1}{N}{\left|X_N\left(e^{\mathrm{jw}}\right)\right|}^2,$

In above formula, the corresponding frequency of data maximum point is exactly the centre frequency of clutter spectrum.

(3) centroid algorithm: say from the statistical significance, to Gauss type clutter power spectrum, the spectral line amplitude that approaches spectrum center should be larger.For the large spectral line of spectral line amplitude, can be weighted the corresponding frequency location of corresponding spectral line by its amplitude completely, to give prominence to the frequency center real by estimated signal.Be similar in form the barycenter of the object of asking an even density, be referred to as " centroid method ".

(a) signal data that input data length is N:

s _m(n),n＝0,1,…,N-1,m＝m ₁,m ₁+1,…,m ₂，

M participates in the range unit that power spectrum signal is estimated;

(b) rated output spectrum:

S _m(n)＝|FFT[s _m(n)]| ²

(c) power spectrum of multiple range units is averaging:

$S\left(n\right)=\frac{1}{m_2-{\mathrm{<figure-callout\; id="1"\; label="m"\; filenames="HDA0000483379010000012.png,HDA0000483379010000021.png"\; state="\{\{state\}\}">m</figure-callout>}}_1+1}\underset{m_1}{\overset{m_2}{\mathrm{\&Sigma;}}}{S}_m\left(n\right)$

(d) phase place corresponding to barycenter vector:

${\mathrm{\&phi;}}_0=\mathrm{phase}\left[\underset{m_1}{\overset{m_2}{\mathrm{\&Sigma;}}}\left(S_m\left(n\right)\exp (j\&CenterDot;2\mathrm{\&pi;n}/N)\right)\right],$

(e) estimated frequency center:

$f_c=={\mathrm{\&phi;}}_0\&CenterDot;\frac{f_p}{2\mathrm{\&pi;}},$

Wherein, f _pit is the target frequency cycle.

In the situation that appointed condition is allowed, expand filter channel, increase group number and the coverage of wave filter, reduce the intersection loss between filter channel, by making the wave filter Doppler center that aims at the mark, reach optimal filtering effect.Consider the bank of filters of 4 passages.Under even weighting scheme, the maximal filter amplitude response that can obtain is 6.021dB (A point), amplitude response 2.323dB in point of crossing between filter channel (C point).Appear in order to reduce echo signal frequency shift (FS) the match penalties that C point causes, be optimized at C point, increase in addition 4 passages, as shown in Figure 3, form the bank of filters of 8 passages, avoided C point to cross over loss.Under this situation, the point of crossing amplitude response 5.164dB of bank of filters (B point).Increase extra filter channel by the frequency place that C point and B are ordered, as shown in Figure 4, can improve and cross over loss, obtain the optimum matching effect of echo signal.

In practice, should comprehensively select the Doppler frequency center of suitable mode estimating target: the velocity variations information of utilizing the target detection of sensor under tracking state; Under detected state, utilize the estimating Doppler frequency centers such as classical spectrum estimate algorithm, period map method, centroid algorithm; In the situation that appointed condition is allowed, expand filter channel, make the wave filter Doppler center that aims at the mark, reach optimal filtering effect.

3. wave filter output data, have mixed a large amount of clutters.Except the clutter of the generations such as isolated buildings can be regarded fixing some clutter as, most of clutters all belong to Distribution Clutter, and have internal motion, and the frequency spectrum of clutter is wider.Clutter spectrum center, at ground clutter or the sea clutter of zero frequency, can adopt moving-target to show that wave filter suppresses clutter, and through clutter map, processing can suppress the impact of ground clutter or sea clutter; For motion clutter, can adopt self-adaptation moving-target to show that wave filter carries out clutter inhibition.In addition, according to different clutter types, the clutter map parameter in dissimilar region can be set according to real work situation, self-adaptation meets different clutters and suppresses needs.

4. the result of clutter map output, through CFAR processing, improves the signal discovery ability under clutter background.In radar system, false-alarm probability and detection probability are one of leading indicators, conventionally require false-alarm probability to be constrained under the condition of certain value, make detection probability maximum.In radar system, too much false target may make follow-up data handling machine saturated, and false-alarm probability is too small, and detection probability will decrease, more on the detectability impact of little target echo signal especially.Adopting CFAR detection, is exactly the in the situation that of noise and noise jamming Strength Changes, and through constant false alarm rate, processing makes false alarm rate keep constant to signal.Actual data handling procedure can arrange according to real work situation the working method of the false alarm rate of the slow thresholding in dissimilar region, artificial threshold value and fast thresholding, determines adaptive targets detection threshold.In addition, adjust on demand false alarm rate, can improve the signal discovery ability of clutter edge.

5. pair CFAR is exported data, carries out the target detection of self-adapting detecting thresholding.Automatic threshold under noise and clutter environment detects, and key is to form with the automatic threshold of noise clutter environment Auto-matching to detect level.The formation that automatic threshold detects level is made up of noise level estimation and coefficient two parts.Because system noise variation is slower, simultaneously in order to eliminate the impact of noise, the sample that noise average level is estimated adopts the sampling in self noise district conventionally.For example, by Clutter Classification, adopting slow thresholding to carry out target detection without clutter region, detection threshold is got a steady state value that meets false alarm rate just can ensure target detection ability, is conventionally arranged on the left and right higher than noise level 6dB; At strong clutter, district adopts fast thresholding to carry out target detection, detection threshold is the dynamic threshold that follower's clutter fluctuating rule changes, and this thresholding can have certain accumulated delays than clutter map conventionally, delay number relevant with clutter map renewal coefficient, conventionally get 1/8 renewal, 7/8 criterion retaining.

Target detection data after self-adapting detecting thresholding detect through sliding window again, improve clutter and suppress ability.

6. the detection signal of each passage same distance unit is synthesized, obtain target detection net result.

As can be seen here, data processing of the present invention is processed FEEDBACK CONTROL to signal, and synchronous signal is processed the forward direction of data processing is supported.Therefore, different from traditional approach, the signal processing module of integrated terminal can, under the control of data processing, be realized adaptive processing according to task and disturbed condition, to improve the ability of discovery of small-signal or to reduce false alarm rate.

The invention provides a kind of radar data moving-target Check processing based on tracking filter; method and the approach of this technical scheme of specific implementation are a lot; the above is only the preferred embodiment of the present invention; should be understood that; for those skilled in the art; under the premise without departing from the principles of the invention, can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.In the present embodiment not clear and definite each ingredient all available prior art realized.

Claims (7)

1. the radar data moving-target Check processing method based on tracking filter, is characterized in that, comprises the following steps:

(1), to Signal for Pulse pre-service, remove interference and clutter in signal;

(2) signal step 1 being obtained carries out filtering, be divided into and detect and follow the trail of two states: detected state adopts the echo signal of MTD wave filter output M road, M road Doppler frequency passage, and tracking state adopts tracking filter to carry out filtering to specific Doppler frequency signal;

(3) the output data to step (2), carry out clutter map processing;

(4) result to step (3), carries out CFAR processing;

(5) distribute according to clutter, noise performance is carried out to analytical calculation, to clutter region class, corresponding various clutters district adopts corresponding detection threshold, carries out target detection; Target detection data detect through sliding window again, according to area type, sliding window detection threshold are set respectively;

(6) detection signal of each passage same distance unit is synthesized, obtain the target detection result of range unit.

2. the radar data moving-target Check processing based on tracking filter according to claim 1, it is characterized in that, in step (2), tracking filter carries out filtering to specific Doppler frequency signal and comprises, by the signal that aims at the mark the maximum zero point of tracking filter, and the Doppler frequency center of estimating target signal.

3. the radar data moving-target Check processing based on tracking filter according to claim 2, it is characterized in that,: if tracking filter in tracking state, utilizes the velocity variations estimating Doppler frequency center of target detection when at the Doppler frequency center of tracking filter estimating target; If tracking filter in detected state, utilizes classical spectrum estimate algorithm period map method, or centroid algorithm estimating Doppler frequency center.

4. the radar data moving-target Check processing based on tracking filter according to claim 3, is characterized in that: in step (2), adopt following formula estimating Doppler frequency center under tracking state:

$f_c=\frac{{2V}_r}{\mathrm{\&lambda;}},$

Wherein, f _cdoppler frequency, V _rbe target radial speed, λ is the operation wavelength of radar.

5. the radar data moving-target Check processing based on tracking filter according to claim 3, is characterized in that: in step (2), under detected state, centroid algorithm estimating Doppler frequency center comprises the following steps:

(a) the Signal for Pulse p obtaining for step (1) _m(n):

p _m(n),n=0,1,…,N-1,m=m ₁,m ₁+1,…,m ₂，

M participates in the range unit that power spectrum signal is estimated, m ₁apart from start element, m ₂that distance stops unit, the length that N is Signal for Pulse, n represents corresponding Signal for Pulse;

(b) rated output spectrum S _m(n):

S _m(n)=|FFT[s _m(n)]| ²；

(c) calculate the power spectrum mean value S (n) of all range units:

$S\left(n\right)=\frac{1}{m_2-m_1+1}\underset{m_1}{\overset{m_2}{\mathrm{\&Sigma;}}}{S}_m\left(n\right);$

(d) calculate phase corresponding to barycenter vector ₀:

${\mathrm{\&phi;}}_0=\mathrm{phase}\left[\underset{m_1}{\overset{m_2}{\mathrm{\&Sigma;}}}\left(S_m\left(n\right)\exp (j\&CenterDot;2\mathrm{\&pi;n}/N)\right)\right],$

φ ₀be the phase place that barycenter vector is corresponding, N is pulse pressure data length, and Phase asks phase place to complex signal, and j is imaginary part unit, and exp represents to ask exponent arithmetic;

(e) estimated frequency center f _c:

$f_c={\mathrm{\&phi;}}_0\&CenterDot;\frac{f_p}{2\mathrm{\&pi;}},$

Wherein, f _pbeing radar pulse repetition frequency, is the radar system parameter of collection signal.

6. the radar data moving-target Check processing based on tracking filter according to claim 3, it is characterized in that: in step (2), under detected state, utilize the period map method estimating Doppler frequency center of classical spectrum estimate: by signal is carried out to FFT processing, the amplitude of frequency spectrum is carried out square, estimated signal frequency center, i.e. Doppler frequency.

7. the radar data moving-target Check processing based on tracking filter according to claim 6, is characterized in that: signal is first weighted to auto-correlation and carries out FFT processing again.

Images