CN104267387A - Target detection method of carrier-free ultra-wide band radar - Google Patents

Abstract

The invention relates to technologies of radar target detection, and discloses a target detection method of carrier-free ultra-wide band radar. According to the technical scheme, the target detection method mainly includes the following steps that a, pulse cancellation processing is carried out on received K-time scanning echo signals; b, sliding window processing is carried out on the signals processed through the step a to accumulate energy; c, the results obtained through the step b are compared with a first threshold Th1 to obtain target data points; d, whether the P target data points exceeding the first threshold Th1 appear continuously or not is judged, if yes, the first data point is reserved, then the follow-up data points are judged, and in the same way other data points are obtained; e, the data points obtained through the step d are combined and processed in a Hough transformation mode; f, the result obtained by processing in the step e is compared with a second threshold Th2, and if the result is larger than the second threshold Th2, a target is detected; otherwise, the target does not exist, wherein the Th2 is set by radar parameters. The amount of calculation of the Hough transformation detection method is effectively reduced.

Description

A kind of object detection method of carrier-free UWB radar

Technical field

The present invention relates to Radar Targets'Detection technology, particularly transmit as the pulse of nanosecond carrierfree and detection technique under target echo form unknown situation.

Background technology

The pulse of carrier-free UWB radar emission nanosecond carrierfree, has the advantage that the normal radars such as High Range Resolution, low probability of intercept, anti-interference, strong penetrating power are incomparable, especially superior anti-stealth capability, makes it the focus becoming research at present.The target echo of carrier-free UWB radar is from each scattering center of whole target, its echo shaping is the convolution responded with target impact that transmits, show as the one-dimensional range profile of target scattering point, and no longer can be regarded as conventional narrow-band radar " duplicate " that transmits.In most cases, because target impact response is unknown, thus conventional in detections of radar matched filtering and related detecting method are no longer applicable to carrier-free UWB radar.

In recent years, along with the development of ultra-wideband radar technology, occurred the plurality of target detection method such as correlation method, Wavelet-denoising Method, Hough transform in short-term, but all there is respective limitation in these detection methods.Correlation method is by the laggard line correlation process of the simple segmentation of signal data in short-term, but due to data segment shorter and by the restriction of Signal-to-Noise, Detection results when signal to noise ratio (S/N ratio) is less, cannot be ensured; Wavelet-denoising Method needs the prior imformation knowing actual signal.And above-mentioned two kinds of detection methods cannot be applied to and transmit as the pulse of nanosecond carrierfree and detection under target echo form unknown situation.Hough transform method is difficult to solve the contradiction between algorithm operation quantity and detection perform, is restricted in actual applications.

Summary of the invention

Technical matters to be solved by this invention, be exactly for existing Hough transform detection method, the shortcoming bringing algorithm operation quantity to increase because improving detection perform, the Hough transform detection method proposing a kind of improvement realizes the target detection of carrier-free UWB radar, while reducing target detection performance loss as far as possible, reduce operand.

The present invention solve the technical problem, the technical scheme adopted is, a kind of object detection method of carrier-free UWB radar, by to objective emission carrierfree pulse signal s (t), and process detection target information is carried out to the scan echo signal received, it is characterized in that, comprise the following steps:

A, K the scan echo signal received is carried out to pulse by following formula and offsets process:

$s_k^d\left(n\right)=s_{k+1}\left(n\right)-s_k\left(n\right)$

Wherein, K is integer, is set by radar parameter; s _kn () is the measured value of the n-th range unit of a kth echoed signal; for the result after once offseting; N=1,2 ..., N _r; N _rfor distance samples is counted, set by radar parameter;

B, by slide window processing, energy accumulation is carried out to the signal after step a process:

$E_k\left(n\right)=\underset{n=1}{\overset{W}{\mathrm{\Σ}}}{\left[s_k^d\left(n\right)\right]}^2$

Wherein, E _kn () is the signal energy of accumulation; W is sliding window length, is set by radar parameter;

C, by E _k(n) and the first thresholding Th ₁relatively, number of targets strong point is obtained; Wherein, Th ₁set by radar parameter;

D, judgement are more than the first thresholding Th ₁number of targets strong point whether there is P continuously, if then retain first data point, then judge follow-up data point, by that analogy, obtain other data points; Wherein, P is positive integer, is set by radar parameter;

E, the data point obtained combined and carry out Hough transform process through steps d;

F, by the result after step e process and the second thresholding Th ₂compare, if be greater than the second thresholding Th ₂, then target is detected; Otherwise then target does not exist; Wherein, Th ₂set by radar parameter.

Concrete, described pulse signal s (t) is Gauss pulse signal, has following expression:

$s\left(t\right)={\mathrm{\αe}}^{{-a}^2{t}^2}$

Wherein, a be the pulsewidth factor and t is the duration of pulse; α is pulse height; A, T, α are set by radar parameter.

Concrete, in described step e, group of data points is combined into D _a:

$D_a=\left[\begin{array}{cccccccccc}1& 1& ...& 1& ...& k& ...& k& ...& K-1\\ n_{l_1}& n_{l_2}& ...& n_{l_j}& ...& n_{k_1}& ...& n_{k_j}& ...& n_{{(K-1)}_j}\end{array}\right]$

Wherein, k represents kth time scanning, k ∈ [1, K]; represent the individual range unit more than the data point place of the first thresholding of jth in kth time scanning.

Concrete, in described step e, the concrete grammar of Hough transform is:

ρ＝xcosθ+ysinθ θ∈[0,π]

Wherein, ρ, θ are data point polar coordinates, represent Distance geometry angle parameter respectively; X, y are the rectangular coordinate of data point.

The invention has the beneficial effects as follows, under the condition that detection perform does not obviously decline, effectively reduce the operand of Hough transform detection method, to a certain degree overcome contradiction between traditional Hough transform detection method operand and detection perform, be particularly suitable for the application of some specific occasions.

Accompanying drawing explanation

Fig. 1 is object module schematic diagram;

Fig. 2 is that the data point of embodiment gathers schematic diagram;

Fig. 3 is the detection perform curve of the present invention and traditional Hough transform method.

Embodiment

Technical scheme of the present invention is described in detail below in conjunction with drawings and Examples.It should be noted that, the parameter in embodiment does not affect generality of the present invention.

Embodiment

This routine carrier-free UWB radar system, transmits as carrierfree Gauss pulse signal s (t), has following form:

$s\left(t\right)={\mathrm{\αe}}^{-a^2{t}^2}$

Wherein a be the pulsewidth factor and t is the duration of pulse, and α is amplitude.A, T, α are set by radar parameter.

According to Transient Electromagnetic field theory, surface uses the target of absorbing material or profile more complicated, and its each scattering center can regard a dispersion passage as.Its target impulse response is h (t):

$h\left(t\right)=\underset{m=1}{\overset{M}{\mathrm{\Σ}}}{A}_m{e}^{j{\mathrm{\φ}}_m}\exp [-a_m^2{(t-{\mathrm{\τ}}_m)}^2]$

Wherein, a _mbe the pulsewidth factor of m scattering center response function, identical with the definition of a, M represents the number of target scattering center, A _mand τ _mrepresent amplitude corresponding to m scattering center and this scattering center and the time delay corresponding to radar radial distance respectively, φ _mfor phase factor.

So, the target echo of carrier-free UWB radar is transmit and the convolution of target impulse response, and corresponding echoed signal is r (t):

r(t)＝s(t)*h(t)

Wherein, * represents convolution.

According to above-mentioned signal model, use matlab (a kind of ALGOL) that the echoed signal of object module as shown in Figure 1 can be produced.The design parameter of emulation is as follows:

Radar parameter is arranged: duration of pulse T=1ns; Pulse height α=1; Sample frequency f _s=5GHz; Pulse repetition rate PRF=550Hz; Scan echo number of times K=20; Distance samples is counted N _r=600; The sliding window length W=15 of sliding window accumulation; The false-alarm probability that first thresholding is corresponding is 0.1; The false-alarm probability (total false-alarm probability) that second thresholding is corresponding is 10 ^-4; P=4; The two-dimensional grid size of the Distance geometry angle parameter spatial division that Hough transform adopts is respectively: M _ρ=800, M _θ=800; The signal to noise ratio scope of the target echo received is-15dB ~ 0dB.

Target component is arranged: as shown in Figure 1, from left to right comprise 9 scattering centers, the amplitude of each scattering center is identical, and phase place is uniformly distributed, and corresponding time delay is respectively 1ns, 6ns, 6ns, 6ns, 6ns, 7ns, 7ns, 10ns, 10ns.Target is flown with the radial velocity of 100m/s.

Concrete testing process is as follows:

A. to 20 the scan echo signal s received _kn () is carried out pulse and is offseted process, with the interference of clutter reduction to target echo.Signal after offseting process is

$s_k^d\left(n\right)=s_{k+1}\left(n\right)-s_k\left(n\right)$

Wherein, k=1,2 ..., 19, n=1,2 ..., 600.

B. the data after each pulse being offseted energy accumulation is carried out, to improve the letter miscellaneous noise ratio of target echo by slide window processing.Signal energy after sliding window accumulation is E _k(n):

$E_k\left(n\right)=\underset{n=1}{\overset{W}{\mathrm{\Σ}}}{\left[s_k^d\left(n\right)\right]}^2$

Wherein W=15.

C. by the signal E after slide window processing _k(n) and the first thresholding Th ₁compare, to reject the less data point of target amplitude.Wherein, the first thresholding Th ₁determined by following formula:

$P_{f_1}=P(E_k\left(n\right)>{\mathrm{Th}}_1)$

In formula being the false-alarm probability that the first thresholding is corresponding, is 0.1 here.If kth time signal E _kn data point that () exceedes thresholding has M _kindividual, then the target data point set by being formed after the first Threshold detection is combined into as data point in Fig. 21 ~ 4,6,8,11 ~ 18 ...

D. the number of targets strong point more than the first thresholding is judged whether occur 4 continuously, if then retain first data point, this step of circular treatment is until k=K-1.Obtain data point A, N, R ..., corresponding data point 1,11,15 respectively ..., as shown in Figure 2.Such process effectively can lower the operand that follow-up Hough transform detects.Reduce data volume process by this step, obtain new target data point set and be combined into as A, N, R in Fig. 2 ... data point.

E. to the set of new target data point combine, obtain the set of data points Da under rectangular coordinate system:

$D_a=\left[\begin{array}{cccccccccc}1& 1& ...& 1& ...& k& ...& k& ...& K-1\\ n_{l_1}& n_{l_2}& ...& n_{l_j}& ...& n_{k_1}& ...& n_{k_j}& ...& n_{{(K-1)}_j}\end{array}\right]$

Then, to the set of data points D under rectangular coordinate system _acarry out Hough transform process:

E1) divide Distance geometry angle two-dimensional grid: (ρ, θ) plane is divided into two-dimensional grid, the meshes number divided along ρ axle is 800, and the meshes number divided along θ axle is 800.So, the length of each grid of ρ axle is wherein (x _max, y _max) for target is in the maximum coordinates of x-axis and y-axis; The length of each grid of θ axle is θ ∈ [0, π].Like this, obtaining transition matrix H is

$H=\left[\begin{array}{cc}\cos {\mathrm{\θ}}_1& \sin {\mathrm{\θ}}_1\\ {\cos \mathrm{\θ}}_2& \sin {\mathrm{\θ}}_2\\ ...& ...\\ \cos {\mathrm{\θ}}_{800}& \sin {\mathrm{\θ}}_{800}\end{array}\right]$

Wherein △ θ=θ _i+1-θ _i.

E2) data-mapping: by the set of data points D under rectangular coordinate system _abe mapped to (ρ, θ) space by transition matrix H, then have

Wherein, Num is D _amiddle data point number, the data point number namely obtained after steps d.

E3) numerical value of element in compute matrix R: definition matrix R, and be initialized as null matrix

Wherein, M _θ=800, M _ρ=800.If certain element ρ in matrix V _ij(i=1 ..., 800; J=1 ..., Num) drop on q grid of ρ axle, then the point on its homologous thread known drops on (i, q) the individual grid in (ρ, θ) space.Like this, by corresponding element r in matrix R _iq(i=1 ..., 800; Q=1 ..., 800) add 1.

F. by all elements in the matrix R that obtains after Hough process and the second thresholding Th ₂compare, if be greater than thresholding, then target detected; Otherwise then target does not exist.Wherein, Th ₂determined by following formula:

$P_{f_2}=P(r_{\mathrm{iq}}>{\mathrm{Th}}_2)$

In formula for total false-alarm probability, be 10 ^-4.

For verifying detection perform of the present invention, the method for Monte Carlo simulation is adopted to add up the detection perform under difference letter miscellaneous noise ratio.If the number of times of Monte Carlo simulation is 500, total false-alarm probability is 10 ^-4, detection perform curve as shown in Figure 3, as can be seen from Figure 3: compared with traditional Hough transform method, the present invention only when signal to noise ratio about-12dB ~-8dB detection probability slightly decline, then obviously do not distinguish in other situations.

For verifying operation efficiency of the present invention further, carry out the CPU time shared by the present invention and traditional Hough transform method contrasting that (CPU is Intel (R) Core (TM) 2 Quad CPU Q6600, dominant frequency is 2.4GHZ, inside saves as 8G).Time shared by these algorithms is averaging process by 1000 times and obtains, as shown in table 1.As can be seen from Table 1: operation efficiency of the present invention improves a lot, be 4 times of traditional Hough transform method.

Table 1

	Tradition Hough transform	The present invention (P=4) Hough transform
			CPU time (s)	32.01	7.91

Claims (4)

1. an object detection method for carrier-free UWB radar, by objective emission carrierfree pulse signal s (t), and carries out process detection target information to the scan echo signal received, it is characterized in that, comprise the following steps:

A, K the scan echo signal received is carried out to pulse by following formula and offsets process:

$s_k^d\left(n\right)=s_{k+1}\left(n\right)-s_k\left(n\right)$

Wherein, K is integer, is set by radar parameter; s _kn () is the measured value of the n-th range unit of a kth echoed signal; for the result after once offseting; N=1,2 ..., N _r; N _rfor distance samples is counted, set by radar parameter;

B, by slide window processing, energy accumulation is carried out to the signal after step a process:

$E_k\left(n\right)=\underset{n=1}{\overset{W}{\mathrm{\Σ}}}{\left[s_k^d\left(n\right)\right]}^2$

Wherein, E _kn () is the signal energy of accumulation; W is sliding window length, is set by radar parameter;

C, by E _k(n) and the first thresholding Th ₁relatively, number of targets strong point is obtained; Wherein, Th ₁set by radar parameter;

D, judgement are more than the first thresholding Th ₁number of targets strong point whether there is P continuously, if then retain first data point, then judge follow-up data point, by that analogy, obtain other data points; Wherein, P is positive integer, is set by radar parameter;

E, the data point obtained combined and carry out Hough transform process through steps d;

F, by the result after step e process and the second thresholding Th ₂compare, if be greater than the second thresholding Th ₂, then target is detected; Otherwise then target does not exist; Wherein, Th ₂set by radar parameter.

2. the object detection method of a kind of carrier-free UWB radar according to claim 1, is characterized in that, described pulse signal s (t) is Gauss pulse signal, has following expression:

$s\left(t\right)=\mathrm{\α}{e}^{-a^2{t}^2}$

Wherein, a be the pulsewidth factor and t is the duration of pulse; α is pulse height; A, T, α are set by radar parameter.

3. the object detection method of a kind of carrier-free UWB radar according to claim 1, it is characterized in that, in described step e, group of data points is combined into D _a:

$D_a=\left[\begin{array}{cccccccccc}1& 1& ...& 1& ...& k& ...& k& ...& K-1\\ n_{l_1}& n_{l_2}& ...& n_{l_j}& ...& n_{k_1}& ...& n_{k_j}& ...& n_{{(K-1)}_j}\end{array}\right]$

Wherein, k represents kth time scanning, k ∈ [1, K]; represent the individual range unit more than the data point place of the first thresholding of jth in kth time scanning.

4. the object detection method of a kind of carrier-free UWB radar according to claim 1, is characterized in that, in described step e, the concrete grammar of Hough transform is:

ρ＝xcosθ+ysinθ θ∈[0,π]

Wherein, ρ, θ are data point polar coordinates, represent Distance geometry angle parameter respectively; X, y are the rectangular coordinate of data point.