CN102058411A - UVB based multi-channel radar life detection instrument - Google Patents

Abstract

The invention discloses a UVB based multi-channel radar life detection instrument for multi-target detection, which comprises a front end of a UWB biologic radar and a computing unit. The front end of the UWB biologic radar comprises a transmitting antenna, three receiving antennas, an impulse oscillator, an electromagnetic pulse generator and a sampling integrator. The transmitting antenna and each receiving antenna form a channel and therefore three channels are formed. The computing unit analyses and processes the collected three radar echo signals and finally extracts life information of multiple human body targets and the two dimensional position information of each target.

Description

A kind of multichannel is based on the UWB radar life-detector

Technical field

The present invention relates to belong to noncontact life parameters Detection Techniques field, particularly a kind of radar life-detector that can be used for the multichannel of multiple target detection based on UWB.

Background technology

Radar life-detector is a kind of fusion Radar Technology and the penetrable nonmetal medium of biomedical engineering technology (brick wall, ruins etc.) noncontact, surveys a kind of emerging special radar of human life's body (breathing, heart beating, body are moving etc.) at a distance.The radar life-detector technology then is to be an emerging technology of the detection of a target with the life entity, is the very important cutting edge technology field that International Technology circle is generally acknowledged.Because this technology does not have any constraint to measured object, need not the connection of contact electrode, pick off, cable etc., and can be every certain distance, penetrate certain medium (as clothes, gauze, brick wall, ruins etc.) human body discerned detection, so can be widely used in fields such as disaster buried person person search and rescue, the monitoring of struggle against terror mid-board and battle reconnaissance, particularly have irreplaceable advantage in fields such as emergency management and rescue, anti-terrorisms.

Target recognition ability and distance, angular resolution are two emphasis of current radar life-detector area research, also are the key issues that this paper need break through.At present, comparatively sophisticated radar life-detector system based on the continuous wave radar system can only provide the unmanned result of people, and can't provide the distance of target and angle information etc., and penetration capacity also remains further to be improved.In view of the advantage that super wide range radar is had, we have adopted advanced in the world at present super wide range technology, and it is combined with noncontact life detection technology, and research is visited people's Radar Technology based on the noncontact of super wide range.

Existing radar type life detection technology is identified as the master with the detection to single goal, and multiobject detection and location are also only limited to moving target.Up to the present, still unresolved a plurality of static human body identification of targets in this field and two-dimensional localization problem.Many quiet target acquisitions identification location technologies are the new research direction and the difficult points in international life detection field, and this technology is the key technology of radar life-detector, and it is restricting the extensive use of radar life-detector.The solution of many quiet target acquisition identifications location difficult problem can greatly improve the detection efficient in the noncontact life detection, satisfies in the real work the localized demand of multiple target quick detection.

Summary of the invention

Technical problem to be solved by this invention is at the deficiencies in the prior art, and a kind of multichannel localized radar life-detector based on UWB of multiple target two-dimensional detection that realizes is provided, and solves the two-dimensional detection and the orientation problem of a plurality of static human body targets.

The present invention adopts following technical scheme:

A kind of multichannel comprises UWB bioradar front end and computing unit based on the radar life-detector of UWB, and described UWB bioradar front end comprises a transmitting antenna, three reception antennas, pulse oscillator, electromagnetic pulse generator, Sampling Integral device; Described transmitting antenna and each described reception antenna are formed a passage, form three passages altogether; Described pulse oscillator produces pulse signal, and this signal triggering electromagnetic pulse generator produces burst pulse, and radiate by described transmitting antenna; Reflected signal is delivered to the Sampling Integral device through each described reception antenna, the pulse signal that is produced by pulse oscillator produces range gate through delay circuit and range gate generator simultaneously, select to received signal, signal is by the Sampling Integral circuit, be detected through accumulation back small-signal, and amplify via amplifier and wave filter, filtering obtains three road radar echo signals, described three road radar echo signals are sent into computing unit after the sampling of high-speed a/d capture card, by computing unit three road radar echo signals that collect are carried out analyzing and processing, finally extract the two-dimensional position information of a plurality of human body target life-informations and each target.

Described multichannel UWB radar life-detector, described transmitting antenna and one of them reception antenna are closely arranged and are placed central authorities, and other two reception antennas are arranged at both sides, form the structure of likeness in form dumbbell.

Described multichannel UWB radar life-detector, described computing unit comprises the signal integration module, the signal decomposition reconstructed module, digital filtering module and numerical differentiation module, spatial-frequency analysis module and filtered back projection's locating module, described signal integration module is carried out integration to three road radar echo signals respectively on distance, described signal decomposition reconstructed module will be decomposed respectively through three road radar echo signals behind the integration, reconstruct, synthetic three road target echo signals and three road distance signals, described digital filtering and numerical differentiation module are carried out digital filtering and numerical differentiation respectively to three road target echo signals, described spatial-frequency analysis module is used for carrying out spatial-frequency analysis according to three road target echo signals after digital filtering and the numerical differentiation and three road distance signals, obtains three projection signals of target; Described filtered back projection locating module is used for determining the two-dimensional position information of target and forming display image according to described three projection signals.

Described multichannel UWB radar life-detector also comprises projection signal's pretreatment module, is used for described three projection signals are gone intermediate value and normalized, and signal after the pretreatment is sent to described filtered back projection locating module.

Described multichannel UWB radar life-detector, described filtered back projection locating module comprise one-dimensional Fourier transform module, one dimension weight factor module, one dimension inverse Fourier transform module, the direct back projection module that connects in turn; Described one-dimensional Fourier transform module is used for projection signal after the pretreatment of three passages is done one-dimensional Fourier transform; Described one dimension weight factor module is used for multiply by the one dimension weight factor to described through the projection signal after the one-dimensional Fourier transform | ρ |; Described one dimension inverse Fourier transform module is used for multiply by the one dimension weight factor | ρ | after projection signal make inverse Fourier transform; Described direct back projection module is used for the projection signal through inverse Fourier transform is done direct back projection.

Described multichannel UWB radar life-detector, the one dimension weight factor | ρ | finally determine be shown below:

g _θ(t) be the projection signal after certain passage processing, F ₁{ g _θ(t) } be projection signal behind the one dimensional fourier transform.

Above-mentioned arbitrary described multichannel UWB radar life-detector also comprises the hangover cancellation module, is used for obtain the display image elimination of trailing through described filtered back projection locating module location.

Described multichannel UWB radar life-detector, described hangover cancellation module adopt the elimination of trailing of following method: the pixel value in the two dimensional surface of viewing area is pre-seted a threshold value, and the pixel that will be lower than described threshold value is painted background color.

Described multichannel UWB radar type life-detection system, the display mode of described display image are the pseudo-color display mode of two dimensional surface, and while range of a signal and angle, realize that multiobject location and result of detection show.

Innovation part of the present invention is:

(1) proposed first realization enhancing, human body identification and the one-dimensional distance of the faint vital signs of static human body target have been distinguished, carried out the new method of multiple target two-dimensional localization again, for a plurality of static human body target localizations of radar life-detector are opened up new approach.

(2) adopting to change the Time-Frequency Analysis Method of shape--the one-dimensional distance of empty frequency analysis (space, frequency) is distinguished algorithm the echo-signal of the super wide range radar life-detector of single channel system acquisition is split, recombinates and relevant processing, and being expected provides new method for the one-dimensional distance differentiation of a plurality of quiet targets in the life detection.

(3) optimal antenna battle array frame mode has been proposed: long dumbbell shape structure.Survey with this frame mode, can make detection system with minimum antenna, the simplest structure obtains best multiple target locating effect.

Description of drawings

Fig. 1 is the super wide range radar life-detector of single channel system principle diagram;

Fig. 2 is the super wide range radar life-detector of multichannel computing unit structural representation;

Fig. 3 is provided with sketch map for super wide range radar parameter;

Fig. 4 is target echo signal and distance signal;

Fig. 5 forms block diagram for the hardware filtering circuit;

Fig. 6 is that the signal waveform before and after the differential algorithm compares (30 seconds data);

Fig. 7 carries out discrimination result for the crest method of discrimination to binocular mark data

Fig. 8 determines the algorithm sketch map for angle;

Fig. 9 is the electromagnetic wave propagation path of bistatic antenna form;

Figure 10 is the antenna echo signal of transceiver and bistatic antenna form;

Figure 11 determines sketch map for the target two-dimensional position of multi-channel system;

Figure 12 is filtered back projection's method;

Figure 13 is the positioning result figure of (threshold value 150) behind the elimination conditions of streaking;

Figure 14 is the positioning result figure of (threshold value 230) behind the elimination conditions of streaking;

Figure 15 is single goal positioning result figure;

Figure 16 demarcates position figure as a result for binocular;

Figure 17 is three target localizations figure as a result;

Figure 18 demarcates position figure (hangover is eliminated) as a result for binocular;

Figure 19 is three target localizations figure (hangover is eliminated) as a result.

The specific embodiment

Below in conjunction with the drawings and specific embodiments, the present invention is described in detail.

Embodiment 1

Present embodiment is the example explanation with a certain passage in the multichannel, and Fig. 1 is the super wide range radar life-detector of single channel system principle diagram.At first pulse oscillator produces pulse signal, and this signal triggering electromagnetic pulse generator produces burst pulse, and radiate by transmitting antenna.Reflected signal is delivered to the Sampling Integral device through reception antenna, the signal that is produced by pulse oscillator produces range gate through delay circuit, select to received signal, signal is by the Sampling Integral circuit, be detected through small-signal after the accumulation of thousands of pulses, and carry out amplification filtering, after the high-speed collection card sampling, send into computing unit again, by computing unit the signal that collects is carried out analyzing and processing and identification, calculate target range at last.

As shown in Figure 1, be radar front end in the frame of broken lines, the mid frequency and the bandwidth of system are all 500MHz, and the wave beam angle of coverage is 60.Computing unit command range door generator obtains the echo-signal of different distance section in the search coverage.

The parameter of computer-controllable system is: initial distance, investigative range, sample frequency and antenna gain.As shown in Figure 3, after antenna penetrated brick wall, search coverage was one fan-shaped, by initial distance and investigative range are set, can realize the scanning probe of the sector region of dash area among the figure, if echo-signal shows target information after by analysis, just can judge in this sector region has target.By the initial distance of continuous adjustment, can realize the tomoscan in certain zone.And adjust investigative range (reception of antenna is counted constant), and then can adjust the sensitivity of detection system, change the target range resolving power of system, realize the coarse scan in certain zone and carefully sweep.

For example, initial distance is set to 6m (40 nanosecond), investigative range is set to 3m (20 nanosecond), and the echo-signal of native system is the sequence that 2048 points are formed, and the effective search coverage of so current radar is antenna dead ahead 6m～9m, angle is 60 sector region, echo-signal only reflects the information that vertically goes up 3m, and the scope of 3m on average is divided into 2048 parts, and promptly each sampling obtains 2048 data, we are referred to as 2048 points, and the distance of n point representative is:

$s=6+\frac{n}{2048}\×3\left(m\right)......\left(1\right)$

In the formula (1): n is an ordinal number.

According to nyquist sampling theorem, sample frequency must be greater than the twice of signal highest frequency, and we set the A/D sample frequency is 64Hz.

Fig. 2 is the super wide range radar life-detector of multichannel of the present invention computing unit structural representation; Described computing unit comprises the signal integration module, the signal decomposition reconstructed module, digital filtering module and numerical differentiation module, the spatial-frequency analysis module, described signal integration module is carried out integration to signal on distance, described signal decomposition reconstructed module is broken up signal and is decomposed, reconstruct, synthetic target echo signal and distance signal, described digital filtering and numerical differentiation module are carried out digital filtering and numerical differentiation to target echo signal, described spatial-frequency analysis module is used for carrying out spatial-frequency analysis according to target echo signal after digital filtering and the numerical differentiation and distance signal, obtains the target one-dimensional distance.

Embodiment 2

Present embodiment is that example describes quiet target small-signal Enhancement Method with a passage in the multichannel,

Realize the static human body identification of targets, at first should the faint life signal of static human body be strengthened.In the present embodiment,, adopt faint processing of biomedical signals method, improve signal to noise ratio, realize basic identification human body target to handle the enhancing with useful signal through the signal after the high-speed sampling at the characteristics of UWB radar echo signal.

4 dot product point-scores carry out integration to signal between adopting at 8 on distance; Again signal is broken up decompose, reconstruct, synthetic target echo signal and distance signal; Target echo signal is carried out digital filtering and numerical differentiation, to realize the enhancing of weak useful signal.

2.1 the integration of signal

The high-speed collection card sample rate that adopts in the present embodiment is 64Hz, and then the data volume after the AD sampling is big, is unfavorable for real-time operation; The data volume minimizing can cause echo-signal to lack enough range informations again too much.So choose between 84 integration methods the back signal of sampling carried out subsection integral guaranteeing to have under the situation of enough range resolutions present embodiment.

4 integration methods are averaged per 8 additions of data exactly between 8, per twice integration interval 4 point (0～7,4～11,8～15, analogize the back), make sampling back signal data amount become 1/4th of original signal through the last integration of distance, under the situation of not losing signal characteristic, reduced the sequence length of signal, reduce operand, accelerated arithmetic speed.

2.2 signal is decomposed and reconstituted

With signal behind the integration by time and space two territories decompose, reconstruct, the synthetic distance signal y (d) that contains the target echo signal x (t) of temporal information and contain spatial information, wherein t is a time variable, d is apart from variable.Target echo signal reflection be the time dependent situation of signal amplitude on the respective distances point, the abscissa of target echo signal is the time; Distance signal then is the sequence that the amplitude of the each point on the synchronization different distance is formed, and the abscissa of distance signal is a distance.Fig. 4 is road target echo signal (1600 points, 25 seconds) and distance signal (60ns, 9m) oscillogram of selecting at random.

Target echo signal has improved signal to noise ratio, more helps the extraction of vital sign signals, and distance signal has guaranteed suitable range resolution again when reducing operand greatly.

2.3 the selection of wave filter

2.3.1 hardware filtering device

The hardware filtering circuit is inserted before the high-speed AD acquisition card in the present embodiment, filter bandwidht is adjustable, in the early stage preliminary experiment, successively having tested bandwidth is several wave filter of 0.08-10Hz, 0.08-100Hz, 0.08-1000Hz, 0.08-2000Hz, 0.08-3000Hz, 0.08-4000Hz, 0.08-5000Hz, contrast by effect, finally selected the passband of 0.08-5000Hz as the hardware filtering circuit, gain is divided into two grades: gain is 1 o'clock, amplification is 1 times, gain is 2 o'clock, and amplification is 2 times.

Adopt the single channel UWB system, single channel UWB system that adds hardware filter circuit (gain is 1) and the single channel UWB system random acquisition data that add hardware filter circuit (gain is 2) each 16 groups (driftlessness, single goal data) that do not add the hardware filter circuit respectively, add up to totally 48 groups of data.The algorithm that adopts computing unit to comprise respectively to these 48 groups of data is handled and is differentiated, the statistics recognition correct rate, and statistical result is as shown in table 1 below.

Recognition correct rate situation (48 groups of data) during table 1 increase and decrease hardware filtering device

Using gain is that the recognition correct rate of 1 hardware filtering circuit is the highest, is 62%.

By relatively finding, adopt the Effect on Detecting of the UWB system that gain is 1, passband is 0.08-5000Hz hardware filtering circuit best.

2.3.2 digital filter

Because phase information is extremely important to quiet target detection in the faint vital sign parameter signals, and quiet target recognition and one dimension are distinguished technology having relatively high expectations to algorithm stability and follow-up Digital Signal Processing, so adopt finite impulse response (FIR) (FIR) wave filter to remove High-frequency Interference in the present embodiment, extract useful signals such as breathing.The system function of FIR wave filter is:

$H\left(z\right)=\underset{n=0}{\overset{N-1}{\mathrm{\Σ}}}h\left(n\right)z^{-n},0\≤n\≤N-1......\left(2\right)$

Difference equation is:

$y\left(n\right)=\underset{k=0}{\overset{N-1}{\mathrm{\Σ}}}{b}_{k}x(n-k)......\left(3\right)$

The selection of filter order has been directly connected to its amplitude-frequency characteristic, and exponent number is high more, and amplitude-frequency characteristic is good more, and filter effect is good more.Also bring some negative effects but unrestrictedly increase exponent number, as increased system's operand, prolonged the time delay of filtering output etc.Comprehensive above two aspects consider that under the situation that system's operational capability allows, we select for use 160 rank FIR wave filter to test.

On Filter Design, adopt the window function method,, finally adopted the hamming window by contrasting the amplitude-frequency characteristic of several window function low pass filters.

Normal condition servant's breathing rate is per minute 15～20 times, considers that its frequency of abnormal condition generally can not surpass 0.4Hz yet.So the digital filter that we adopt, its cut-off frequency is this index of main reference 0.4Hz also, and promptly the low pass filter that is not less than 0.4Hz with cut-off frequency carries out filtering to target echo signal, with the performance of each wave filter relatively.We test the lowpass digital filter that cut-off frequency is respectively 0.4Hz, 0.5Hz, 0.6Hz, 0.7Hz, 0.8Hz in this article.

Picked at random investigative range (window when being radar) be 20 nanoseconds (3m) and 60 nanoseconds (9m) data each 48 groups, amount to 96 groups, these signals are signal (data that contain driftlessness, single goal) after the sampling of single channel UWB system, and the algorithm that these 96 groups of The data computing units comprise is differentiated.In the contrast experiment, only change filter cutoff frequency, other each software and hardware parameter constants, statistics is differentiated result's accuracy, and it is as shown in table 2 that it differentiates accuracy.

Table 2 changes filter cutoff frequency to differentiating the influence of accuracy

According to above experimental result, by comprehensive comparison, the hamming window Finite Impulse Response filter of finally choosing 160 rank, cut-off frequency and be 0.5Hz comes target echo signal filtering High-frequency Interference, keeps vital sign signals such as breathing.

2.4 the selection of differentiator

Because the existence of DC component and baseline drift phenomenon often comprises the very big extremely low frequency composition of energy in the target echo signal, make signal substantial deviation baseline faint life signal identification to be produced very big influence.Present embodiment proposes to adopt the method for numerical differentiation to come filtering DC component in time and extremely low frequency to disturb, and makes useful signal center on zero base line and fluctuates up and down, to reach the purpose that strengthens vital sign signals such as breathing.The computational process of differentiator is as shown in Equation (4):

$y\left(n\right)=x\left(n\right)-\frac{\underset{k=n-m}{\overset{n-1}{\mathrm{\Σ}}}x\left(k\right)}{m}......\left(4\right)$

In the formula: y is an output signal, and x is an input signal, and m is an exponent number, and n is the sequence number of point.

Picked at random investigative range (window during radar) be 20 nanoseconds (3m) and 60 nanoseconds (9m) data each 48 groups, amount to 96 groups, these signals are signal (data that contain driftlessness, single goal) after the sampling of single channel UWB system, adopt the digital differentiator on 20 rank, 40 rank, 60 rank, 80 rank, 100 rank, 120 rank, 140 rank, 160 rank, 180 rank to carry out discerning after differential is handled and distance calculating respectively to these data, it is as shown in table 3 that it differentiates accuracy:

Table 3 changes the differentiator exponent number to differentiating the influence of accuracy

As can be seen, the differentiation accuracy of the signal after 60 exponent number word differentiators are handled is the highest, and total accuracy of its 96 groups of data is 60.78%.By comparing, present embodiment has selected for use 60 exponent number word differentiators to remove DC component and extremely low frequency disturb, the vital signs of enhancing signal.

From Fig. 6 more as can be seen, through after target echo signal carried out 60 rank differential in time, signal has been got back near the baseline and tightly and has been fluctuateed up and down around baseline, DC component and extremely low frequency composition have obtained inhibition, useful signal has obtained enhancing.

Embodiment 3

Present embodiment is that example describes one-dimensional distance differentiating method and spatial-frequency analysis method with a certain passage in the multichannel:

After finishing quiet target small-signal enhancing, will on distance, distinguish human body target.Because distance signal is the ultra-low frequency signal of reflection target range information, so contain spatial information and characteristics such as non-stationary at distance signal, present embodiment has made up the joint distribution function of space, frequency, adopt spatial frequency conjoint analysis method (changing the time frequency analysis of shape) the signal analysis of adjusting the distance, describe energy density and the intensity of signal on different distance, frequency, thereby provide the range information of each human body target.

What time frequency analysis was represented is the situation of change of signal spectrum on time shaft, after becoming time variable apart from variable, what the time frequency analysis result represented is exactly frequency spectrum situation of change spatially, so utilizing these characteristics of time frequency analysis comes the target on the different distance is carried out spectrum analysis, and then the acquisition human body is differentiated result and target one-dimensional distance information, so just formed space, application form that this time frequency analysis of frequency conjoint analysis is new, its essence remains time frequency analysis.

In the present embodiment, time variable in the time frequency analysis is become space (distance) variable, make up space, frequency associating function, make it can utilize space, frequency information to describe the energy density of input signal simultaneously, make this method possess " location " function of space, frequency, thereby the method for good non-stationary signal Frequency Estimation in a certain distance range is provided for us.

Time-frequency conversion comprises single linear conversion such as short time discrete Fourier transform, and bilinear transformation such as Wei Na-Wei Er distributes, wavelet transformation etc.Spatial frequency transforms in the present embodiment is the time variable in the time frequency analysis to be replaced to space (distance) variable and next, and its range resolution is to determine in advance, does not need to change by changing window width; And object of experiment is a static human body, its breath signal is comparatively stable in long-time, belong to steadily local and the big non-stationary signal of length, be fit to analyze for this class signal with short time discrete Fourier transform, carry out spatial-frequency analysis so chosen the short time discrete Fourier transform of single linear conversion in the present embodiment, and result is analyzed and compared.

In the present embodiment with the time window be that distance signal evenly is divided into 100 sections on distance behind the differential of 60ns (corresponding 9m investigative range, initial distance are 1ns), corresponding range resolution is about 0.09m.For removing the influence of antenna direct wave, preceding 12 of distance signal is abandoned, do not participate in segmentation, 500 of backs are divided into 100 sections, then 5 on each section amplitude is done an addition in the section, obtain and as the value of this section, thereby form the new distance signal that has only 100 numerical value to form, these 100 numerical value correspondences be to finish since 12 * 9/512=0.21m to 9m, target echo signal of the point that equally distributed each distance is last.

After the segmentation,, take out all new distance signals in this time period every 10 seconds, amount to 64 * 10=640 new distance signal (64 is sample rate) according to the needed positioning result refresh rate of actual detection; Each distance signal is split into point (100 point), and in chronological sequence order is with the sequence reorganization with each point again, and formation contains the fresh target echo-signal (amounting to 100 groups of target echo signals) of temporal information; Each new target echo signal is joined end to end by distance antenna order from the close-by examples to those far off, constitute the input signal of spatial-frequency analysis.

Synthetic input signal is made spatial-frequency analysis, promptly make short time discrete Fourier transform, wherein window width is corresponding to the length of fresh target echo-signal, be decided to be 64 * 10=640, the each sliding distance of window is corresponding to the range resolution of distance signal, according to the count principle that is not less than window width and differentiate the result to select Fourier transform to count to the demand of frequency resolution be 1024 points of conversion.Determine after the above parameter input signal to be carried out short time discrete Fourier transform, and draw figure as a result.The short time discrete Fourier transform formula is as the formula (5):

STFT(t，w)＝∫S(τ)γ(τ-t)e ^-jwτdτ ......(5)

Wherein S (τ) is an input signal, and γ (t) is a window function.

In the selection of window function length,, usually require the window function time width of selection short as far as possible in order to improve the temporal resolution of short time discrete Fourier transform.On the other hand, short time discrete Fourier transform will be expected high frequency resolution, then require the window function time width of selection long as far as possible, so the raising of temporal resolution contradicts with the raising of frequency resolution.In the reality, the width of the window function γ (t) of selection should adapt with the steady length of the local of signal.In this experiment, the eupnea frequency of detected object human body is per minute 15-20 time, be that 3-4 finishes the respiration motion second, for influence and the assurance frequency resolution that reduces human body respiration accidentalia, individual variation, the time width of the window function that we choose is 10 seconds, and the window width that corresponds in the spatial-frequency analysis is 640.

Embodiment 4

To be example with a certain passage in the multichannel describe the setting of crest method of discrimination and threshold value present embodiment:

The result of space, frequency analysis is one 3 dimension (space, frequency, energy) corresponding relation, and two coordinate axess are respectively distance and frequency, and energy intensity is to come corresponding by the depth of color.By suitable mode and suitable human life feature decision threshold is set, can realize that promptly single channel is distinguished the distance of a plurality of quiet targets and distance is calculated.If on a certain distance, signal energy is big, the spectrum peak is concentrated, the signal energy on the neighbor distance, and meet decision threshold, then thinking has the static human body target (one-dimensional distance is determined) on the respective distance in this reception antenna investigative range; If have the signal of macro-energy to occur on a plurality of distances, and meet threshold value, then thinking has the static human body target to exist on a plurality of distances, writes down the one-dimensional distance value of these targets by algorithm, is the distance of each target to antenna.

It is as follows that multiobject differentiation and distance are calculated concrete steps:

Find out energy value maximum in 100 sections 15 sections, and find out all energy crests in these 15 sections, be designated as E respectively by the energy size _Peak1, E _Peak2, E _Peak3... crest is such regulation: promptly the energy value of this section is greater than adjacent two sections energy values, and then this section is a crest.Find out after the energy crest, the section sequence number of record crest place section is used for the subsequent calculations target range.

The average energy value of 10 sections of energy value minimum is designated as in calculating 100 sections: E _Mean, utilize crest energy and the average energy value of minimum 10 sections to make comparisons to determine the number of target.Compare threshold is as follows:

(1) if the ENERGY E of energy crest 1 _Peak1Minimum average B configuration energy value E greater than 4 times _Mean, i.e. E _Peak1＞4E _Mean, think that then crest 1 position has target to exist, the distance of target is calculated by the sequence number of crest 1;

(2) if the ENERGY E of energy crest 2 _Peak2Minimum average B configuration energy value E greater than 3 times _Mean, i.e. E _Peak2＞3E _Mean, think that then crest 2 positions have target to exist, the distance of target is calculated by the sequence number of crest 2;

(3) if the ENERGY E of energy crest 3 _Peak2Minimum average B configuration energy value E greater than 2.5 times _Mean, i.e. E _Peak3＞2.5E _Mean, think that then crest 3 positions have target to exist, the distance of target is calculated by the sequence number of crest 3.

Fig. 7 be according to the crest method of discrimination and decide the discrimination result that threshold value is carried out binocular mark data.

As can be seen, there are two crests in 100 segment signals, lay respectively at the 54th section and the 82nd section, the energy of these two crests through with separately threshold ratio, draw place, two crest present positions and be target, by calculating, the distance of two targets is respectively: (54-1) * and 0.09+0.21=4.98m and (82-1) * 0.09+0.21=7.50m.

So far, we have finished each passage to the identification of a plurality of static targets on the different distance and the calculating of each target range, multiobject one-dimensional distance information has promptly been arranged, on this basis, the subsequent treatment that the one dimension result of three passages is correlated with forms the projection signal of each passage on two dimensional surface again.

Embodiment 5

5.1 the angle of transceiver antenna form is determined

Range information has been arranged, realize two-dimensional localization, also needed angle information, for the antenna form of transceiver, we adopt the cosine law just can solve the problem that angle is determined.Its algorithm sketch map as shown in Figure 8.

Target is to the distance A and target being drawn by distance differentiation algorithm apart from B to antenna 2 of antenna 1, spacing C between antenna 1 and the antenna 2 is known, we can solve the angular relationship between target and the antenna 1 according to formula (6), be angle [alpha], utilize the combination of polar coordinate middle distance and angle promptly can draw the two-dimensional position information of target again.

2AC?cosα＝A ²+C ²-B ² ......(6)

5.2 the angle of bistatic antenna form is determined

In the multi-channel system of reality, transmitting antenna and reception antenna separate, as shown in Figure 9.Suppose that transmitting antenna Tx is D to the distance between the reception antenna Rx, the Tx range-to-go is S ₀(t), the Rx range-to-go is S ₁(t), then the electromagnetic wave that sends of transmitting antenna Tx needs time D/c (c is the aerial spread speed of electromagnetic wave) could arrive reception antenna Rx, in the echo-signal that Rx receives, just have one section direct wave like this, the waveform of this section is a straight line, it does not comprise any target information, this segment distance should be taken into account when signal processing.The electromagnetic wave propagation path as shown in Figure 9 during the antenna form target acquisition of actual bistatic.

Figure 10 is the contrast of the reception antenna echo-signal of transceiver antenna form and bistatic antenna form, wherein figure (a) is the echo-signal of transceiver antenna form, figure (b) is the echo-signal of bistatic antenna form, and reception antenna is according to transmitting antenna 1.5m.The signal location setting of two paths of signals is 1ns, the time window setting be 20ns.As can be seen: the transceiver antenna form because receive, the transmitting antenna distance is very near, so all include the reflective information of medium when its echo-signal is whole in the window; And the bistatic antenna form is because reception and transmitting antenna have left a segment distance, so be baseline direct wave stably the last period during its echo-signal in the window, the electromagnetic wave that direct wave length is sent by transmitting antenna directly arrives the required time decision of reception antenna, do not comprise reflected by objects information in any detecting area, the second half section of echo-signal just begins to occur comprising the waveform of target reflection information.

Target two-dimensional position for the multi-channel system of bistatic antenna form is determined as shown in figure 11.

This is the location sketch map of a simple single goal, and it realizes two-dimensional localization by a transmitting antenna Tx and a group of received antenna array (comprising two reception antenna Rx1 and Rx2).The position of target is to determining by following electromagnetic wave stroke is calculated, that is: the electromagnetic wave that sends of transmitting antenna arrives target, after being returned by target reflection again, arrive again each reception antenna the stroke of process.The Tx range-to-go is S ₀(t), target is S1 (t) to the distance of Rx1, and target is S to the distance of Rx2 ₂(t).Electromagnetic wave emits the arrival target from Tx, returns to arrive Rx1 and the used time (when walking) of Rx2 is respectively τ from target reflection again ₁=(S ₀(t)+S ₁(t))/and c, τ ₂=(S ₀(t)+S ₂(t))/c.By τ ₁And τ ₂Can determine two ellipses, oval focus is the position of transmitting antenna Tx and corresponding reception antenna Rx1 and Rx2.Like this, the position of target can be determined that oval computational process sees that formula (7) is to formula (10) by the intersection of two ellipses.

${\left(\frac{x\left(t\right)+\frac{D}{2}}{a_1\left(t\right)}\right)}^2+{\left(\frac{y\left(t\right)}{b_1\left(t\right)}\right)}^2=1......\left(7\right)$

${\left(\frac{x\left(t\right)-\frac{D}{2}}{a_2\left(t\right)}\right)}^2+{\left(\frac{y\left(t\right)}{b_2\left(t\right)}\right)}^2=1......\left(8\right)$

Here 2a _iBe long axis of ellipse, electromagnetic wave reflected back stroke that reception antenna is walked again from the transmitting antenna to the target just, it is by time τ _iCalculate, its computational process as shown in Equation (9), wherein i gets 1,2, is the numbering of reception antenna.

2a _i(t)＝S ₀(t)+S _i(t)＝cτ _i(t) ......(9)

Oval minor axis 2b _iCan calculate by formula (10).

${\left(\frac{D}{2}\right)}^2+{b_i}^2\left(t\right)={a_i}^2\left(t\right)......\left(10\right)$

The precision of this algorithm depends on the distance between the antenna array, the size dimension of target and the time delay τ that is gone out by the antenna echo calculated signals _iPrecision, this also requires us to use high-precision UWB radar cell, the theoretical full accuracy of the super wide range of the multichannel that uses in present embodiment system is 4ns/2048 ≈ 2ps (psec), be converted into range accuracy be about 2ps * c=0.03cm (centimetre), satisfy required precision.

5.3 antenna array frame mode

Before experiment is carried out, following priori has been arranged:

(1) the super wide range of multichannel system can not work by two or more transmitting antennas simultaneously, otherwise can interfere with each other, so only need adopt a transmitting antenna when design.

(2) single channel can only be determined multiobject distance, and can not provide the angle information of target, thus need plural passage just may realize two-dimensional localization of target at least, so will in experiment, select to position more than the mode of two reception antennas.

(3) under the identical condition of locating effect, should select a kind of mode of using number of antennas minimum, so can reduce volume, the weight of system, improved portability, apply after convenient, the minimizing of antenna simultaneously also greatly reduces the complexity of systematic sampling, computing etc., has improved operation efficiency.

(4) because we will realize two-dimensional localization of target rather than three-dimensional imaging, so only need only need antenna is placed on the same horizontal line at concrete laboratory test platform with selected antenna as for getting final product on the same horizontal plane.

Based on above 4 points, carried out following experiment:

5.3.1 choosing of reception antenna number

Two passages can be realized the location of how quiet target, but wherein may contain pseudo-shadow, and at this moment the projection signal by third channel verifies and eliminate pseudo-shadow, and promptly the crossing point of the elliptic arc of the projection signal of three passages is only real place, target location.In actual detection experiment, also proved this point, so finally on antenna amount, selected the form of a transmitting antenna, three reception antennas.

5.3.2 determining of transmitting antenna position

Situation high according to the station of normal adult and sitting height is added up, in order to ensure all reaching best Effect on Detecting to target stance and sitting posture, the antenna height of antenna is decided to be 1.2m, and promptly a transmitting antenna and three reception antennas all are on the horizontal line of height 1.2m.

5.3.3 determining of reception antenna position

Determined after the position of transmitting antenna that adjacent transmitting antenna has been placed a reception antenna, so just formed a single channel system that is similar to the transceiver form, this passage is mainly used to target is carried out the last differentiation of distance.In order to guarantee the symmetry of search coverage angular resolution, two remaining reception antennas are the center with the transmitting antenna, the symmetric both sides that are arranged on the high 1.2m horizontal line, and the range transmission antenna is respectively 0.5m, 1.0m and 1.5m.Found through experiments, distance is near more, and the angular resolution of target is low more, and distance is far away more, and angular resolution is high more, that is: there is proximate inverse relation in the distance of both sides reception antenna and central transmitting antenna with angle on target resolution.When distance is infinitely small when the adjacent transmitting antenna, the effect of three passages is the same, and this moment, three passages all only can carry out distance differentiation to multiple target, can say and have no angular resolution.According to above experiment situation, in order to guarantee maximum angular resolution, finally selected an adjacent transmitting antenna of reception antenna, and two other reception antenna range transmission antenna 1.5m, and be distributed in the transmitting antenna both sides symmetrically.

5.3.4 brief summary

One three receipts on antenna amount, for realizing the localized minimum antenna number of multiple target, wherein a pair of dual-mode antenna is closely arranged and is placed central authorities, two other reception antenna to set up the frame mode of the likeness in form dumbbell that places both sides.We also find by experiment, and both sides reception antenna and the distance L of central transmitting antenna become the relation of approximate reverse ratio, i.e. L ∝ 1/ θ with angular resolution θ.

Embodiment 6

6.1 filtered back projection's algorithm for reconstructing

What present embodiment filtered back projection method for reconstructing adopted is the way of first correction, back back projection, can obtain comparatively accurate original density function, promptly each passage is revised earlier through the data for projection that calculates, and then back projection is on each pixel on perspective plane, thereby recovers primary density function.

At first projection signal is gone intermediate value, normalization correction:

Projection signal's (100 point) to each passage finds out its median, will be less than the value zero setting of median, and all the other are constant.Consistent to the contribute energy power of final two-dimensional localization figure for the projection signal that makes three passages, by formula (11) are to removing the signal normalization after the intermediate value:

$r\left(t\right)=2\×\frac{e\left(t\right)-\underset{0\≤t\≤T}{\min }\left[e\left(t\right)\right]}{\underset{0\≤t\≤T}{\max }\left[e\left(t\right)\right]-\underset{0\≤t\≤T}{\min }\left[e\left(t\right)\right]}-1......\left(11\right)$

Here the time variable that refers to of t, T is the length of projection signal, and e is input, and r is output.

Correction has been got well after the projection signal, just the projection signal of three passages can be carried out filtered back projection toward search coverage.

The basic thought of this filter back-projection algorithm is: after extracting projection function (one dimension function) in the echo-signal of a certain reception antenna, this one dimension projection function is done Filtering Processing, obtain a projection function through revising, and then this revised projection function made backprojection operation, draw required density function.The process of filtered back projection's method reconstructed image as shown in figure 12.

The step of filtered back projection's method reconstructed image is as follows:

(1) projection function of certain reception antenna is done one-dimensional Fourier transform;

(2) transformation results to (1) is multiplied by the one dimension weight factor;

(3) weighted results of (2) is made the one dimension inverse Fourier transform;

(4) the corrected projection function that draws in (3) is done direct back projection;

(5) repeat the process that (1) arrives (4), up to the back projection that finishes each passage projection signal;

According to the scope of respiratory frequency, in conjunction with the contrast of a large amount of experiment effects, weight factor | ρ | finally determine as the formula (12).

Compare with the method for reconstructing that first back projection, back are revised, filtered back projection's method only need be done one dimensional fourier transform, thereby shorten the time of image reconstruction when image reconstruction.

6.2 conditions of streaking and solution

As can be seen, the target location draws by three elliptic arcs are crossing, can have conditions of streaking to each target like this.The solution of this problem is by the pixel value setting threshold to two dimensional surface, and the pixel that will be lower than threshold value paints that background color solves.Because the pixel value 0～255 of pcolor is corresponding cool colour (indigo plant)～warm colour (red) respectively, here we decide threshold value be 150, promptly the color more than the light green color shows on plane graph, and threshold value is lower than 150 the whole display background colors of pixel.The result who does has like this removed the hangover part of target, has given prominence to the position of target more.Remove the later network for location of hangover as shown in figure 13 by setting threshold.

Find that in experiment after threshold value was further raise, as threshold value is brought up to 230, the network for location target of single goal was more outstanding, effect is more obvious, threshold value be the later network for location of 230 removal hangover as shown in figure 14.

Threshold value is brought up to after 230 as can be seen, and the position of single goal is more outstanding, and conditions of streaking further is eliminated, and positioning accuracy further improves.But, the raising of threshold value neither be unconfined, threshold value is carried too high after, in the time of can causing multiple target detection, failing to judge of the target that energy is less, through groping of experiment repeatedly, balance is eliminated conditions of streaking as possible and is not caused multiple target this two principles of failing to judge as far as possible, and finally the threshold value of selecting is 150.

6.3 filtered back projection's algorithm for reconstructing positioning result

Multichannel based on the UWB radar life-detector on, adopt filtered back projection's method to carry out the detection identification positioning experiment actual through walls of driftlessness, single goal, binocular mark, three targets several situations such as (being the target that stands still), experiment is to carry out at laboratory, according to breadboard orientation, survey the left side of display result and point to south, the north is pointed on the right side.That Figure 15, Figure 16, Figure 17 are respectively that the employing filtered back projection method of actual measurement rebuilds is single, double, three quiet target localizations figure as a result.This three width of cloth figure is the imaging results of trailing and eliminating, and setting is: signal location 20ns, the time window 20ns, by calculating, its investigative range is 3-6m.Wherein Figure 15 is the positioning result of single goal, and the physical location of target is the 4m center; Figure 16 is a binocular target positioning result, and two target physical locations are respectively 4m 30 degree by north and 5m center, and the target location and the target actual position of the demonstration of the zone of the warm colour in the positioning result are identical substantially as can be seen; Figure 17 is the positioning result of three targets, and three target physical locations are respectively 3m 30 degree by north, 4m 30 degree by north and 5.5m center, and same warm colour zone is also identical substantially with the target actual position.

From the positioning result of top single, double, three targets as can be seen, filtered back projection's algorithm for reconstructing can comparatively accurately be discerned three human body targets that stand still with interior (containing three) under state through walls and locate, thereby has proved that filter back-projection algorithm can be applied to the detection location of a plurality of static human body targets of multichannel radar life-detector.

Figure 18, Figure 19 are respectively binocular mark after the elimination hangover of actual measurement, three target localizations figure as a result.Parameter is set to: signal location 15ns, the time window 20ns, its investigative range is 2-5m.Wherein Figure 18 is binocular target positioning result figure, and the physical location of two targets is respectively 4m 20 degree by north and 5m center; Figure 19 is the positioning result figure of three targets, the physical location of three targets be respectively 3m by north 30 the degree, 4m just in and 5m by north 20 the degree.After eliminating by hangover as can be seen, the red area in the positioning result is more outstanding, and target is more obvious, and the resolution of target is improved to a certain extent.

6.4 filtered back projection's algorithm for reconstructing efficiency evaluation

The filter back-projection algorithm efficiency evaluation is finished on based on UWB radar life-detector test platform at multichannel, and the frame mode of antenna array has been selected long dumbbell shape.The major parameter of system is set to: signal location 15ns, the time window 20ns.All experimental datas all penetrate the 30cm brick wall and gather, and experimental subject is needs according to a target number picked at random from 16 volunteers, and target was the state of standing still when all experimental datas were gathered, i.e. static human body target acquisition experiment.The distribution situation regulation of target is as follows: the fore-and-aft distance of promptly any two targets is 0.5m at least at interval, and lateral angles is 20 degree at least at interval.

Driftlessness data through walls have been gathered 17 groups altogether.According to above-mentioned mode classification the result of adopting data is added up, because all data are the driftlessness data, so there is not the situation of failing to judge and misjudging.Statistical result is as shown in table 4.

Table 4 is worn the correct localization situation of 30cm brick wall driftlessness data

Single goal data through walls are gathered 48 groups altogether, and wherein, target is positioned at 2m 30 degree by north, 2m center, 2m 30 degree by north; 3m 30 degree by north, 3m center, 3m 30 degree by north; 4m 20 degree by north, 4m center, 4m 20 degree by north; Each 4 groups of these data such as 5m 20 degree by north, 5m center, 5m 20 degree by north, totally 48 groups of data.According to above-mentioned mode classification the result of adopting data is added up equally, statistical result is as shown in table 5.

The correct localization situation of single goal data when table 5 is worn the 30cm brick wall

Binocular mark data in the data through walls have been gathered 60 groups altogether.In the investigative range of 2-5m, the distribution form that makes up with the different distance different angles stood still after two target locations were randomly dispersed in wall on the basis of satisfying the described condition of preamble.According to above-mentioned mode classification the result of adopting data is added up equally, statistical result is as shown in table 6.

Table 6 is worn the correct localization situation of 30cm brick wall binocular mark data

Three target datas in the data through walls have been gathered 85 groups altogether.In the investigative range of 2-5m, the distribution form that makes up with the different distance different angles stood still after the position of three targets was randomly dispersed in wall on the basis of satisfying the described condition of preamble.Because in algorithm, at most also only considered to three identification of targets and location, so there is not the situation of erroneous judgement.According to above-mentioned mode classification the result of adopting data is added up equally, statistical result is as shown in table 7.

Table 7 is worn the correct localization situation of 30cm brick wall three target datas

The result who is discerned after the localization process by above various data to different target number, target different distributions situation adds up, filter back-projection algorithm is 94% to 17 groups of driftlessness discriminating data accuracy, to 48 groups of single goal data locking accuracy is 81%, to 60 groups of binocular mark data locking accuracy is 78%, is 67% to 85 group of three target data correct localization.As seen, filter back-projection algorithm is the highest to the recognition correct rate of driftlessness data, and is minimum to three identification of targets accuracy.

On the whole, filtered back projection's algorithm for reconstructing can be applied to maximum three static human body identification of targets and location.

Should be understood that, for those of ordinary skills, can be improved according to the above description or conversion, and all these improvement and conversion all should belong to the protection domain of claims of the present invention.

Claims (9)

1. a multichannel is based on the radar life-detector of UWB, it is characterized in that, comprise UWB bioradar front end and computing unit, described UWB bioradar front end comprises a transmitting antenna, three reception antennas, pulse oscillator, electromagnetic pulse generator, Sampling Integral device; Described transmitting antenna and each described reception antenna are formed a passage, form three passages altogether; Described pulse oscillator produces pulse signal, and this signal triggering electromagnetic pulse generator produces burst pulse, and radiate by described transmitting antenna; Reflected signal is delivered to the Sampling Integral device through each described reception antenna, the pulse signal that is produced by pulse oscillator produces range gate through delay circuit and range gate generator simultaneously, select to received signal, signal is by the Sampling Integral circuit, be detected through accumulation back small-signal, and amplify via amplifier and wave filter, filtering obtains three road radar echo signals, described three road radar echo signals are sent into computing unit after the sampling of high-speed a/d capture card, by computing unit three road radar echo signals that collect are carried out analyzing and processing, finally extract the two-dimensional position information of a plurality of human body target life-informations and each target.

2. multichannel UWB radar life-detector according to claim 1 is characterized in that, described transmitting antenna and one of them reception antenna are closely arranged and placed central authorities, and other two reception antennas are arranged at both sides, form the structure of likeness in form dumbbell.

3. multichannel UWB radar life-detector according to claim 1, it is characterized in that, described computing unit comprises the signal integration module, the signal decomposition reconstructed module, digital filtering module and numerical differentiation module, spatial-frequency analysis module and filtered back projection's locating module, described signal integration module is carried out integration to three road radar echo signals respectively on distance, described signal decomposition reconstructed module will be decomposed respectively through three road radar echo signals behind the integration, reconstruct, synthetic three road target echo signals and three road distance signals, described digital filtering and numerical differentiation module are carried out digital filtering and numerical differentiation respectively to three road target echo signals, described spatial-frequency analysis module is used for carrying out spatial-frequency analysis according to three road target echo signals after digital filtering and the numerical differentiation and three road distance signals, obtains three projection signals of target; Described filtered back projection locating module is used for determining the two-dimensional position information of target and forming display image according to described three projection signals.

4. multichannel UWB radar life-detector according to claim 3, it is characterized in that, also comprise projection signal's pretreatment module, be used for described three projection signals are gone intermediate value and normalized, signal after the pretreatment is sent to described filtered back projection locating module.

5. multichannel UWB radar life-detector according to claim 4, it is characterized in that described filtered back projection locating module comprises one-dimensional Fourier transform module, one dimension weight factor module, one dimension inverse Fourier transform module, the direct back projection module that connects in turn; Described one-dimensional Fourier transform module is used for projection signal after the pretreatment of three passages is done one-dimensional Fourier transform; Described one dimension weight factor module is used for multiply by the one dimension weight factor to described through the projection signal after the one-dimensional Fourier transform | ρ |; Described one dimension inverse Fourier transform module is used for multiply by the one dimension weight factor | ρ | after projection signal make inverse Fourier transform; Described direct back projection module is used for the projection signal through inverse Fourier transform is done direct back projection.

6. multichannel UWB radar life-detector according to claim 5 is characterized in that the one dimension weight factor | ρ | finally determine be shown below:

g _θ(t) be the projection signal after certain passage processing, F ₁{ g _θ(t) } be projection signal behind the one dimensional fourier transform.

7. according to the arbitrary described multichannel UWB radar life-detector of claim 3 to 6, it is characterized in that, also comprise the hangover cancellation module, be used for the display image that obtains through the described filtered back projection locating module location elimination of trailing.

8. multichannel UWB radar life-detector according to claim 7, it is characterized in that, described hangover cancellation module adopts the elimination of trailing of following method: the pixel value in the two dimensional surface of viewing area is pre-seted a threshold value, and the pixel that will be lower than described threshold value is painted background color.

9. multichannel UWB radar life-detector according to claim 7, it is characterized in that, the display mode of described display image is the pseudo-color display mode of two dimensional surface, and while range of a signal and angle, realizes that multiobject location and result of detection show.

Images