CN106405571A - 2D aperture imaging algorithm inhibiting influence of object thickness in Terahertz single frequency point - Google Patents

Abstract

The invention relates to a two-dimensional aperture imaging algorithm for suppressing the influence of target thickness at a terahertz single frequency point, which is applied to the reconstruction of target two-dimensional direction-dimensional scattering characteristic distribution during terahertz short-distance imaging. The present invention uses a plurality of distance tangent planes parallel to the observation plane to perform phase compensation on all target strong scattering points, gathers the imaging results under all distance tangent planes and performs weighted average processing, thereby improving the distance dimension information between strong scattering points of volume targets The difference in the negative impact on 2D aperture imaging. In the two-dimensional and direction-dimensional imaging system of terahertz single frequency point, the accuracy of phase compensation in the distance direction of the target directly affects the imaging effect, so the algorithm described in the present invention has a more general application range and stronger applicability.

Description

Two-dimensional aperture imaging algorithm for suppressing the influence of target thickness at a terahertz single frequency point

technical field

The invention relates to an imaging technology, in particular to a two-dimensional aperture imaging algorithm for suppressing the influence of target thickness at a terahertz single frequency point.

Background technique

Terahertz waves: Terahertz waves (THz waves) generally refer to electromagnetic waves with a frequency in the range of 0.1THz-10THz (wavelength 3mm-30pm). characteristics and important research value.

Terahertz imaging: Terahertz imaging is one of the important application fields of terahertz waves. Based on the safety, high resolution and penetrability of terahertz waves, terahertz imaging has unique advantages: Compared with infrared and optical imaging, terahertz imaging can penetrate non-polar materials such as clothing, wood boards and plastics: Compared with X-ray radio frequency and microwave imaging, terahertz imaging can often obtain images with higher spatial resolution: Compared with X-ray imaging, terahertz imaging with milliwatt-level power is generally considered harmless to the human body. According to whether there is a terahertz source, the terahertz imaging technology applied to human security inspection can be divided into two types: active and passive. Passive imaging uses bolometers to detect terahertz waves emitted by the body to form an image of the body's surface. Because the terahertz waves radiated by the human body are very weak, the imaging results of passive imaging often have low contrast and the image is not clear enough. Active imaging generally uses antennas to emit milliwatt-level terahertz waves. Terahertz waves penetrate the clothes on the surface of the human body, and the image of the human body surface can be obtained by measuring the reflected terahertz signals. Active imaging not only has high image contrast, but also can achieve high azimuth resolution through the method of synthetic aperture. Active terahertz imaging technology can be used for close-range human security detection in airports, subway stations and other places.

Synthetic aperture: The synthetic aperture imaging system changes the position of the transceiver antenna unit to form a synthetic aperture by sequentially switching the electronic switch in one direction, and forms a synthetic aperture through mechanical scanning in the other dimension. The magnitude and phase of the scattered field data are recorded, and then the original image is reconstructed by focusing imaging on the scattered field data.

Short-distance imaging: The division of short-distance and long-distance imaging is related to the division of the radiation field of the antenna. The radiation field of the antenna can be generally divided into the induction short-distance area, the radiation short-distance area (Fresnel area) and the far-field area from near to far. Define d as the effective radiation aperture length of the antenna, λ as the electromagnetic wave wavelength radiated by the antenna, R as the distance between the observation point and the antenna, and the close-range area of the antenna satisfies Since the main application scenario of the inventive algorithm is human body security inspection, it is approximately considered as close-range imaging. For an antenna with aperture length d, the spatial resolution is about The schematic diagram of the close-range imaging geometry is shown in Figure 2. In short-distance imaging algorithms, phase error correction in the range dimension is an important link in imaging, and the accuracy of the correction directly affects the performance of imaging. The range-dimension phase correction in the conventional short-range imaging algorithm is based on the assumption that the target is an ideal planar target, that is, the target point is in the same range tangent plane parallel to the XOY observation plane. According to this idea, the literature on the range-dimension phase compensation mainly includes :

(1) Yinsheng Zhang, Jungang Yang, Wei An. Single frequency two-dimensional synthetic aperture imaging system and imaging algorithm. Journal of National University of Defense Technology. 2014;

(2) David Sheen, Douglas McMakin, Thomas Hall. Near-field three-dimensional radar imaging techniques and applications. Applied Optics. 2010;

(3) Shichao Li, Chao Li, Wei Liu, et al. Study of Terahertz Superresolution Imaging Scheme With Real-Time Capability Based on Frequency Scanning Antenna. IEEE Transaction Terahertz Science and Technology, 2016.

The above-mentioned documents all use a single frequency point for two-dimensional synthetic aperture imaging, and their imaging performance is limited to the ideal situation where the default volume target has no thickness or the thickness of the volume target is negligible relative to the distance from the observation plane. Case. Therefore, it is necessary to propose a terahertz single-frequency two-dimensional synthetic aperture algorithm for volume targets with a certain thickness, which is used to suppress the degradation of imaging performance caused by the thickness of the volume target.

Contents of the invention

The present invention aims at the problem that the conventional short-distance imaging algorithm causes the decline of the imaging effect when there is a difference in the thickness of the volume target, and proposes a two-dimensional aperture imaging algorithm that suppresses the influence of the target thickness at a terahertz single frequency point. The distance tangent plane of the plane performs phase compensation for all strong scattering points of the target, and gathers the imaging results under all the distance tangent planes and then performs weighted average processing, so as to improve the negative impact of the distance dimension information difference between the strong scattering points of the volume target on imaging. . In the two-dimensional and direction-dimensional imaging system of terahertz single frequency point, the accuracy of phase compensation in the direction of target distance directly affects the imaging effect.

The technical solution of the present invention is: a two-dimensional aperture imaging algorithm that suppresses the influence of target thickness at a terahertz single frequency point, specifically including the following steps:

1) The center plane of the target is XOY, and an observation plane _X'OY ' is established parallel to and at a distance R0 from the center plane of the target, and the central point of the observation plane z= _-R0 ; the strong scattering point coordinates of the body target are (x, y, Δz), where Δz is the distance that the target point deviates from the default target distance tangent plane z=0 in the distance dimension direction; the transceiver antenna is on the observation plane of z=-R ₀ , and the coordinates of the transceiver antenna are (x', y', -R ₀ );

2) The terahertz single-frequency point signal sent by the transmitter is: s _t (t) = e ^j2πft , where f is the carrier frequency, and the echo signal received by the receiving antenna from a strong scattering point is:

Where c is the speed of light in a vacuum state, g(x, y) is the scattering coefficient of the target strong scattering point (x, y, Δz); x, y are the coordinates of the strong scattering point on the default z=0 target XOY plane, Δz is the actual deviation value of the strong scattering point at (x, y) in the distance dimension and the default value z=0; x', y' are the coordinates of the transceiver on the observation plane;

Perform down-conversion processing on the transmitted signal and received signal, eliminate the time variable t, and get:

3) Terahertz security imaging, the imaging scene is in the short range, after the step 2) the frequency-converted echo signal is decomposed by spherical waves, the echo transformed from the space domain to the wave number domain is obtained, and the echo is expressed as:

in is the wavenumber domain frequency; k _x' , _ky' , k _z' are the wavenumber domain frequency components in the X, Y, and Z directions in the wavenumber domain;

4) Assume that the Z-direction coordinates of the strong scattering point target are distributed in the interval (-△z _max , △z _max ), that is, △z∈(-△z _max , △z _max ), and use multiple distance planes to compensate the imaging independently. Then gather all imaging results for weighted average, and perform pixel-level superposition of a series of distance slice imaging results, where the phase compensation of the mth distance slice plane is:

z _m ∈ (-Δz _max ,Δz _max ), m=1,...,M; z _m is one of the tangent planes introduced for distance dimension phase compensation, and the corrected distance dimension compensation factor is

After performing phase compensation according to a series of distance tangent planes, the weighted weight value is determined by using the two-dimensional inverse Fourier transform to obtain the size relationship of the image entropy of the image, that is, the target image obtained by inversion according to the mth distance tangent plane is:

Weighted weight w=[w ₁ ,w ₂ ,...,w _M ] ^T , satisfying w _m ∈ (0,1);

Then the optimal image obtained according to M different reconstruction results and their weighted weights is:

In the step 3), the imaging scene is in the short-distance range, and at this time, the wave front form of the signal transmitted by the transmitting antenna and reaching the target plane retains the characteristics of spherical waves.

In said step 4), determine the weight according to the size relationship of the image entropy of the image after each compensation plane compensation, specifically:

It is assumed that there are N gray levels in the m-th image, and the probabilities between the various gray levels are p ₁ , p ₂ ,...,p _N , then according to Shannon's theorem, the image entropy of the image is expressed as:

The total image entropy of M images is:

The weighted weight of the corresponding image is:

The beneficial effect of the present invention is that: the two-dimensional aperture imaging algorithm for suppressing the influence of the target thickness at the terahertz single frequency point of the present invention can effectively suppress the influence of the thickness of the volume target on the imaging by using the pixel-level superposition of multiple distance slices to compensate the phase. The impact caused by the algorithm can more accurately reconstruct the distribution of scattering characteristics of strong scattering point targets.

Description of drawings

Fig. 1 is the flowchart of two-dimensional imaging algorithm of the present invention;

Fig. 2 is a schematic diagram of the spatial distribution of the body target and the observation plane of the present invention;

Fig. 3 is the geometrical schematic diagram of close-range imaging of the present invention;

Fig. 4 is the three-dimensional image figure of target point of the present invention;

Fig. 5 is the tangent plane view of the target point of the present invention along the direction;

Figure 6 is a reconstructed image without weighted average processing;

Fig. 7 is a reconstructed image processed by weighted average in the present invention.

detailed description

The present invention proposes a single-frequency point two-dimensional synthetic aperture imaging algorithm based on the terahertz frequency band for suppressing the influence caused by the thickness of the body target. As shown in Figure 1, the specific steps of the algorithm are as follows:

1. Figure 2 is a schematic diagram of the spatial distribution of volume targets and observation planes. The observation plane X'OY' is located at z=-R ₀ , the target plane is XOY, and the two planes are parallel, that is, X'OY'//XOY; the coordinates of the strong scattering point of the volume target are (x, y, Δz), Where Δz is the distance from the target point to the default target distance tangent plane z=0 in the distance dimension; the coordinates of the transmitting and receiving antennas on the observation plane z=-R ₀ are (x',y',-R ₀ ). Then the distance from the antenna to the target point is:

2. The terahertz single-frequency point signal sent by the transmitter is: s _t (t)=e ^j2πft , where f is the carrier frequency. The echo signal received by the receiving antenna from the strong scattering point is:

Where c is the speed of light in a vacuum state, g(x, y) is the scattering coefficient of the target strong scattering point (x, y, Δz); x, y are the coordinates of the strong scattering point on the default z=0 target XOY plane, Δz is the actual deviation value of the strong scattering point at (x, y) in the distance dimension and the default value z=0; x', y' are the coordinates of the transceiver on the observation plane X'OY'.

Perform down-conversion processing on the transmitted signal and received signal, eliminate the time variable t, and get:

3. The invention is mainly applied to terahertz security imaging, and its imaging scene is in a short range, so the imaging scene is in a short distance. At this time, the form of the wave front of the signal transmitted by the transmitting antenna reaching the target plane cannot be approximated as a plane wave as in the long-distance scene. The wave front retains the characteristics of a spherical wave, so it should be treated as a spherical wave (as shown in Figure 3).

A variety of current optimization inversion methods utilize the method of decomposing the spherical wave into two superimposed plane waves in mutually perpendicular directions. For the spherical integral in the above echo, after superposition and expansion of plane waves, it is mapped to the wave number domain along the X and Y directions, that is, after spherical wave decomposition, the echo transformed from the space domain to the wave number domain can be obtained. The echo is expressed as:

in is the wavenumber domain frequency; k _x' , _ky' , k _z' are the wavenumber domain frequency components in the X, Y, and Z directions in the wavenumber domain.

The frequency range of the wavenumber domain in the X, Y, and Z directions is:

Among them, L _x and L _y are the effective aperture lengths of the transceiver in the X and Y directions of the observation plane, respectively (as shown in Figure 2).

4. Compensating the phase shift in the distance dimension direction existing in the echo signal. Then, the two-dimensional Fourier inverse transform is performed, so as to realize the function of reconstructing the distribution of target scattering characteristics from the target echo signal.

In the two-dimensional and direction-dimensional imaging based on a single frequency point, the high resolution of the two-dimensional direction dimension is contributed by the effective length of the synthetic aperture, and since only a single-point frequency signal is used in imaging, its resolution in the distance dimension is limited, and it can be considered that there is no distance dimension resolution. There are coordinate deviation Δz and constant R ₀ related to the distance dimension in the echo signal. In order to eliminate its influence on imaging, the phase shift caused by them needs to be compensated. Since the information of the distance dimension cannot be obtained only by using the single-point frequency signal, the coordinate deviation Δz in the distance dimension of the actual volume target makes the phase compensation only using the default distance R ₀ between the detection plane and the target center plane cannot eliminate the existing phase Offset, will also introduce a new phase offset, causing imaging performance degradation, or even failure. This invention aims at the fact that the target to be measured usually has a certain thickness difference, which leads to the phenomenon that the distance information of the scattering point is different. In order to suppress its influence on the imaging, considering that the target has certain thickness information, the Z-direction coordinate of the strong scattering point target is set Distributed in the (-Δz _max , Δz _max ) interval, that is, Δz∈(-Δz _max , Δz _max ). Multiple distance planes are used to independently compensate for imaging, and then the weighted average of all imaging results is combined to perform pixel-level superposition of a series of distance slice imaging results for image reconstruction. The phase compensation of the mth distance tangent plane is:

z _m ∈(-Δz _max ,Δz _max ), m=1,...,M; z _m is one of the tangent planes introduced for distance dimension phase compensation. The corrected distance dimension compensation factor is

After performing phase compensation according to a series of distance tangent planes, the weighted weight value is determined by using the two-dimensional inverse Fourier transform to obtain the size relationship of the image entropy of the image, that is, the target image obtained by inversion according to the mth distance tangent plane is:

Weighted weight w=[w ₁ ,w ₂ ,...,w _M ] ^T , satisfying w _m ∈ (0,1);

The weight is determined according to the size relationship of the image entropy of the image after each compensation plane is compensated.

Assuming that the m-th image has N gray levels in total, and the probabilities between the various gray levels are p ₁ , p ₂ ,..., p _N , then according to Shannon's theorem, the image entropy of the image is expressed as follows:

The total image entropy of M images is:

The weighted weight of the corresponding image is:

According to the target image obtained by each inversion and its corresponding weighted weight value, the final image is obtained:

Compensating the phase by pixel-level superposition of multiple distance slices can effectively suppress the influence of the thickness of the volume target on the imaging algorithm. Compared with images processed without weighted mean, the distribution of scattering characteristics of strong scattering points can be reconstructed more accurately.

Fig. 6 is an image processed without weighted average, and Fig. 7 is an image processed with weighted average. After comparison, it is obvious that Figure 7 is clearer.

Claims (3)

1. the two-dimentional aperture imaging algorithm that under a kind of Terahertz single-frequency point, suppression target thickness affects is it is characterised in that concrete wrap Include following steps：

1) target's center's plane be XOY, parallel and with target's center plan range R₀Place sets up plane of vision X'OY', and observation is flat Face central point z=-R₀；The strong scattering point coordinates of body target is that (x, y, Δ z), wherein Δ z are impact points on distance dimension direction Deviate the distance apart from section z=0 for the default objects；Dual-mode antenna is in z=-R₀Plane of vision on, the seat of dual-mode antenna It is designated as (x', y' ,-R₀)；

2) the Terahertz single frequency point signal that emitter sends is：s_t(t)=e^j2πft, wherein f is carrier frequency, and reception antenna receives Echo-signal to strong scattering point is：

$s_r\left(x^{\′},y^{\′},t\right)=\∫\∫g\left(x,y\right)e^{j2\mathrm{\π}f\left(t-\frac{2R\left(x^{\′},y^{\′},R_0\right)}{c}\right)}dxdy$

Wherein c is the light velocity under vacuum state, and g (x, y) is target strong scattering point (x, y, the scattering coefficient of Δ z)；X, y are to dissipate by force Coordinate on acquiescence z=0 target XOY plane for the exit point, Δ z is (x, y) place strong scattering point in distance dimension and default value z=0 Actual deviation value；X', y' are the coordinates of transceiver on plane of vision；

Down-converted is carried out to transmission signal and receipt signal, eliminates time variable t, obtain：

$s_r(x^{\′},y^{\′})=s_r(x^{\′},y^{\′},t)e^{-j2\mathrm{\π}ft}=\∫\∫g(x,y)e^{-j2\mathrm{\π}f\left(\frac{2R(x^{\′},y^{\′},R_0)}{c}\right)}dxdy;$

3) Terahertz safety check imaging, image scene is in short range, by step 2) through frequency conversion echo-signal through ball After the Wave Decomposition of face, obtain changing to the echo of wave-number domain from transform of spatial domain, echo is expressed as：

$\begin{array}{c}{S}_{FT2}\left(k_{x^{\′}},k_{y^{\′}}\right)=\∫\∫s_r\left(x^{\′},y^{\′}\right)e^{-{\mathrm{jk}}_{x^{\′}}{x}^{\′}}{e}^{-{\mathrm{jk}}_{y^{\′}}{y}^{\′}}{\mathrm{dx}}^{\′}{\mathrm{dy}}^{\′}\\ =\∫\∫g\left(x,y\right)\left(\∫\∫e^{-{\mathrm{jk}}_{x^{\′}}x-{\mathrm{jk}}_{y^{\′}}y-{\mathrm{jk}}_{z^{\′}}\left(\mathrm{\Δ}z+R_0\right)}{\mathrm{dx}}^{\′}{\mathrm{dy}}^{\′}\right)dxdy\end{array}$

$k_{z^{\′}}=\sqrt{4k^2-k_{x^{\′}}^2-k_{y^{\′}}^2}$

WhereinFor wave-number domain frequency；k_x'、k_y'、k_z'It is the wave-number domain frequency component of X in wave-number domain, Y, Z-direction；

4) the Z-direction coordinate setting strong scattering point target is distributed in (- △ z_max,△z_max) interval interior, i.e. △ z ∈ (- △ z_max,△ z_max), using multiple apart from plane compensating image alone, then gather all imaging results and be weighted average mode, carry out one Series is superimposed apart from the Pixel-level of tangent plane imaging results, and wherein m-th phase compensation apart from section is：

$S(k_{x^{\′}},k_{y^{\′}},z_m)=S_{FT2}(k_{x^{\′}},k_{y^{\′}})e^{{\mathrm{jk}}_{z^{\′}}{R}_0}{e}^{{\mathrm{jk}}_{z^{\′}}{z}_m}$

z_m∈(-Δz_max,Δz_max), m=1 ..., M；z_mTie up the section introducing during phase compensation for one of for distance, Revised distance ties up compensating factor

According to a series of image entropies obtaining image after section carries out phase compensation using bidimensional inverse Fourier transform Magnitude relationship determining weighting weights, the target picture being obtained apart from section inverting according to m-th is：

$g_m(x,y)=\∫\∫S(k_{x^{\′}},k_{y^{\′}},z_m)e^{{\mathrm{jk}}_{x^{\′}}x}{e}^{{\mathrm{jk}}_{y^{\′}}y}{\mathrm{dk}}_{x^{\′}}{\mathrm{dk}}_{y^{\′}}$

Weighting weight w=[w₁,w₂,...,w_M]^T, meetw_m∈(0,1)；

Then obtaining optimum image according to M different reconstruction result and its weighting weights is：

$\hat{g}(x,y)=\underset{m=1}{\overset{M}{\Σ}}{w}_m{g}_m(x,y).$

2. the two-dimentional aperture imaging algorithm that under Terahertz single-frequency point, suppression target thickness affects according to claim 1, it is special Levy and be, described step 3) in image scene be in short range, now transmitting antenna transmission signal reaches objective plane Wavefront form retains the characteristic of spherical wave.

3. the two-dimentional aperture imaging algorithm that under Terahertz single-frequency point, suppression target thickness affects according to claim 2, it is special Levy and be, described step 4) in the image entropy of image after plane compensation is compensated according to each magnitude relationship, to determine power Value, specially：

Set m width image and have N kind gray scale, and the probability between various gray scale is respectively p₁,p₂,...,p_N, then according to perfume (or spice) Agriculture theorem, the image entropy of image is expressed as：

The total image entropy of M width image is：

The weighting weights of correspondence image are：