RU2612193C1 - Method of forming objects images in double-channel radiometric system - Google Patents

Abstract

FIELD: radars.

SUBSTANCE: invention relates to passive radar scanning systems using a double-channel scanning radiometer operating in the millimeter wavelength range and can be used in optical systems of the infrared range. Images forming method means a different scanning order of the antennae by angular coordinates with further combined processing the obtained in two measurement channels two measurement matrices, as the result of which formed is a matrix of objects images with a higher resolution by angular coordinates.

EFFECT: technical result is aimed at improvement of accuracy of recovery and resolution of objects images in a double-channel radiometric system operating with a higher scanning pitch by the elevation angle.

1 cl, 1 tbl

Description

The invention relates to passive systems for radio surveillance of objects [1, 2] using a two-channel scanning radiometer operating in the millimeter wavelength range (3-8 mm) with an antenna radiation pattern width of 1-3 °. Each channel is an antenna that receives the emitted signal in a certain wavelength range, and a primary processing path, including a high-frequency amplifier, a quadratic detector, a low-pass filter (low-pass filter), an analog-to-frequency converter, and a recording device that digitally stores a matrix (frame) images of objects.

,

, элементы дискретизации которого x(θ_i,ϕ_j) имеют смысл интенсивности излучения в i,j-м направлении и рассматриваются в системе угловых координат наблюдателя: θ_i - по углу места и ϕ_j - по азимуту. Число N определяет размер поля X по строке и столбцу. Антенная система построчно сканирует участок местности, смещаясь по азимуту непрерывно с определенной скоростью, зависящей от времени накопления сигнала в ФНЧ, и по углу места дискретно путем механического переключения. При каждом θ_i,ϕ_j-м угловом положении антенны принимаемая часть поля X усиливается радиометром и после ФНЧ регистрируется в виде напряжения y_k(θ_i,ϕ_j)=y_k(i,j) в каждом k-м канале:

, K - число каналов. Величины y_k(i,j) носят интегральный (суммарный) характер и на множестве значений i,j подчинены модели измерений вида:Distant objects emit a field X = {x (θ _i , ϕ _j )},

,

whose sampling elements x (θ _i , ϕ _j ) have the meaning of radiation intensity in the i, jth direction and are considered in the observer's angular coordinate system: θ _i - in elevation and ϕ _j - in azimuth. The number N determines the size of the field X by row and column. The antenna system line-by-line scans a site, shifting in azimuth continuously at a certain speed, depending on the time of signal accumulation in the low-pass filter, and discretely in elevation angle by mechanical switching. For each θ _i , ϕ _j angular position of the antenna, the received part of the field X is amplified by a radiometer and after the low-pass filter is recorded as the voltage y _k (θ _i , ϕ _j ) = y _k (i, j) in each kth channel:

, K is the number of channels. The quantities y _k (i, j) are integral (total) in nature and, on the set of values i, j, the measurement models of the form are subordinate:

where 2n + 1 is the width of the antenna pattern (BOTTOM) in elevation (in i) and azimuth (in j) at 0.5 power in the number of i, j-th sampling elements; α _k (i, j) - normalized values of DND in the k-th channel; p _k (i, j) are the noise of the equipment in the form of white noise.

The set Y _k = {y _k (i, j)} forms the image matrix in the expanded field of view corresponding to the size of the matrix X at the output of the k-th channel. If there is one channel, the task is to increase the resolution of the radiometric image Y ₁ in azimuth and elevation due to the restoration of the unobservable field X = {x (i, j)} based on observations (1) at K = 1 and is solved by well-known image restoration methods , for example [3, 4].

There is a method of image formation in multichannel radar (RTLS) and radar (radar) systems [5], which we will consider as a prototype. The method as applied to a two-channel (K = 2) radiometric system and measurement model (1) for Y ₁ and Y ₂ is as follows:

1. Two antennas with different DND characteristics scan the line of sight line by line, shifting in azimuth (in j) and elevation angle (in i) with a small step equal to the sampling step.

2. The primary processing system of the received signals measures in each k-th channel (k = 1, 2) the signals at discrete time instants that coincide with the i, j-th sampling steps at angles that make up the (2n + 1) th part of the bottom width , and forms from them two image matrices Y ₁ and Y ₂ .

.3. The resulting matrices Y ₁ and Y _{2 are} sequentially and row-wise folded into one measurement vector

.

умножается справа на матрицу весовых коэффициентов Н, вычисляемую заранее по методу, представленному в расчетной части заявки, тем самым получается вектор оценок

.4. Vector

it is multiplied on the right by the matrix of weight coefficients H, calculated in advance by the method presented in the calculation part of the application, thereby obtaining a vector of estimates

.

разворачивается построчно в матрицу X, представляющую восстановленное изображение в зоне обзора с повышенным разрешением по угловым координатам.5. Vector ratings

expands line by line into matrix X, which represents the reconstructed image in the field of view with increased resolution in angular coordinates.

This method has the following disadvantage. The azimuthal scanning speed is determined by the accumulation time of the signal in the low-pass filter, which is from 0.1 s to 1 s [2]. Therefore, line-by-line scanning of the viewing area, for example, with angular dimensions of 30 ° × 30 ° at a small sampling step, takes tens of minutes, which is unacceptable when observing moving objects. To reduce the observation time (image frame formation), the scanning step is increased in elevation (several times as compared to the sampling step). However, this reduces the accuracy of image recovery and, accordingly, resolution.

The technical result is aimed at eliminating this drawback, namely, improving the accuracy of the restoration and resolution of the image in a two-channel radiometric system, working with an increased scanning step in elevation.

, который умножают справа на матрицу весовых коэффициентов Н, вычисляемую заранее, тем самым получают вектор оценок

, затем переписывают вектор

построчно в матрицу X, которая представляет восстановленное изображение объектов в зоне обзора с повышенным разрешением по угловым координатам.The technical result of the proposed technical solution is achieved by the fact that in the method of imaging objects in a two-channel radiometric system, which consists in the fact that when observing distant objects using two scanning antennas, the line of sight of the first antenna is shifted in azimuth (j) in each i-th line the magnitude of the sampling step and elevation angle (i) by an amount several times greater than the sampling step is measured at each i, jth position of the antenna in the first measuring channel, the amplitudes at volume signals and form from them an image matrix Y ₁ , which is further processed, according to the invention at the same time use a second scanning antenna, the line of sight of which is shifted in elevation (in i) in each j-th column by the value of the sampling step and in azimuth (in j ) by an amount also several times greater than the sampling step, for each i, jth position of the antenna in the second measuring channel, the amplitudes of the receiving signals are measured and an image matrix Y _{2 is} formed from them, which is further processed together with triceet Y ₁ , while the elements of the obtained measurement matrices Y ₁ and Y _{2 are} copied sequentially and line by line into one measurement vector

, which is multiplied on the right by the matrix of weights H calculated in advance, thereby obtaining a vector of estimates

then rewrite the vector

line-by-line into matrix X, which represents a reconstructed image of objects in the viewing area with increased resolution in angular coordinates.

The method is as follows.

1. Two antennas simultaneously scan a field of view of size N × N sampling elements in azimuth and elevation. The first antenna moves continuously along the line (in azimuth) with data acquisition with a small sampling step h and moves to another line with an increased step mh, where m is an integer that takes values from 1 to n, that is, mh is at most half the width DNA in 2n + 1 bins. The second antenna, on the contrary, moves continuously along the column (in elevation) with the data being taken with a small sampling step h and moves to another column also with an increased step mh.

,

и Y₂={y₂(i,j)},

,

. Пропущенные при сканировании строки или столбцы в матрицах Y₁ и Y₂ не рассматриваются.2. The signals from the first and second antennas simultaneously pass the primary processing paths in two measuring channels, and two measurement matrices are formed from the scan results: Y ₁ = {y ₁ (i, j)},

,

and Y ₂ = {y ₂ (i, j)},

,

. Rows or columns that were missed during scanning in the matrices Y ₁ and Y _{2 are} not considered.

размером в 2(N-2n)² строк.3. Elements of the matrices Y ₁ and Y ₂ sequentially (first Y ₁ , then Y ₂ ) and are written row by row into one column vector

2 (N-2n) ² lines in size.

умножается справа на матрицу весовых коэффициентов Н, размером в N² строк и 2(N-2n)(N-2n) столбцов, вычисленную заранее по определенному правилу, изложенному в расчетной части заявки. В результате умножения получается вектор

размером в N² строк.4. Vector

multiplied on the right by the matrix of weight coefficients H, the size of N ² rows and 2 (N-2n) (N-2n) columns, calculated in advance according to a certain rule set forth in the calculation part of the application. As a result of multiplication, we get a vector

size N ² lines.

переписываются построчно в матрицу X={x(i,j)},

,

, которая представляет восстановленное изображение объектов в зоне обзора с повышенным разрешением по угловым координатам.5. Vector elements

correspond row by row to the matrix X = {x (i, j)},

,

, which represents a reconstructed image of objects in the viewing area with a high resolution in angular coordinates.

Settlement part

When scanning the field of view with two antennas, the measurement model (1) takes the form of the following system:

which can be represented in vector-matrix form:

- вектор измерений;

- вектор искомого изображения;

- вектор помех; A={a(i,j)}-N²x2(N-2n)² - матрица значений ДНА, элементы которой a(i,j) получены из α(i,j) по определенному правилу в соответствии с (2). Ниже показан пример (первый столбец и затем второй столбец программы) заполнения предварительно обнуленной матрицы A={a(i,j)} значениями alfa(i,j)=α{i-n-1,j-n-1),

,

на языке Matlab, где N₁ - номер последних строки и столбца искомой матрицы X, участвующих в образовании Y₁ и Y₂ (N₁=N при m=1 и N₁≤N при m>1):Where

- vector of measurements;

- vector of the desired image;

- interference vector; A = { a (i, j)} - N ² x2 (N-2n) ² is the matrix of DND values, the elements of which a (i, j) are obtained from α (i, j) according to a certain rule in accordance with (2) . An example is shown below (the first column and then the second column of the program) filling the previously zeroed matrix A = { a (i, j)} with the values alfa (i, j) = α {in-1, jn-1),

,

in Matlab, where N ₁ is the number of the last row and column of the desired matrix X involved in the formation of Y ₁ and Y ₂ (N ₁ = N for m = 1 and N ₁ ≤N for m> 1):

подчиняем критерию минимума квадрата евклидовой нормы:In accordance with the least squares method (least squares) search estimate

subject to the criterion of the minimum square of the Euclidean norm:

:where "T" is the transpose symbol. From the necessary condition for the existence of an extremum of functional (3), we find OLS estimates

:

where E is the identity matrix; δ> 0 is a small regularization parameter necessary for stable inversion of the matrix A ^T A. The matrix A ⁺ in (4) is pseudoinverse for A and can also be found by the singular decomposition of A, for example, in the Matlab medium: A ⁺ = pinv (A, δ).

Simulation results

The table shows the data of computer simulation of the proposed method. The width of the BOTTOM was 2n + 1 = 7 sampling elements, the signal-to-noise ratio (S-N) 30 and 50 with a maximum amplitude of 5, the scanning step in the number of sampling elements m = 1, 2, 3, the size of the observed object was 5 × 5 = 25 elements of discretization. DND was set by an exponential dependence with a quadratic exponent. In addition, noise effects on the reconstructed image were removed by a small threshold. The cells of the table give estimates of the standard deviation (RMS) of the reconstruction error obtained by comparing the simulated and reconstructed images.

The first line of values of the standard deviation shows the results of the proposed method. The second row of RMS values for comparison shows the results obtained under similar modeling conditions when using only one antenna (matrix Y _{1 was} to be processed).

It can be seen that the use of two simultaneously scanning antennas with different characteristics increases the accuracy of restoration compared to a single antenna, which leads to an increase in spatial resolution. At the same time, an increase in accuracy compensates for a decrease in accuracy due to the increased scanning step. Increased m times the scan step, in turn, increases m times the speed of formation of the image matrix, which shows the advantage of the proposed method compared to the prototype.

The proposed method may find application in existing microwave radiometric systems [6], as well as in infrared optical systems designed to detect and recognize objects from their reconstructed image.

Literature

1. Nikolaev A.G., Pertsov S.V. Radiolocation (passive radar). - M .: Owls. Radio, 1964.335 s.

2. Sharkov EA Radiothermal remote sensing of the Earth: physical foundations: in 2 tons / T. 1. M.: IKI RAS, 2014.554 p.

3. Vasilenko G.I., Taratin A.M. Image recovery. - M .: Radio and communications, 1986. 304 p.

4. Pirogov Yu.A., Timanovsky A.L. Superresolution in systems of passive radio-vision of millimeter range / Radio engineering, 2006. No. 3. S. 14-19.

5. Patent RU 2368917 C1. The method of image formation in multichannel RTLS and radar / V.K. Shit. IPC: G01S 13/89. Priority 12/21/2007. Published: 09/27/2009. Bull. Number 27.

6. Passive radar: methods for detecting objects / Ed. R.P. Bystrova and A.V. Sokolova. - M .: Radio engineering. 2008.320 s.

Claims (1)

The method of imaging objects in a two-channel radiometric system, which consists in the fact that when observing distant objects using two scanning antennas, the line of sight of the first antenna is shifted in azimuth (j) in each i-th line by the value of the sampling step and elevation angle (by i) by an amount several times greater than the sampling step, for each i, jth position of the antenna in the first measuring channel, the amplitudes of the receiving signals are measured and an image matrix Y _{1 is} formed from them, which is further processed UT, characterized in that at the same time they use a second scanning antenna, the line of sight of which is shifted in elevation (in i) in each jth column by the value of the sampling step and in azimuth (in j) by an amount also several times greater than the sampling step measured at every i, j-th position of the antenna in the second measuring channel values of the amplitudes of the reception signals and generating one image matrix Y _2, which is further treated together with a matrix of Y _1, with the elements of measurements obtained matrices Y ₁ and Y ₂ rewrite after ovatelno row and one measurement vector

, which is multiplied on the right by the matrix of weights H calculated in advance, thereby obtaining a vector of estimates

then rewrite the vector

line-by-line into matrix X, which represents a reconstructed image of objects in the viewing area with increased resolution in angular coordinates.