RU2413238C1 - Interference suppression method - Google Patents

Abstract

FIELD: radio engineering.

SUBSTANCE: in suppression method of interference representing passive interference (PI) with unknown parametres, from received signals and interference in N repetition periods of coherent block, where N≥2, there formed are references of complex envelope x_i,k, where i - number of repetition period of coherent block, 1≤i≤N, k - number of range element; as per references of complex envelope in window of N_w range elements enveloping k element there formed are evaluations of correlation parametres of interference, which are used to determine weight ratios of rejector filter to the input of which there supplied are references of complex envelope x_ik; at that, references of complex envelope x_i,k are also supplied to M additional rejector filters the weight ratios of which are obtained by using the evaluations of correlation parametres of interference in the appropriate M windows of the size which is less than N_w, and consisting of range elements enveloping k element; then, in window from N_p range elements enveloping k element, where N_p is chosen on the basis of allowable PI suppression reduction, there measured is interference power

at outputs of additional rejector filters, where j=1,…,M - number of additional rejector filter, and interference power

at the output of the above rejector filter; r filter is chosen from additional rejector filters, where 1≤r≤M with minimum evaluation of interference power at its output

after that, the following condition is checked:

where d≤1 - coefficient the value of which is determined with allowable reduction of suppression of PI which are homogeneous as to range, depending on the fact whether this condition is met or not; mixture of signals and suppressed interference is picked up from the output either of r additional rejector filter or the above rejector filter.

EFFECT: improving suppression of passive interference which is non-homogeneous as to range.

6 dwg

Description

The invention relates to radar and can be used to suppress interference when signals are detected in pulsed radar stations. In particular, to suppress non-uniform in range passive interference (PP).

A known method of suppressing interference ("The use of digital signal processing" edited by E. Opengheim: Mir, 1980), in which the complex envelope readings are formed from the signals received by the radar in N repetition periods of the coherent burst, the samples in each range element are processed in the notch filter with weighting factors W = (w _l , w ₂ , ..., w _N ) ^T , pre-selected in such a way as to ensure the minimum value of the interference power at the output of the notch filter. This method provides suppression of certain types of passive interference with a priori known parameters, for example, interference from stationary local objects (MP). In this case, use binomial weights found from the expression

where i is the number of the repetition period of the pack.

A significant drawback of this method is the low suppression of PP with obviously unknown parameters (Doppler frequency, correlation coefficient), including, for example, interference from weather patterns or dipoles dropped from an aircraft.

по отсчетам в N_w элементах дальности, окружающих обрабатываемый k-й элемент, определяют весовые коэффициенты режекторного фильтра W={w_l,w₂,…,w_N)^T из уравнения:The closest technical solution is a method of suppressing interference (Popov D.I. Synthesis of digital adaptive notch filters, Radio Engineering, 1981, vol. 36, No. 10), in which complex envelope readings are formed from the signals received by the radar in N repetition periods of the coherent burst , when processing samples in the kth range element, the interference covariance matrix is estimated

from the counts in N _w range elements surrounding the processed k-th element, the weight coefficients of the notch filter W = {w _l , w ₂ , ..., w _N ) ^{T are} determined from the equation:

размером N×N, обрабатывают отсчеты k-го элемента дальности в данном режекторном фильтре. Причем для некоторых видов сигналов и помех существуют достаточно простые решения уравнения (1). Например, при обработке когерентной пачки сигналов с одинаковыми периодами повторения Т в условиях ПП элементы матрицы

задаются в виде оценок where I is the identity matrix of size N × N, λ _min is the minimum eigenvalue of the matrix

size N × N, process the samples of the k-th element of the range in this notch filter. Moreover, for some types of signals and interference, there are fairly simple solutions to equation (1). For example, when processing a coherent packet of signals with the same repetition periods T in the conditions of PP, the elements of the matrix

are given in the form of estimates

- оценка межпериодной фазы ПП,

- оценка коэффициента корреляции ПП между n-ым и m-ым периодами, где 1≤n, m≤N. При этом весовые коэффициенты режекторного фильтра

а значения ν_i следующим образом выражаются через оценки

при разных N, где N≥2:Where

- assessment of the interperiod phase of PP,

- assessment of the correlation coefficient of the PP between the n-th and m-th periods, where 1≤n, m≤N. At the same time, the notch filter weights

and ν _i values are expressed as follows in terms of estimates

for different N, where N≥2:

1. ν _i = 1, ν ₂ = -ρ _1,2 , for N = 2;

2. ν _l = ν ₃ = 1, ν ₂ = -2ρ _1,2 , for N = 3;

3. ν ₁ = -ν ₄ = 1,

4. ν ₁ = ν ₅ = 1,

For PPs encountered in most situations with a unimodal Gaussian spectrum, the values of ν _i are expressed in the following even simpler way through only estimating the correlation coefficient between the 1st and 2nd periods of the coherent burst:

1. ν _i = 1, ν ₂ = -ρ _1,2 , for N = 2;

2. ν _l = ν ₃ = 1, ν ₂ = -2ρ _1,2 , for N = 3;

_1,2+

), для N=4;3. ν _l = -ν ₄ = 1, ν ₂ = -ν ₃ = - (1+

_1.2 +

), for N = 4;

4. ν _{1 =} ν ₅ = 1,

In this method, suppression of various types of interference is provided, including PP with known unknown parameters. In this case, the minimum value of the interference power at the output of the notch filter is achieved.

(соответственно, оценки доплеровской фазы и коэффициента корреляции), неверному определению весовых коэффициентов The main disadvantage of this method is to reduce the suppression of non-uniform range PPs, when PPs with different correlation parameters fall into the window of N _w range elements surrounding the kth processed element. This leads to errors in the estimation of the correlation parameters of PP

(respectively, estimates of the Doppler phase and correlation coefficient), incorrect determination of weight coefficients

and, as a consequence, an increase in interference power at the output of the notch filter.

The technical result (the problem being solved) of the invention, therefore, is to eliminate the aforementioned drawback, namely, to increase the suppression of PP inhomogeneous in range.

на выходах дополнительных режекторных фильтров, где j=1, …, M - номер дополнительного режекторного фильтра, и мощность помех

на выходе упомянутого режекторного фильтра, из дополнительных режекторных фильтров выбирают r-й, где 1≤r≤M, с минимальной оценкой мощности помех на его выходе

после чего проверяют следующее условие:The technical result (the problem to be solved) in the proposed method is achieved by the fact that in the known method of suppressing interference, which is passive interference (PP) with unknown parameters, in which of the received radar signals and interference in N repetition periods of the coherent packet, where N≥2, form the samples of the complex envelope x _{i, k} , where i is the number of the repetition period of the coherent burst, 1≤i≤N, k is the number of the range element, according to the samples of the complex envelope in the window of N _w range elements surrounding the kth element, form estimates correl tional parameters interference, are used to determine weighting coefficients of the notch filter, the input of which is fed the samples of the complex envelope x _{i, k,} according to the invention the samples of the complex envelope x _{i, k} are also fed to the M additional bandstop filter, the weighting coefficients are obtained from the estimation of the correlation interference parameters in the respective windows M smaller than N _w, size, consisting of a range of elements surrounding the k-th element, then in the range of N _p elements surrounding a k-th element, where N _p ybiraetsya based on the acceptable reduction PP suppression, measured interference power

at the outputs of the additional notch filters, where j = 1, ..., M is the number of the additional notch filter, and the interference power

at the output of the said notch filter, from the additional notch filters, choose the rth, where 1≤r≤M, with a minimum estimate of the interference power at its output

then check the following condition:

where d≤1 - coefficient, the value of which is determined by the allowable reduction of suppression uniform in range PP,

depending on whether this condition is fulfilled or not, the mixture of signals and suppressed interference is removed from the output of either the r-th additional notch filter or the mentioned notch filter.

New significant features of the proposed method are the following:

- samples of the complex envelope x _ik are fed to M additional notch filters, the weight coefficients of which are obtained using estimates of the correlation parameters of interference in the corresponding M windows of different and smaller than N _w sizes consisting of range elements surrounding the kth element;

на выходах дополнительных режекторных фильтров, где j=1,…, M - номер дополнительного режекторного фильтра, и мощность помех

на выходе упомянутого режекторного фильтра;- in the window of N range elements surrounding the k-th element, where N _p is selected based on the allowable reduction in the suppression of the PP, the interference power is measured

at the outputs of the additional notch filters, where j = 1, ..., M is the number of the additional notch filter, and the interference power

at the output of said notch filter;

- r-th is selected from additional notch filters, where 1≤r≤M is the minimum estimate of the interference power at its output

- check the condition (2), in which d≤1 is a coefficient, the value of which is determined by the allowable decrease in the suppression of uniform in range PP;

- depending on whether or not this condition is fulfilled, the mixture of signals and suppressed interference is removed from the output of either the rth additional notch filter or the mentioned notch filter.

The use of all new features in conjunction with the features of the prototype will increase the suppression of heterogeneous in the range of PP. Since the nonuniformity of the PP in the window causes an increase in the interference power at the output of the notch filter, the weight coefficients of which are determined by the estimates of the correlation parameters of the PP in this window, condition (2) is fulfilled when the PP is not uniform in range. An additional notch filter, the weight coefficients of which are determined by estimates of the correlation parameters of the PP in the window, the size of which most closely matches the size of the homogeneous region of the PP, provides maximum suppression of the PP and, therefore, the minimum value of the output interference power

The invention is illustrated by drawings.

Figure 1. The structural diagram of a device that implements the proposed method of suppressing interference.

Figure 2. The structural diagram of a device that implements the adaptive notch filter in figure 1.

Figure 3. Block diagram of a device that implements a power meter in figure 1.

Figure 4. Block diagram of a device that implements a channel determination unit with a minimum value in FIG. 1.

Figure 5. Dependence of the power of the threshold signal on the number of the element in the window (inhomogeneous PP).

6. Dependences of the threshold signal power on the element number in the window (inhomogeneous MP).

Technical implementation of the proposed method of suppressing interference is possible based on the device shown in figure 1. The device comprises a complex envelope sampler 1 whose input is the device input, M + 1 adaptive notch filters 2, M + 1 power meters 3, a multiplier 4, a channel determination unit with a minimum value of 5, a comparison circuit 6, a first multiplexer 7 and a second multiplexer 7, the output of which is the output of the device. In this case, the adaptive notch filter 2 performs both the function of evaluating the correlation parameters of the PP in the window, the size of which is set by the information at its second input, and the functions of the notch filter itself, the weight coefficients of which vary (are tuned) depending on the obtained estimates of the correlation parameters of the PP.

The technical implementation of the adaptive notch filter 2 in figure 1 is possible in the form of the device shown in figure 2. The device contains a first multiplier 4, a delay unit for a period of 9, the inputs of which are connected to the first input of the correlation coefficient estimation unit PP 10 and are the first input of the device, the second input of the correlation coefficient estimation unit PP 10 is the second input of the device, register 11, the second multiplier 4, arithmetic unit for calculating the weight coefficients of the notch filter 12, tunable notch filter 8, the output of which is the output of the device.

The technical implementation of the power meter 3 in FIG. 1 is possible in the form of the device shown in FIG. 3. The device contains a square computer module 13, the input of which is the input of the device, (N _p +1) -tap register 14, adder 15, divider 16, the output of which is the output of the device.

The technical implementation of the channel determination unit with a minimum value of 5 in FIG. 1 is possible in the form of the device shown in FIG. 4. The device contains a circuit for selecting the minimum value 17, the M + 1 inputs of which are the inputs of the device, M + 1 comparison circuits 6, and the encoder 18, the output of which is the output of the device.

полученные на выходах 2,…, M+1-го измерителей мощности 3, подаются на входы блока определения канала с минимальным значением 5. Минимальная из оценок мощности помех

с первого выхода блока определения канала с минимальным значением 5 подается на второй вход схемы сравнения 6. Оценка мощности помех

с выхода 1-го измерителя мощности 3 подается на первый вход умножителя 4, после умножения в котором на постоянный коэффициент d, подающийся на второй вход умножителя 4, величина

поступает на первый вход схемы сравнения 6. При выполнении условия

единичный сигнал с выхода схемы сравнения 6 поступает на управляющий вход первого мультиплексора 7 и разрешает прохождение на его выход сигнала r с его второго информационного входа, поступающего со второго выхода блока определения канала с минимальным значением 5. Если упомянутое выше условие не выполняется, то есть

то на выходе схемы сравнения 6 формируется нулевой сигнал, который, поступая на управляющий вход первого мультиплексора 7, разрешает прохождение на его выход сигнала 0 с его первого информационного входа. Таким образом, с выхода первого мультиплексора 7 на управляющий вход второго мультиплексора 7 подается либо сигнал, равный номеру r, где 1≤r≤M, дополнительного адаптивного режекторного фильтра 2, либо 0-й сигнал при выборе 1-го адаптивного режекторного фильтра 2. По данному управляющему сигналу на выход устройства передаются отсчеты, взятые с выхода r-го дополнительного адаптивного режекторного фильтра 2 u_k,r или 1-го адаптивного режекторного фильтра 2 y_k. The device in figure 1 works as follows. The complex envelope 1 sampler receives x (t) signals from the output of the radar receiver. From the output of the complex envelope sampler 1, the complex envelope samples x _{i, k} are sent to M + 1 adaptive notch filters 2, and the weighting coefficients of the first adaptive notch filter 2 are determined by estimating the correlation parameters of interference in a window of N _w range elements surrounding the kth element. The weights of each of j = 1, ..., M of additional adaptive notch filters 2 are determined by estimating the correlation parameters of the interference in the windows of the corresponding smaller than N _w size L _w (j) surrounding the k-th element of the range. The samples at _k from the output of the first adaptive notch filter 2 and the samples u _{k, j} from the outputs of M of the additional adaptive notch filter 2 are sent to M + 1 power meters 3 and the information inputs of the second multiplexer 7. Estimates of the interference power at outputs 1, ..., M- th additional adaptive notch filters 2

obtained at the outputs 2, ..., M + of the 1st power meter 3 are fed to the inputs of the channel definition block with a minimum value of 5. The minimum of the interference power estimates

from the first output of the channel definition block with a minimum value of 5 is fed to the second input of the comparison circuit 6. Assessment of interference power

from the output of the 1st power meter 3 is fed to the first input of the multiplier 4, after multiplication in which by a constant coefficient d, fed to the second input of the multiplier 4, the value

arrives at the first input of the comparison circuit 6. When the condition

a single signal from the output of the comparison circuit 6 is supplied to the control input of the first multiplexer 7 and allows the signal r to pass to its output from its second information input coming from the second output of the channel determination unit with a minimum value of 5. If the above condition is not satisfied, that is

then at the output of the comparison circuit 6 a zero signal is formed, which, entering the control input of the first multiplexer 7, allows the signal 0 to pass from its first information input to its output. Thus, from the output of the first multiplexer 7, either a signal equal to the number r, where 1≤r≤M, of the additional adaptive notch filter 2, or the 0th signal when choosing the 1st adaptive notch filter 2 is supplied to the control input of the second multiplexer 7. According to this control signal, samples taken from the output of the rth additional adaptive notch filter 2 u _{k, r} or the 1st adaptive notch filter 2 y _k are transmitted to the device output _.

The complex envelope sampler 1 is a device known, for example, from (L.D. Shirman, V.N. Manzhos. Theory and technique of processing radar information against a background of interference. M: Radio and communications, 1981, 416 pp.) .

Оценка фазового вектора ПП

с первого выхода блока оценки коэффициента корреляции ПП 10 подается на первый вход второго умножителя 4, на второй вход которого с выхода регистра 11 поступает хранящийся в нем вектор поворота отсчетов по фазе в p-м периоде повторения

когерентной пачки

С выхода второго умножителя 4 новое значение вектора поворота отсчетов по фазе

записывается в регистр 11 для использования в следующем (p+1)-м периоде повторения когерентной пачки. Вектор поворота отсчетов по фазе в р-м периоде повторения когерентной пачки

с выхода регистра 11 подается на второй вход первого умножителя 4, с выхода которого отсчеты

с нулевой после поворота доплеровской фазой ПП подаются на информационный вход перестраиваемого режекторного фильтра 8. Оценка модуля коэффициента корреляции ПП

со второго выхода блока оценки коэффициента корреляции ПП 10 подается на вход арифметического блока расчета весовых коэффициентов режекторного фильтра 12, с выхода которого весовые коэффициенты режекторного фильтра ν₁, ν₂,…, ν_N подаются на управляющий вход перестраиваемого режекторного фильтра 8. В арифметическом блоке расчета весовых коэффициентов режекторного фильтра 12 в зависимости от N используются приведенные в таблице 2 расчетные формулы. В перестраиваемом режекторном фильтре 8 происходит оптимальное подавление ПП с нулевой доплеровской фазой, гауссовым спектром и модулем коэффициента корреляции

Смесь сигналов и подавленных помех с выхода перестраиваемого режекторного фильтра 8 поступает на выход данного устройства.The device in figure 2 works as follows. In the rth repetition period of the coherent burst, the samples x _pk from the output of the complex envelope shaper 1 are fed to the first input of the correlation coefficient estimator 10; the input of the delay unit for period 9 and the first input of the first multiplier 4. _p-1.k from the output of the delay unit for a period of 9 at the same time go to the third input of the correlation coefficient estimation unit of software 10, in which, in accordance with the information about the size of the window supplied to its second input, the coefficient is evaluated in this range window t correlation of PP between the 7th and 2nd repetition periods of a coherent burst

PP phase vector estimation

from the first output of the correlation coefficient estimation block, the PP 10 is fed to the first input of the second multiplier 4, the second input of which from the output of register 11 receives the phase rotation vector of the samples stored in it in the pth repetition period

coherent pack

From the output of the second multiplier 4, the new value of the phase rotation vector of samples

written to register 11 for use in the next (p + 1) -th repetition period of the coherent burst. Phase rotation vector of the samples in the rth repetition period of the coherent burst

from the output of the register 11 is fed to the second input of the first multiplier 4, the output of which is counted

with zero after rotation, the Doppler phase of the PP are fed to the information input of the tunable notch filter 8. Evaluation of the module of the correlation coefficient of PP

from the second output of the correlation coefficient estimator, the PP 10 is fed to the input of the arithmetic unit for calculating the weight coefficients of the notch filter 12, from the output of which the weight coefficients of the notch filter ν ₁ , ν ₂ , ..., ν _N are fed to the control input of the tunable notch filter 8. In the arithmetic unit calculation of the weight coefficients of the notch filter 12, depending on N, the calculation formulas given in table 2 are used. In tunable notch filter 8 there is an optimal suppression of PP with zero Doppler phase, a Gaussian spectrum and the module of the correlation coefficient

A mixture of signals and suppressed interference from the output of the tunable notch filter 8 is fed to the output of this device.

In the block for estimating the correlation coefficient PP 10, known, for example, from (D.I. Popov. Synthesis of Digital Adaptive Notch Filters, Radio Engineering, 1981, vol. 36, No. 10), the estimation of the correlation coefficient between the samples of the complex envelope j- of the 2nd and 2nd periods of the repetition packet of the coherent packet according to the following formula for the j-th additional adaptive notch filter 2 in FIG. 1:

Перестраиваемый режекторный фильтр 8 аналогичен известному режекторному фильтру с постоянными весовыми коэффициентами, например, из (Л.Рабинер, Б.Гоулд. Теория и применение цифровой обработки сигналов. М.: Мир, 1978), за исключением наличия управляющего входа, по которому передаются значения используемых в данный момент времени весовых коэффициентов трансверсального фильтра.

The tunable notch filter 8 is similar to the well-known notch filter with constant weighting factors, for example, from (L. Rabiner, B. Gould. Theory and application of digital signal processing. M .: Mir, 1978), except for the presence of a control input through which values are transmitted currently used transverse filter weights.

отводов которого сигналы подаются на сумматор 15 и, далее на первый вход делителя 16, на второй вход которого подается коэффициент N_p. Величина среднего значения квадратов модулей сигналов с выхода делителя 16 подается на выход данного устройства.The device in figure 3 works as follows. The square calculator module 13 receives signals from the output of the adaptive notch filter 2. Then, from the output of the square calculator module 13, the square values of the input sample module are recorded in (N _p +1) - tap register 14, from 1, ...,

the taps of which the signals are fed to the adder 15 and then to the first input of the divider 16, to the second input of which a coefficient N _p is supplied _. The average value of the squares of the signal modules from the output of the divider 16 is fed to the output of this device.

, j=1, …, М с выхода измерителей мощности 3 поступают на 1, …, М входы схемы выбора минимального значения 17 и первые входы M схем сравнения 6. Минимальная оценка мощности помех

с выхода схемы выбора минимального значения 17 подается на вторые входы M схем сравнения 6. На выходах схем сравнения 6 единичный уровень появляется при равенстве входных сигналов, то есть при

в противном случае формируется нулевой уровень. С выходов M схем сравнения 6 логические сигналы подаются на 1,2,…, M входы шифратора 18, на выходе которого формируется число, равное номеру входа с единичным логическим уровнем, в данном случае это номер r.The device in figure 4 works as follows. Estimates of interference power

, j = 1, ..., M from the output of the power meters 3 go to 1, ..., M the inputs of the minimum value selection circuit 17 and the first inputs of the M comparison circuits 6. The minimum estimate of the interference power

from the output of the minimum value selection circuit 17, it is supplied to the second inputs of the comparison circuits 6. At the outputs of the comparison circuits 6, the unit level appears when the input signals are equal, that is, when

otherwise, a zero level is formed. From the outputs of the M comparison circuits 6, the logical signals are fed to the 1,2, ..., M inputs of the encoder 18, the output of which forms a number equal to the input number with a single logical level, in this case it is the number r.

The calculation of the detection characteristics of the prototype algorithm and the invention was carried out by the method of statistical modeling in the environment of MatLab ν.7.1. The number of trials was 10 ³ . Algorithm parameters: N = 4; N _w = 56, L _w (l) = 8; the number of additional notch filters M = 1. The probability of false alarms P _f = 10 ^-6 . The inhomogeneity of the PP in the window was set by the Butterworth function with parameters: duration τ = 8, 16; the smoothness parameter q = 2, 8. In this case, the heterogeneity was set in the center of the large window, and the detection characteristics were calculated in all elements of the window. N _p = 8 is the size of the window in which the power of the residual interference is estimated.

The following interference situations were considered:

- a mixture of interference from homogeneous local objects (MP) with a Gaussian or fractional rational spectrum with a width Δf _m = 10 Hz, noise with a level of P _n = -40 dB relative to MP and inhomogeneous PPs with a Gaussian or fractional rational spectrum with a width Δf _m = 50 Hz and the offset of the Doppler frequency f ₀ = 0 ... 500 Hz;

- a mixture of interference from homogeneous SPs with a Gaussian or fractionally rational spectrum with a width of Δf _m = 50 Hz and a Doppler frequency offset f ₀ = 0 ... 500 Hz, noise level P _n = -40 dB relative to SP and interference from inhomogeneous MPs with Gaussian or fractional -rational spectrum with a width Δf _m = 10 Hz.

Separate results are shown in Fig. 5, Fig. 6 in the form of dependences of the threshold signal power P _sd (dB) on the element number in the window. Obviously, when using the invention, there is a significant reduction in the area of the “shadow”, which is caused by inhomogeneous interference (due to the transient in the channel selection scheme with the minimum residual power, the region of the “shadow” is ≈τ + 2 · N _p elements). In this case, there is a significant gain in the magnitude of the threshold signal in comparison with the prototype, which indicates an increase in the suppression of nonuniform in range PP in the proposed method.

Claims (1)

A method for suppressing interference, which is passive interference (IF) with unknown parameters, in which of the signals received by the radar and interference in N repetition periods of the coherent burst, where N≥2, the samples of the complex envelope x _{i, k are formed} , where i is the number of the repetition period coherent packs, 1≤i≤N, k - distance element number of counts in the window of the complex envelope of N _w elements range surrounding a k-th element, form interference correlation estimation parameters are used to determine weighting coefficients rezhektornog a filter whose input is fed the samples of the complex envelope x _{i, k,} characterized in that the samples of the complex envelope x _{i, k} are also fed to the M further notch filter, the weighting coefficients are obtained from the estimation of the correlation noise parameters in the respective M windows smaller than N _w , of the size consisting of range elements surrounding the k-th element, then in the window of N _p range elements surrounding the k-th element, where N _p is selected based on the allowable reduction in the suppression of PP, the interference power is measured

at the outputs of the additional notch filters, where j = 1, ..., M is the number of the additional notch filter, and the interference power

at the output of the said notch filter, from the additional notch filters, choose the rth, where 1≤r≤M, with a minimum estimate of the interference power at its output

then check the following condition:

where d≤1 - coefficient, the value of which is determined by the allowable reduction of suppression uniform in range PP,
depending on whether this condition is fulfilled or not, the output signal is a mixture of signals and suppressed interference from the output of either the r-th additional notch filter or the mentioned notch filter.

Images