JP2678072B2 - Radar signal processing method and apparatus - Google Patents

Description

TECHNICAL FIELD The present invention relates to a radar signal processing method and apparatus for suppressing a sea surface reflection signal and detecting a target object signal of a ship or the like.

[Prior Art] Conventionally, Log-CFAR is known as a signal processing method for suppressing a clutter signal in a marine radar.

In Log-CFAR, the clutter signal is a Rayleigh distribution signal (Ray
This is a method of keeping a false alarm at a constant value by utilizing the property that the deviation around the average value of the log amp output signal of the Rayleigh signal becomes constant when it is a leigh distribution signal).

That is, as shown in FIG. 8, the clutter signal W is input to the Log amplifier 100 to obtain the output signal Y, the average value <Y> is calculated from the Log amplifier output signal Y by the averaging circuit 200, and finally, Difference circuit 300 averages Log Amp output signal Y
By subtracting Y>, the CFAR signal V with a constant false alarm can be obtained.

[Problems to be Solved by the Invention] However, it has been confirmed that the amplitude distribution characteristic of the sea surface reflection signal, which is one of the clutter signals, does not have a Rayleigh distribution but a Weibull distribution. Applying the Log-CFAR method for Rayleigh distribution signals shown in Fig. 8 to sea surface reflection signals,
Due to the difference in the statistical properties of the two, the deviation around the average value <Y> is not constant, the probability of false alarm increases, and the remaining sea surface reflection remains.

In particular, it is difficult for a ship radar to distinguish between a sea surface reflection signal and a ship image, which may lead to a very dangerous situation in ship navigation.

The present invention has been made in view of the above conventional problems, and calculates a signal having a constant false alarm probability with respect to a radar signal including a sea surface reflection signal according to a Weibull distribution received by a radar device such as a ship. An object of the present invention is to provide a radar signal processing method and device capable of reliably suppressing a sea surface reflection signal.

[Means for Solving the Problem] In order to achieve this object, according to the present invention, first, as a radar signal processing method including a sea surface reflection signal according to a Weibull distribution, a moving average of a logarithmically converted radar signal is calculated. The difference signal is obtained by subtracting the difference signal, the difference signal is inversely logarithmically converted into a linear signal, and the first to nth moment signals of this linear signal are calculated, and a predetermined coefficient is used for each moment signal. Then, the threshold signal is calculated as the sum of the moment signals multiplied by the coefficient, and finally the threshold signal is subtracted from the linear signal to obtain an output signal.

Further, in the present invention as a radar signal processing device including a sea surface reflection signal according to the Weibull distribution, in the present invention, moving average means for calculating a moving average of the logarithmically converted radar signal; moving average signal from the moving average means And a logarithmic-converted radar signal to output a differential signal; an inverse logarithmic conversion unit that inversely logarithmically converts the differential signal from the first differential unit to output a linear signal; Moment calculation means for generating in parallel the first to nth moment signals of the linear signal from the inverse logarithmic conversion means;
Moment coefficient means for multiplying the coefficient using a predetermined coefficient for each moment signal from the 1st to nth order generated by the moment calculation means; and the 1st to nth coefficient multiplied by the moment coefficient means Summing means for calculating a threshold value signal as the sum of the next moment signals; and second difference means for calculating an output signal by subtracting the threshold value signal obtained by the summing means from the linear signal converted by the antilogarithmic transformation means. Prepare

Further, as a radar signal processing method and apparatus, a threshold signal obtained as a sum of moment signals multiplied by a coefficient may be used as it is as an output signal.

[Operation] According to the radar signal processing method and apparatus of the present invention having such a configuration, the false alarm probability becomes constant by the signal processing on the sea surface reflection signal according to the Weibull distribution.
An output signal can be obtained, the sea surface reflection signal is surely suppressed so as not to remain, and even if the sea surface reflection is received, a target image of a ship or the like can be clearly displayed to ensure navigation safety.

[Embodiment] First, a radar signal processing method of the present invention for suppressing a sea surface reflection signal according to a Weibull distribution included in a radar video signal (hereinafter, simply referred to as a radar signal) is obtained in the following processing steps.

The radar signal is logarithmically converted.

The moving average is calculated from the logarithmically converted radar signal.

The differential signal is obtained by subtracting the moving average obtained from the logarithmically converted radar signal.

The differential signal is inversely logarithmically converted into a linear signal.

The first to nth moment signals of the linear signal are calculated.

The coefficient is multiplied by using a predetermined coefficient determined for each moment signal.

The threshold signal is calculated as the sum of the first to nth moment signals multiplied by the coefficient.

Finally, the threshold signal is subtracted from the linear signal obtained above to obtain an output signal.

Such a radar signal processing method of the present invention is based on the following principle.

First, the probability density function of the radar signal X that follows the Weibull distribution
Pc (X) is given by the following expression (1).

Here, b is called a scale parameter and c is called a shape (shape) parameter, which is a parameter that determines the characteristics of the Weibull distribution.

The logarithmic conversion output Y of the radar signal X is represented by Y = lnX (2), and the moving average value <Y> of the logarithmic conversion output Y is represented by the following equation by the parameters b and c of the equation (1).

<Y> = lnb− (1 / c) · γ (3) where γ is the Euler constant.

Next, the Log-CFAR output V of the logarithmic conversion output Y is a value obtained by subtracting the moving average <Y> from the logarithmic conversion output Y, so that V = Y- <Y> (4), and this Log-CFAR output The inverse logarithmic conversion output Z of V is expressed by the following equation.

Z = ^ev (5) By substituting the equations (2), (3) and (4) into the equation (5), the antilogarithmic transformation output Z can be transformed as follows.

Z = (X / b) e ^{γ / c} (6) Next, the Jth moment of the antilogarithmic transformation output Z given by the equation (6) is obtained as follows.

<Z ^J > = e ^{(J / c) γ} Γ (J / c + 1) (7) where Γ is a gamma function.

On the other hand, the antilogarithmic transformation output Z given by the equation (6)
The false alarm probability Pfa of T is the threshold (threshold level)
Then, it is expressed by the following equation.

Here, when the threshold T is represented by a linear sum (sum) of the Jth moments of the antilogarithmic transformation output Z, Becomes Therefore, the false alarm probability Pfa of the inverse logarithmic transformation output Z given by the equation (8) is Given as

Here, J is the Jth moment of the inverse logarithmic transformation output Z
= 1,2,3 and J = 1,2,3,4, the coefficient a _{J in} the equation (9) is determined as shown in FIG. The false alarm probability Pfa given by the formula is J
= 1 to 3 becomes substantially constant as shown in FIG. 5, and when J = 1 to 4 becomes more constant as shown in FIG. 6, the sea surface reflection signal having a Weibull distribution is obtained by the present invention. It has been confirmed that a sufficiently suppressed CFAR output can be obtained.

The values of the coefficients a1 to a3 and a1 to a4 of J = 1 to 3rd order or J = 1 to 4th order shown in FIG. 4 are calculated as the moving average value <Y> given by the above equation (3). Is a value obtained by the optimization method with the number N of logarithmic conversion outputs Y for N = 16, and this coefficient a _J depends on the false alarm probability Pfa and the number N of logarithmic conversion outputs Y for obtaining the moving average <Y>. To do. Furthermore, the value of a _J is not uniquely determined by each parameter, and there are several optimal solution combinations.

FIG. 1 is a block diagram of an embodiment showing an embodiment of a radar signal processing device according to the present invention.

In FIG. 1, reference numeral 100 is a log amplifier, which inputs a radar signal X sampled at a predetermined cycle, logarithmically amplifies it, and outputs a logarithmic conversion signal Y.

Here, the radar signal to be processed by the apparatus of FIG. 1 may be an analog signal or a quantized radar signal obtained by converting a sampled analog signal into a digital signal. As a matter of course, each circuit unit adopts a circuit system adapted to an analog signal and a digital signal.

The logarithmic conversion signal Y from the log amp 100 is the moving average circuit 1
A predetermined number N of logarithmic conversion signals Y given to 0
Calculate the moving average of.

FIG. 2 is a block diagram of an embodiment showing one embodiment of the moving average circuit 10 of FIG. 1, and is composed of a delay circuit 24, an adding circuit 26 and a dividing circuit 28.

The delay circuit 24 has a number N of moving average values, for example, N =
Sixteen delay elements are connected in series, and the logarithmic conversion signal Y input in series is sequentially converted into N parallel outputs. The adder circuit 26 obtains the sum of N = 16 parallel outputs obtained for each new logarithmic conversion signal Y input from the delay circuit 24, and outputs the sum to the divider circuit 28. The division circuit 28 divides the output of the addition circuit 26 by the number N of average values to output a moving average value signal <Y>.

FIG. 3 is a block diagram of an embodiment showing another embodiment of the moving average circuit 10 of FIG.

In FIG. 3, the moving average circuit 10 is an N / 2 delay circuit 30,3.
4, the adder circuits 36 and 38, the adder 40 and the divider circuit 28. Further, the output of the N / 2 delay circuit 30 is given to the next N / 2 delay circuit 34 via the delay element 32 of one stage. ing.

When each of the N / 2 delay circuits 30 and 34 has N = 16,
Eight delay elements are connected in series, and the logarithmic conversion signal Y that is sequentially input is converted into eight parallel outputs to add circuits 36 and 3
Output to 8. The adder circuits 36 and 38 are N of the respective N / 2 delay circuits 30 and 34.
= The total sum of eight parallel outputs is obtained, and the sum is given to the adder 40 to obtain the total of N pieces, and the dividing circuit 28 divides by the number N to obtain the moving average, and the moving average value signal <Y> is output. To do.

One-stage delay element 3 provided between the N / 2 delay circuits 30 and 34
2 is used to output the logarithmic conversion signal Y to the first difference circuit 12 in the next stage shown in FIG.

Referring again to FIG. 1, the moving average circuit 10 is followed by a first difference circuit 12, and the first difference circuit 12 uses the logarithmic conversion signal Y obtained from the log amplifier 100 to calculate the moving average circuit 10. The N moving average value signals <Y> calculated in step 1 are subtracted and the difference signal V is output. This difference signal V becomes the Log-CFAR output given by the equation (4).

Further, the first difference circuit 12 can be realized by a difference circuit using an operational amplifier if it is an analog system, while it can be realized by a digital adder if it is a digital system.

The difference signal V from the first difference circuit 12 is an antilogarithm calculation circuit.
The antilogarithmic operation circuit 14 is provided to the antilogarithmic conversion circuit 14 and outputs the antilogarithmic conversion signal (linear signal) Z given by the equation (5).
The inverse logarithmic arithmetic circuit 14 is realized by a function arithmetic unit having an exponential function characteristic in the analog system, and in the digital system by a look-up table LUT storing the antilogarithm precalculated with the differential signal V as an address. realizable.

The antilogarithmic conversion signal Z from the antilogarithmic calculation circuit 14 is given to the threshold value calculation unit 50 for calculating the threshold value T given by the equation (9). In this embodiment, the threshold value calculation unit 50 uses the first to fourth order moment calculation as an example.

First, the inverse logarithmic conversion signal Z is the first moment calculation circuit.
16-1, 2nd moment calculation circuit 16-2, 3rd moment calculation circuit 16-3 and 4th moment calculation circuit 16-4 are respectively input, and the inverse logarithmic conversion signal Z is raised to the nth power, and then the moving average is calculated. Similar to the circuit 10, the average value is calculated by the average calculation algorithm. That is, in the above equation (9) For the terms on the right side except for, the moment is calculated for each of J = 1, 2, 3, and 4.

Subsequently, a first-order moment coefficient circuit 18-1, a second-order moment coefficient circuit 18-2, a third-order moment coefficient circuit 18-3 and a fourth-order moment coefficient circuit 18-4 are provided, and the first-order moment calculated in the preceding stage Each of the 4th moments is multiplied by a predetermined moment coefficient a1, a2, a3, a4. As the first to fourth order moment coefficients a1 to a4, the values a1 to a4 shown on the right side of FIG. 4 are used.

The first to fourth order moments thus multiplied by the predetermined moment coefficient are finally input to the summing circuit 20, and in the summing circuit 20, J = 1 to J shown in the equation (9) above. A threshold value T of 4 is calculated. The threshold calculation unit
The moment coefficient circuits 18-1 to 18-4 in 50 can be realized by an operational amplifier in the analog system, and can be realized by a look-up table in the digital system.
Similarly for 20, the analog type can be realized by an operational amplifier, and the digital type can be realized by a combination of a lookup table or a digital adding circuit.

The antilogarithmic conversion signal Z from the antilogarithmic calculation circuit 14 and the threshold signal T calculated by the threshold calculation unit 50 are finally output to the second difference circuit 22, and the second difference circuit 22 calculates the antilogarithmic conversion signal. The threshold signal T is subtracted from the (linear signal) Z to output the final CFAR signal. This second difference circuit
Since the output signal obtained from 22 uses the threshold value T calculated by the first to fourth moments according to the moment coefficient shown on the right side of FIG. 4, it follows the Weibull distribution signal as shown in FIG. Even for the sea surface reflection signal, an output signal with a constant false alarm probability Pfa can be calculated regardless of the characteristics of the distribution.

FIG. 7 is a block diagram of an embodiment showing another embodiment of the present invention, in which the threshold value signal T calculated by the threshold value calculation unit 50 is finally obtained except for the second difference circuit 22 shown in FIG. The output signal is That is, in the embodiment of FIG. 1, the output signal is the inverse logarithmic conversion signal (linear signal) Z minus the threshold signal T, but the false alarm is generated as in the embodiment of FIG. Even if the threshold signal T itself having a constant probability Pfa is displayed as an output signal,
An output signal in which the sea surface reflection signal is suppressed can be obtained.

The embodiment shown in FIGS. 1 and 7 is based on the calculation of the first to fourth moments. For example, the third-order moment calculation using the moment coefficients a1, a2, a3 shown on the left side of FIG. It may be a moment calculation, and the order of the moment calculation can be appropriately determined as needed.

Further, although the above embodiment has been described by taking the case where the sea surface reflection signal has a Weibull distribution as an example, also in the case of taking a distribution other than the Weibull distribution, similarly, the moment calculation of the inverse logarithmically converted linear signal is performed. Results and optimized coefficients
The threshold signal can be obtained by a _J to reliably prevent the sea surface reflection signal.

[Effect of the Invention] As described above, according to the present invention, even for a sea surface reflection signal according to the Weibull distribution, an output signal with a constant false alarm probability is calculated regardless of the characteristics of the distribution. By reliably suppressing the sea surface reflection signal and clearly displaying the ship image, it is possible to ensure the safety of navigation of the ship.

[Brief description of the drawings]

FIG. 1 is a block diagram of an embodiment showing one embodiment of the device configuration of the present invention; FIGS. 2 and 3 are configuration diagrams of an embodiment of the moving average circuit of FIG. 1; FIG. 4 is a moment calculation of the present invention. illustration of the coefficient a _J used; FIG. 5 is an explanatory view of a false alarm probability Pfa by third order moment calculation of the present invention; illustrates false alarm probability Pfa by fourth order moment calculation in FIG. 6 is the invention FIG. 7 is a configuration diagram of an embodiment showing another embodiment of the device configuration of the present invention; FIG. 8 is an explanatory diagram of a conventional system. 10: Moving average circuit 12: First difference circuit 14: Inverse logarithm calculation circuit 16-1 to 16-4: 1st to 4th moment calculation circuit 18-1 to 18-4: 1st to 4th moment coefficient circuit 20: Summing circuit 22: Second difference circuit 24: Delay circuit 26, 36, 38, 40: Adder circuit 28: Division circuit 30,34: N / 2 delay circuit 32: Delay element 50: Threshold calculation unit 100: Log amp

Claims (4)

(57) [Claims] 1. A differential signal is obtained by subtracting the moving average from a logarithmically converted radar signal, the differential signal is inversely logarithmically converted into a linear signal, and a moment signal from the first to nth order of the linear signal is obtained. The coefficient signal is calculated and multiplied by a predetermined coefficient for each moment signal, and a threshold signal is calculated as the sum of the coefficient signals of the 1st to nth moment signals, and finally the threshold signal is calculated. A radar signal processing method, characterized by subtracting from the linear signal to obtain an output signal. 2. A moving average means for calculating a moving average signal of a logarithmically converted radar signal; a first means for subtracting the moving average signal from the moving average means from the logarithmically converted radar signal to output a differential signal. Differential means for inversely logarithmically converting the differential signal from the first differential means to output a linear signal; first to nth moments of the linear signal from the antilogarithmic conversion means A moment calculating means for generating signals in parallel; a moment coefficient means for multiplying a coefficient using a predetermined coefficient for each of the first to nth moment signals generated by the moment calculating means; Summing means for calculating a threshold signal as a sum of first to nth moment signals multiplied by a coefficient by the coefficient means; and converting the threshold signal obtained by the summing means by the inverse logarithmic conversion means. Second which is subtracted from the linear signal obtaining an output signal
A radar signal processing device, comprising: 3. A differential signal is obtained by subtracting the moving average from the radar signal that has been logarithmically converted, and the differential signal is inversely logarithmically converted into a linear signal. Moment signals from the first order to the nth order of the linear signal are obtained. The coefficient signal is calculated and multiplied by a predetermined coefficient for each moment signal, and the threshold signal is calculated as the sum of the coefficient signals of 1st to nth moment signals multiplied by the coefficient. A radar signal processing method comprising: 4. A moving average means for calculating a moving average signal of a logarithmically converted radar signal; a first which outputs a difference signal by subtracting the moving average signal from the moving average means from the logarithmically converted radar signal. Differential means for inversely logarithmically converting the differential signal from the first differential means to output a linear signal; first to nth moments of the linear signal from the antilogarithmic conversion means A moment calculating means for generating signals in parallel; a moment coefficient means for multiplying a coefficient using a predetermined coefficient for each of the first to nth moment signals generated by the moment calculating means; A radar signal processing device, comprising: a summing means for calculating a threshold value signal as a sum of the first to nth moment signals multiplied by the coefficient means.