JPH0527018A - Radar signal processing device - Google Patents

Abstract

PURPOSE:To automatically discriminate and recognize the kind or the like of a moving target in a radar system of a radar signal processing device. CONSTITUTION:A radar signal processing device is provided with a frequency analyzing means 81 to obtain the frequency spectrum of the phase detection signal where phase detection of the radar received signal is made, a collating means 82 to collate the pattern of the frequency spectrum obtained by the frequency analyzing means 81 with the model pattern of various known frequency spectra of the moving target, and a recognizing means 83 to recognize the kind of the moving target on the basis of the collated result by the collating means 82.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is based on information received from a MIT (moving target indicating device) radar receiver such as a coherent pulse Doppler radar to determine what kind of moving body the moving target detected is. The present invention relates to a radar signal processing device that automatically estimates and recognizes.

[0002]

2. Description of the Related Art A conventional moving target indicating device indicates the presence of a moving target, and what kind of moving target is (for example, type of person, jeep, truck, tank, aircraft, missile, etc.)? Recognition cannot be done automatically. Further, a technique for recognizing a target under the condition that many targets are included in the instantaneous field of view of the radar device, that is, the antenna beam width, has not yet been established.

Conventionally, these recognitions are based on the empirical judgment of the operator. That is, a skilled operator listens to the strength and tone of the signal sound of the received echo signal, and based on his many years of experience, what kind of moving target is and how many are there. I'm guessing there is.

[0004]

As described above, conventionally, a trained operator recognizes the type and number of moving targets by listening to signal sounds based on many years of experience. When trying to automate such recognition, it is difficult to perform signal processing for target recognition in real time, or the environment at the time of measurement processing, such as weather conditions (temperature, time, season,
Others) and clutter (water waves, clouds, fluctuations of grass and trees, etc.) are difficult to consider, and automation was not easy.

However, for example, in a radar device for defense that requires countermeasures for each type of moving target within a very short time, high-speed detection / recognition of the type of moving target is absolutely necessary. Since it is not possible to rely solely on human experience, there is an increasing need for automation of identification and recognition of target types.

The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a radar signal processing device capable of automatically identifying and recognizing the type of a moving target in a radar system. is there.

[0007]

FIG. 1 is a diagram illustrating the principle of the present invention. As one form, the radar signal processing device according to the present invention is a frequency analysis means 81 for obtaining a frequency spectrum of a phase detection signal obtained by phase detection of a radar reception signal.
And a matching means 82 for matching the frequency spectrum pattern obtained by the frequency analysis means 81 with a known frequency spectrum model pattern for various moving targets, and recognizing the type of the moving target based on the matching result of the matching means 82. And a recognition means 83 for doing so.

A radar signal processing device according to the present invention is
As another mode, in the above radar signal processing device,
A size calculating means 84 for calculating the size of the moving target based on the amplitude strength of the received signal of the moving target is further provided, and the recognizing means 83 is configured to also consider the size information when recognizing the type of the moving target. It

A radar signal processing device according to the present invention is
As another form, each of the radar signal processing devices described above further includes a target number calculation means 85 for estimating the number of moving targets by monitoring the fluctuation of the amplitude intensity of the received signal of the moving target, and the recognition means 83 The number of moving targets is also recognized based on the target number information of the target number calculating means 85.

A radar signal processing device according to the present invention is
As still another form, each of the radar signal processing devices described above further includes a clutter detection unit 86 for detecting the presence or absence of clutter in the received signal, and the recognition unit 83.
Is configured to determine the recognition rate for the moving target based on this clutter information.

[0011]

When the operator recognizes the sound of the reflected wave and estimates the type of the moving target, the speed is determined from the tone color of the signal and the type of the moving target according to the speed (for example, a person or an automobile). Echo sound that distinguishes the type of vehicle (for example, passenger car, jeep, truck, tank, etc.) based on the difference in the timbre of the reflected wave depending on its shape if it is an automobile. It can be considered that the pattern recognition such as determining the size of the moving target (for example, a jeep or a tank in the case of a car) is unconscious according to the strength of the. The present invention is based on these findings.

That is, the radar signal processing device of the present invention is
The phase detection signal obtained by phase-detecting the radar reception signal is subjected to frequency spectrum analysis by the frequency analysis means 81 to obtain a frequency spectrum distribution curve of the reception signal including the moving target and the like. Then, based on the Doppler frequency position where the reflected wave from the moving target exists, the speed of the moving target is known, and it is presumed what the moving target is, and the matching means 82 is used for the moving target candidate with a high possibility. In the above step, the model patterns of the frequency spectra of the moving targets as those candidates obtained in advance are collated. The recognition unit 83 estimates the type of the moving target based on the matching result, for example, the magnitude of the degree of correlation for each model pattern.
recognize.

The signal intensity of the echo from the moving target is the intensity of the echo at the echo reflection point, and therefore corresponds to the size of the moving target. Therefore, the size calculating means 84 calculates the size of the moving target based on this. This size can also largely determine what the moving target is. For example, people, passenger cars, tanks, aircrafts, etc. are obviously different in size, so it is possible to roughly estimate what the moving target is based only on the size information of these moving targets. Therefore, the recognition unit 83 also considers this size information when recognizing the moving target, and determines the type of the moving target that is appropriate for both the pattern matching result and the size.

Further, by monitoring the change in the amplitude of the echo from the moving target, it is possible to obtain a probability density distribution of the amplitude value. If mountains occur in this probability distribution, the number of the mountains is counted and the moving target is calculated. The number of can be roughly estimated. The target number calculation means 84 generally calculates the number of movement targets. The result is reflected in the recognition by the recognition means 83.

If clutter is further generated, the recognizing means 83 depends on the amount of clutter.
Influences the recognition rate of recognition (that is, probability of recognition). Therefore, the clutter detection means 86
The clutter situation is monitored with and the recognition means 83 changes the recognition rate according to the detection result. For example, if many clutters appear, the recognition rate is lowered.

[0016]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 2 shows a moving target indicating device by a coherent pulse Doppler radar incorporating a radar signal processing device as an embodiment of the present invention. In the figure, 1 is a radar transmission / reception unit, 2 is a frequency analysis unit, 3 is an intensity processing unit, 4
Is a clutter processing unit, 5 is a frequency processing unit, and 6 is a recognition processing unit.

The radar transmitting / receiving unit 1 includes a transmitting / receiving antenna 11, a pulse generator 12, a transmitting wave oscillator 13, and a Palace modulator 1.
4, circulator 15, amplifier 16, phase detector 17
Etc. Here, the phase detector 17 phase-detects the echo signal received through the antenna 11, the circulator 15, and the amplifier 16 with two local oscillation signals having a phase difference of 90 ° as reference waves, and outputs an I video signal. It is a circuit that creates and outputs a Q video signal.

The frequency analysis unit 2 includes a sum of squares circuit 21, an optimum leveling circuit 22, and a fast Fourier transform (FFT) circuit 23.
It is composed including. The sum-of-squares circuit 21 is a circuit for obtaining the sum-of-squares of the I video signal and the Q video signal from the phase detector 17, and has a configuration as shown in FIG. 3, for example. That is, the I video signal and the Q video signal are respectively transmitted at high speed A
A / D conversion is performed for digitization, these are respectively put in two memories MEM1 and MEM2 having a 12-bit width, the outputs of these two memories MEM1 and MEM2 are multiplied by a high-speed multiplier MP, and then these Squared I
And the Q video signal are added by the adder ADD to obtain the sum of squares, which is taken as one video signal on the time axis.

In this way, the sum of squares is obtained by the sum of squares circuit 21 to make one video signal, so that the phase detection output changes or sometimes disappears depending on the signal reflection position (distance) from the target. Is lost. That is, for example, the phase detection A · Bsin (θ _a −θ _b ) obtained by multiplying two waves Asin (θ _a ) and Bsin (θ _b ) is sometimes output at a certain distance due to the influence of the sine wave term. Since it disappears, two waves A ・ B sin (θ _a −
θ _b ) and A · B cos (θ _a −θ _b ), squared and summed (A · B) ² {sin ² (θ _a −θ _b ) + cos ²
By setting (θ _a −θ _b )} = (A · B) ² , the output video signal is not affected even if the phase, for example, θ _a , changes depending on the distance, thereby eliminating the phase blind.

The video signal from the sum-of-squares circuit 21 is then input to the optimum leveling circuit 22 and normalized by using the peak value to be adjusted to an appropriate level. This level-adjusted video signal is then input to the fast Fourier transform circuit 23. The fast Fourier transform circuit 23 fast Fourier transforms the video signal of the time function into a power spectrum of the frequency function in order to analyze the spectrum distribution (curve) on the frequency axis of the video signal. This power spectrum distribution is, for example, as shown in FIG. 5, the horizontal axis is the Doppler frequency, the vertical axis is the amplitude intensity, and the frequency positions f ₀₁ , f corresponding to the speeds of various moving targets are set.
The maximum value of the amplitude intensity of the reflection signal from the moving target appears at ₀₂ , f ₀₃ , f _04, ...

The analysis results of the fast Fourier transform circuit 23 are input to the intensity processing section 3, crack processing section 4 and frequency processing section 5, respectively.

The frequency processing unit 5 is a fast Fourier transform circuit 2
The frequency spectrum distribution patterns of the echo signals input from 3 are obtained in advance for various moving targets (people, automobiles such as jeep and trucks, tanks, aircraft, missiles, etc.) when they are moving. This is a circuit that matches the model pattern of a typical frequency spectrum and finds the correlation with them. The frequency processing unit 5, the processing section setting unit 51, the pattern source 52 ₁ to 52 _n, the center frequency f
₀ matching unit 53 _{1 to} 53 _n , pattern presenting unit 54 _{1 to} 5
4 _n , correlation units 55 _{1 to} 55 _n , coefficient output units 56 _{1 to} 56
_It is configured to include _n etc.

For example, as shown in FIG. 5, the processing section setting unit 51 determines the Doppler frequency positions f ₀₁ and f in accordance with the type of moving target, such as a person, an automobile, or an aircraft.
_When the maximum values of the amplitude intensity occur at ₀₂ , f ₀₃ , f _04, ..., the ranges E1 and E in the vicinity including those maximum values
It is a circuit that sets sections such that 2, E3, E4, ... Are each one processing section. The type of movement target detected in each processing section set in this way can be limited to some extent. For example, it is possible to determine that the moving target detected in the processing range E1 having a low Doppler frequency is likely to be a low-speed moving object such as a person or a vehicle traveling at a low speed, and is unlikely to be an aircraft, and processing having a high Doppler frequency is performed. It can be determined that the moving target detected by the intention E4 is highly likely to be a high-speed moving body such as an aircraft, and very unlikely to be a person.

The pattern sources 52 _{1 to} 52 _n are memory circuits which hold typical model patterns of the frequency spectrum of the moving target which are likely to be detected in the respective processing sections set by the setting section 51. These model patterns are obtained in advance by experiments etc. when various moving targets are moving at a typical moving speed.

[0025] f ₀ matching unit 53 ₁ to 53 _n are example, by changing the clock speed of the shift and read clock of the read timing when reading of the model pattern from the pattern source 52 ₁ to 52 _n, the center frequency of the model patterns and distribution It operates to match the width with the center frequency and distribution width of the frequency spectrum of the echo signal. It is possible that even the same moving target, for example a car, is traveling at different speeds, but it is not economical to have a model pattern for every speed of the car, so at typical speeds for cars. By preparing one model pattern and changing its center frequency and distribution width, a matching model pattern for each speed is realized, thereby reducing the amount of model patterns prepared in advance.

The pattern presentation unit 54 ₁ through 54 _n in the circuit for presenting each matching model pattern for changing the center frequency and distribution width f ₀ registration unit 53 ₁ to 53 _n to the correlation unit 55 ₁ to 55 _n is there.

The correlating units 55 _{1 to} 55 _n calculate the correlation coefficient by correlating the matching model pattern presented by the pattern presenting units 54 _{1 to} 54 _n with the frequency spectrum distribution pattern of the echo signal. And has a configuration as shown in FIG. 4, for example. That is, it includes a variable clock source CLK1, a memory MEM1 to which a frequency spectrum distribution pattern of an echo signal is input, a pattern memory Pat1 to which a matching model pattern is input, a correlator COR1 for obtaining a correlation between these patterns, and the like. . This correlation coefficient is obtained as follows, for example. Correlation coefficient = {Σ (x _i y _i ) / N−x _av · y _av } / S _x · S _y Here, x _i and y _i respectively represent the i-th data,
x _av and y _av are average values of each data, N is the number of data, S _x ,
S _y represents the variance of x and y data. Each correlator 55 _1-5
The correlation coefficient calculated by 5 _n is the coefficient output unit 56 _{1 to} 56 _n.
Is output to the recognition processing unit 6.

The frequency processing unit 5 performs parallel processing and pipeline processing for correlating the frequency spectrum pattern of the echo signal with various typical model patterns (person, truck, jeep, tank, aircraft, missile, etc.). It is performed by processing, and the processing time is greatly shortened to achieve high-speed processing.

The intensity processing unit 3 is configured including a timer 31, a target cross-sectional area calculation section 32, f ₀ different intensity management data 33, the same f ₀ target speed calculating section 34, a target size estimating unit 35 or the like. The timer 31 is the pulse generator 12 of the radar transmitter / receiver 1.
Generate distance information of moving target based on pulse from
This is given to the target cross-sectional area calculation unit 32.

Based on the distance information, the video signal from the sum of squares circuit 21, and the data from the intensity management data 33, the target cross-sectional area calculating unit 32 determines the absolute value of the reflection intensity of the echo signal from the moving target to determine the moving target. Size (reflection cross section)
Figure out. That is, radar equation: reception intensity Pr = Pt · G ² · λ ² · σ / (4π) ³ R ⁴ (where Pt is transmission power, G is antenna gain, λ is transmission wavelength, and σ is target reflection cross section , R are distances), the reflection cross section σ can be calculated by measuring the reception intensity Pr when the distance R is determined. However, this is on condition that the signal from the moving target is detected larger than the clutter.

The target number calculation unit 34 is a circuit for estimating the number of moving targets included in the antenna beam width for each moving target having center frequencies f ₀₁ , f ₀₂ , f ₀₃ ,. This operation will be described in detail later. Size estimating unit 35
Is a circuit for obtaining the size of the moving target from the target sectional area information from the target sectional area calculating unit 32 and the target number information from the target number calculating unit 34 and then estimating the type of the moving target. For example, the moving target can be estimated as a person << Jeep << truck <Tank <Airplane etc. in ascending order of the target cross-sectional area. The type information estimated by the size estimation unit 35 and the target number information calculated by the target number calculation unit 34 are recognized by the recognition processing unit 6.
Given to.

The clutter processing unit 4 includes a clutter measuring unit 41,
It is configured to include a selection unit 42, a clutter unit 43, a clutter correlation unit 44, a state determination unit 45, and the like, and determines the presence or absence of clutter and gives the clutter presence information to the recognition processing unit 6.

The recognition processing unit 6 includes information on the type and number of moving targets from the intensity processing unit 3, clutter presence / absence information from the crack processing unit 4, patterns of reflected signals of the moving target from the frequency processing unit 5, and a matching model. This is a circuit that comprehensively determines the type of movement target based on the correlation information with the pattern, and includes a target-specific total recognition processing unit 61, a final target number / recognition rate determination unit 62, a target-specific recognition display data output unit 63, and the like. Consists of inclusive.

The operation of the apparatus of this embodiment will be described below. (1) The radar transmission / reception unit 1 radiates a pulse transmission wave toward a moving target. This pulse transmission wave is reflected by the moving target, and part of it returns to the radar device as a reflected wave (echo).

(2) The antenna 1 receives the reflected wave from the moving target.
It is received by 1 and amplified by the amplifier 16 and then 9 by the phase detector 17.
Phase detection is performed with a local oscillation signal having a 0 ° phase difference to generate an I video signal and a Q video signal.

(3) The sum of squares of the I video signal and the Q video signal is obtained by the sum of squares circuit 21, the video signal of the sum of squares is optimized to an appropriate level by the optimum leveling circuit 22, and further this optimum leveling is performed. The obtained video signal is subjected to fast Fourier transform by the fast Fourier transform circuit 23 to obtain the frequency spectrum distribution curve of the received wave. This frequency spectrum distribution curve is as shown in FIG. 5, for example. In FIG. 5, the horizontal axis represents the Doppler frequency and the vertical axis represents the reception intensity.

[0037] (4) a fast Fourier transform and spectral distribution curves obtained, in order to parallel processing of recognizing a target at high speed, the correlation unit 55 _1-5 connected in parallel in the frequency processing unit 5
Put in 5 _n of memory. Further, in the frequency processing unit 5, the processing section setting unit 51 detects a moving target signal from the frequency spectrum distribution and determines the processing range. In this, the envelope of the frequency spectrum is differentiated to obtain local maxima and local minima with no signal, and the local minimum points on both sides of one local maxima are set as one processing range. For example, in the case of FIG. 5, the processing range E1 is defined between the local minimum points on both sides of the local maximum point f ₀₁ where the reflection signal from the moving target exists, and the processing ranges E2 and E are similarly set.
3, E4 ... are set. Here, when a large number of maximums continuously exist in a narrow range such as the processing range E3, it is regarded as a reflection signal from one moving target. This is a phenomenon that occurs when the moving target is a cloud or the like.

(5) From the maximum frequency (signal center) of this frequency spectrum distribution, some pattern sources of known model patterns considered to be appropriate as the type of the moving target are selected. This maximum frequency is the velocity of the moving target. Therefore, it is possible to roughly know what kind of moving object the moving target is from the speed, and the range of the model pattern to be used can be limited to some extent.

(6) In order to correlate the selected model pattern with the reflected signal pattern of the moving target, the frequency distribution (distribution center, distribution width, etc.) near the center of the model pattern is compared with that of the received signal pattern from the moving target. The f ₀ aligning units 53 _{1 to} 53 _n adjust the center alignment (by changing the read timing) and the width alignment (by changing the clock speed of the read clock) so that they coincide with each other. In this way the processing range, based on the model pattern of the pattern source, currently it is possible to create a verification model pattern suitable for matching the received signal pattern from the movement target that has received, which pattern presentation unit 54 ₁
And outputs to the correlation unit 55 ₁ to 55 _n via 54 _n.

(7) In the correlation units 55 _{1 to} 55 _n , the correlation between the matching model pattern from the pattern presenting units 54 _{1 to} 54 _n and the reflection signal pattern from the moving target is obtained, and the correlation coefficient or the correlation coefficient is calculated. The squared value is calculated and given to the recognition processing unit 6.

(8) In the recognition processing unit 6, the target-based total recognition processing unit 61 arranges the matching model patterns in descending order of the degree of correlation (coincidence degree) based on this correlation information. Then, the types of moving bodies corresponding to the model patterns are displayed in order from the one having the highest degree of coincidence. For example, when it is determined that the moving target appearing in a certain processing range is a vehicle, the type of the vehicle is further displayed in the order of high accuracy such as jeep, truck, bus, and tank.

(9) In the target number calculation unit 34, the fluctuation of the amplitude fluctuation near each center frequency f ₀ (f ₀ ± Δf) is monitored for a certain period of time, and the probability density distribution of the amplitude value is measured. Looking at this probability distribution, if mountains are formed in it, the number of mountains is counted and used as the number of movement targets. This is because the moving target within the radar antenna beam width moves or stops, or when one moving target is hidden behind another similar moving target, the width of the moving target is equivalent to the cross-sectional area. Since a continuous change in the amplitude value occurs, the number of units is roughly estimated by capturing this as a probability density distribution.

(10) In the intensity processing unit 3, the target cross-section calculation unit 32 estimates the target reflection cross-section by observing the intensity and change in intensity of the video signal before the fast Fourier transform, and the target size is calculated. Give judgment. As described above, this calculates the cross-sectional area of the moving target from the absolute value of the signal reflection intensity according to the radar equation. Once the cross-sectional area of the moving target is determined, it is then possible to determine the size of the moving target and thus the type of the moving target. For example, by estimating that the person is a jeep, a truck, a tank, and an airplane, an appropriate type of the moving target is determined from the reflection cross section of the moving target.

(11) The recognition processing unit 6 determines the type of the moving target that is appropriate for any of these types of information, based on the correlation information between the type / number information and the model pattern from the frequency processing unit 5. To do.

(12) Further, the clutter processing section 4 judges whether or not clutter is occurring in the received echo. Since clutter determination is a conventional technique, it will not be described in detail here. Clutter, which is a disturbing component when recognizing a moving target, has a spectral distribution that changes depending on environmental conditions such as weather (temperature, humidity, wind direction, wind speed, rain, snow, fog, sandstorm, etc.), so the type of moving target is determined. In doing so, it is necessary to consider these effects.

(A) Clutter is usually highly random.
Therefore, while the signals from the moving target are added and become large when the integration processing is performed, the clutter grows small with respect to the number of additions. Therefore, the signal component of the moving target can be extracted from the received signal by repeating the addition.

(B) Clutter is distributed like 1 / F noise centering on the carrier on the frequency axis. Therefore, clutter can be removed from the received signal by using a low-pass removal filter.

(C) Since the clutter enhances the amplitude intensity and the frequency spread by the wind, the clutter is treated by strengthening the above processes (a) and (b).

As described above, the increase of clutter makes the detection of the signal from the moving target and the judgment to be regarded as the signal erroneous, and therefore the judgment criteria and the specification are made gentle in the recognition, depending on the climate condition.

(13) In the recognition processing unit 6, when the clutter processing unit 4 notifies that the amount of extraneous noise and clutter is increasing, the recognition rate (that is, the estimation is correct) when the type of the moving target is specified. Display)
Notify the operator by sound or display. In addition, the specifications of the moving target (size, relative moving speed, etc.) are also notified. In addition, it automatically connects to other responding devices to take appropriate measures and provide information promptly.

In this embodiment, the processing speed is increased by performing the fast Fourier transform in order to obtain the frequency spectrum, and the parallel signal processing and the pipeline processing are performed in real time. In addition, by setting the processing range, the collation with the model pattern can be performed efficiently.

As described above, by automatically recognizing the target of the radar signal, for example, the type of the target can be identified and recognized at high speed.
Thereby, when the present invention is applied to a collision prevention device, it is possible to avoid a collision by appropriate countermeasures, and also to perform warning monitoring.
When applied to a system such as interception, it is possible to avoid or shoot down the threat from the moving target in front.

[0053]

As described above, according to the present invention, it is possible to automatically identify and recognize the type of moving target in the radar system. Such a device can be made very compact (for example, as a small unit or module), small in size, and lightweight by making it into an LSI, and easily connected to a conventional radar device to recognize a moving target at high speed. However, various other connection applications are possible.

[Brief description of drawings]

FIG. 1 is a diagram illustrating the principle of the present invention.

FIG. 2 is a block diagram showing an MTI radar device incorporating a radar signal processing device as an embodiment of the present invention.

FIG. 3 is a block diagram showing a configuration example of a square sum circuit in the apparatus of the embodiment.

FIG. 4 is a block diagram showing a configuration example of a correlation unit in the apparatus of the embodiment.

FIG. 5 is a diagram showing an example of a frequency spectrum distribution curve of a radar reception signal.

[Explanation of symbols]

1 radar transmission / reception unit 2 frequency analysis unit 3 intensity processing unit 4 clutter processing unit 5 frequency processing unit 6 recognition processing unit 11 radar antenna 17 phase detector 21 square sum circuit 22 optimal leveling circuit 23 fast Fourier transform circuit 32 target cross-section calculation Unit 33 Central frequency-specific intensity management data 34 Same central frequency target calculation unit 35 Target size-specific estimation unit 51 Processing section setting unit 52 _{1 to} 52 _n Pattern source 53 _{1 to} 53 _n Central frequency matching unit 53 _{1 to} 53 54 ₁ ˜54 _n pattern presenting unit 55 _{1 to} 55 _n correlation unit 56 _{1 to} 56 _n coefficient output unit 61 target-specific total recognition processing unit 62 final target number / recognition rate determination unit 63 target-specific recognition display data output unit

Claims (4)

[Claims] 1. A frequency analysis means (8) for obtaining a frequency spectrum of a phase detection signal obtained by phase detection of a radar reception signal.
1), a matching means (82) for matching the pattern of the frequency spectrum obtained by the frequency analyzing means with a model pattern of the frequency spectrum of various known moving targets, and a moving target based on the matching result of the matching means. Radar signal processing apparatus comprising a recognition means (83) for recognizing the type of. 2. A size calculation means (8) for calculating the size of the moving target based on the amplitude strength of the received signal of the moving target.
4. The radar signal processing device according to claim 1, further comprising 4), wherein the recognizing means is configured to consider the size information when recognizing the type of the moving target. 3. A target number calculating means (85) for estimating the number of moving targets by monitoring a change in the amplitude intensity of the received signal of the moving target, the recognizing means including the target of the target number calculating means. The radar signal processing device according to claim 2, wherein the number of moving targets is also recognized based on the number information. 4. A clutter detection means (86) for detecting the presence or absence of clutter in the received signal, the recognition means being configured to determine a recognition rate for a moving target based on the clutter information. The radar signal processing device according to any one of Items 1 to 3.