JP2015161662A - radar signal processing apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make it possible to suppress both clutter components and azimuth ambiguity components even if a DPCA condition is not met.SOLUTION: A radar signal processing apparatus comprises an unnecessary-component suppression unit 4 estimating clutter components and azimuth ambiguity components contained in signal components divided into a positive region by a signal-component division unit 3 and suppressing the clutter components and the azimuth ambiguity components contained in the signal components divided into the positive region, and estimating clutter components and azimuth ambiguity components contained in signal components divided into a negative region and suppressing the clutter components and the azimuth ambiguity components contained in the signal components divided into the positive region.

Description

  The present invention relates to a radar signal processing device that detects a moving target by suppressing clutter components and azimuth ambiguity components included in a SAR image that is an image of a synthetic aperture radar (SAR). is there.

In the radar signal processing apparatus disclosed in Non-Patent Document 1 below, a plurality of transmission / reception antennas are provided in the orbital direction, so that a plurality of channels are provided and transmission / reception phase centers are observed in different states in the orbital direction. A plurality of SAR images (a plurality of SAR images having the same observation area and different observation times) are acquired.
In this radar signal processing device, after performing alignment (registration) between a plurality of SAR images, the difference between the plurality of SAR images is taken to suppress the clutter component and detect the moving target. (Moving Target Indicator) processing is performed.

However, in some SAR images, in addition to the clutter component, an azimuth ambiguity component due to the side lobe of the antenna pattern may occur.
At this time, when a DPCA (Dilated Phase Center Antenna) condition, which is a condition in which the phase centers of transmission and reception between channels overlap, registration between a plurality of SAR images can be performed in units of pixels. The moving target can be detected by suppressing the azimuth ambiguity component in the same way as the clutter component.
However, when the DPCA condition is not satisfied, registration between a plurality of SAR images is performed in units of subpixels. When performing registration with resampling in units of subpixels, the azimuth ambiguity component behaves differently from the clutter component, and thus phase rotation occurs in the azimuth ambiguity component in the SAR image after registration. As a result, there arises a problem that azimuth ambiguity components remain even if differences between a plurality of SAR images are taken.

Here, in order to confirm the above problem, a target detection method of the radar signal processing device disclosed in Non-Patent Document 1 will be briefly described.
FIG. 9 is a block diagram showing a radar signal processing device disclosed in Non-Patent Document 1, and FIG. 10 is a flowchart showing processing contents of the radar signal processing device of FIG.
This radar signal processing apparatus receives two reflected SAR images (1) and (2) by receiving a reflected wave of the above-described pulse reflected from a target by repeatedly radiating a pulse from one transmitting antenna. And the moving target is detected from the two SAR images (1) and (2).
At this time, the distance between the two receiving antennas is expressed by 2 mV _p / PRF + 2LV _p / PRF with respect to the platform speed V _p . PFR (Pulse Repetition Frequency) is a pulse repetition frequency, m integer, L is a real number, and 0 <L <1.
In the case of L = 0, the DPCA condition in which the transmission / reception antenna phase centers overlap is satisfied as shown in FIG. 11, but here, since L ≠ 0, the DPCA condition is satisfied as shown in FIG. Is not satisfied.

First, the registration unit 101 performs registration to match the position of the SAR image (2) with the position of the SAR image (1) (step ST101 in FIG. 10).
At this time, the registration unit 101 performs azimuth resampling in order to correct an image shift of (m + L) pixels in the azimuth direction. As a result, the received signal r ₁ (p, q) of the SAR image (1) in the p-th range bin and the q-th azimuth bin is described as follows.

In equation (1), s (p, q) is a clutter component 111 as shown in FIG. 13, and a ₊ (p, q) is a primary azimuth ambiguity component 112 as shown in FIG. a ₋ (p, q) is a primary azimuth ambiguity component 113 as shown in FIG. 13.
The primary azimuth ambiguity component is generated by the sidelobe pattern of the antenna, and the secondary and subsequent components are assumed to be negligibly small.
n ₁ (p, q) is a noise component in the SAR image (1), and t ₁ (p, q) is a moving target component in the SAR image (1).
Note that there are P range bins and Q azimuth bins.

式（２）において、ｎ_２（ｐ，ｑ）はＳＡＲ画像（２）内の雑音成分、ｔ_２（ｐ，ｑ）はＳＡＲ画像（２）内の移動目標成分である。 On the other hand, the received signal r ₂ (p, q) of the SAR image (2) in the p-th range bin and the q-th azimuth bin is described as follows.

In Expression (2), n ₂ (p, q) is a noise component in the SAR image (2), and t ₂ (p, q) is a moving target component in the SAR image (2).

As described above, phase rotation occurs in the azimuth ambiguity component generated in the SAR images (1) and (2), although the frequency band of the azimuth ambiguity component differs from the clutter component. This is because azimuth resampling similar to that of the clutter component is performed.
However, since the Doppler frequency band limitation less than the PRF is actually performed with the bandwidth _BD , the signal components are as shown in the lower diagram of FIG. However, also in this case, the azimuth ambiguity component resulting from the sidelobe pattern of the antenna remains.

The difference image generation unit 102 suppresses clutter components included in the SAR images (1) and (2) by taking the difference between the SAR image (1) and the registered SAR image (2) ( In step ST102, a difference image between the two images is output (step ST103).
The signal component of the difference image between the SAR image (1) and the registered SAR image (2) is expressed by the following equation (3).

If the DPCA condition is satisfied when L = 0, the moving target component can be confirmed from the difference image without the azimuth ambiguity components 112 and 113 remaining.
However, if L ≠ 0 and the DPCA condition is not satisfied, the azimuth ambiguity component remains as shown in FIG.

CE Livingstone, I. Sikaneta, CH Gierull, S. Chiu, A. Beaudoin, J. ampbell,. Beaudoin, S. Gong, and TA Knight, “An airborne synthetic aperture radar (SAR) experiment to support RADARSAT-2ground moving target indication (GMTI) ”, Can. J. Remote Sensing, Vol. 28, No. 6, pp. 794-813, 2002

  Since the conventional radar signal processing apparatus is configured as described above, when the DPCA condition that the phase centers of the plurality of antennas that receive the reflected waves of the pulse reflected by the target overlap is satisfied, Taking the difference between the SAR image (1) and the registered SAR image (2), the azimuth ambiguity component can be removed. However, if the DPCA condition is not satisfied, the moving target can be detected because the azimuth ambiguity component remains even if the difference between the SAR image (1) and the registered SAR image (2) is taken. There was a problem that could be impossible.

  The present invention has been made to solve the above-described problems. Even when the DPCA condition is not satisfied, the moving target can be detected by suppressing both the clutter component and the azimuth ambiguity component. An object of the present invention is to obtain a radar signal processing device capable of performing the above.

  A radar signal processing apparatus according to the present invention includes an alignment unit that aligns a plurality of synthetic aperture radar images having the same observation region and different observation times, and a plurality of synthetic aperture radars that are aligned by the alignment unit A signal component classification unit that classifies signal components in the Doppler frequency domain in an image into a positive region and a negative region; The clutter component and the azimuth ambiguity component contained in the signal component classified into the positive region are suppressed and included in the signal component separated into the negative region by the signal component separation means. Clutter component and azimuth ambiguity component are estimated and included in the signal component sorted into the negative region. An unnecessary component suppression unit that suppresses the clutter component and the azimuth ambiguity component is provided. It is intended to be detected.

  According to the present invention, the signal component classification unit that classifies the signal component in the Doppler frequency domain in the plurality of synthetic aperture radar images that have been aligned by the registration unit into the positive region and the negative region, and the signal component classification unit Clutter component and azimuth ambiguity component included in the signal component classified into the positive region by estimating the clutter component and azimuth ambiguity component included in the signal component classified into the positive region And the clutter component and the azimuth ambiguity component included in the signal component classified into the negative region by the signal component classification means are estimated and included in the signal component classified into the negative region. Unnecessary component suppression means for suppressing the existing clutter component and azimuth ambiguity component. Even if the phase centers of the antenna do not overlap, by suppressing both of the clutter components and azimuth ambiguity components, there is an effect capable of detecting a moving target.

BRIEF DESCRIPTION OF THE DRAWINGS It is a block diagram which shows the radar signal processing apparatus by Embodiment 1 of this invention. It is a flowchart which shows the processing content of the radar signal processing apparatus by Embodiment 1 of this invention. It is explanatory drawing which shows a spectrum when a 0 component is given to the negative area | region of the Doppler frequency area | region as a signal component. It is explanatory drawing which shows a spectrum when a 0 component is given to the positive area | region of the Doppler frequency area | region as a signal component. It is explanatory drawing which shows the image which has the signal component by which the clutter component and the azimuth ambiguity component were suppressed by the unnecessary component suppression part. It is a block diagram which shows the radar signal processing apparatus by Embodiment 2 of this invention. It is explanatory drawing which shows the example in which two types of azimuth ambiguity components will be mixed even if it isolate | separates a signal component in a Doppler frequency domain. It is a block diagram which shows the radar signal processing apparatus by Embodiment 3 of this invention. 1 is a configuration diagram illustrating a radar signal processing device disclosed in Non-Patent Document 1. FIG. It is a flowchart which shows the processing content of the radar signal processing apparatus of FIG. It is explanatory drawing which shows the example in which the phase center of transmission / reception between two channels overlaps. It is explanatory drawing which shows the example in which the phase center of transmission / reception between two channels does not overlap. It is explanatory drawing which shows the various components in the received signal of a SAR image. It is explanatory drawing which shows the difference image in which the azimuth ambiguity component remains.

Embodiment 1 FIG.
1 is a block diagram showing a radar signal processing apparatus according to Embodiment 1 of the present invention.
In FIG. 1, when a registration unit 1 is provided with a plurality of synthetic aperture radar images SAR image (1) and SAR image (2) having the same observation area and different observation times, the registration unit 1 determines the position of the SAR image (2). Registration is performed to match the position of the SAR image (1). The registration unit 1 constitutes an alignment unit.

The signal component estimation unit 2 includes a signal component classification unit 3 and an unnecessary component suppression unit 4, and performs a process of outputting a SAR image (1) and a SAR image (2) in which clutter components and azimuth ambiguity components are suppressed. carry out.
The signal component classification unit 3 performs a process of classifying the signal components in the Doppler frequency region in the SAR images (1) and (2), which are aligned by the registration unit 1, into a positive region and a negative region. The signal component sorting unit 3 constitutes signal component sorting means.

The unnecessary component suppression unit 4 assumes a component other than the clutter component and the azimuth ambiguity component included in the signal component classified into the positive area by the signal component classification unit 3, and the clutter component and the azimuth ambiguity. By estimating the component, the clutter component and the azimuth ambiguity component contained in the signal component classified into the positive region are suppressed, and the signal component classified into the negative region by the signal component classification unit 3 is suppressed. Assuming components other than the included clutter component and azimuth ambiguity component, and estimating the clutter component and azimuth ambiguity component, the clutter included in the signal component classified into the negative region The process which suppresses a component and an azimuth ambiguity component is implemented. The unnecessary component suppressing unit 4 constitutes an unnecessary component suppressing unit.
The target detection unit 5 performs processing for detecting a moving target from an image having a signal component in which the clutter component and the azimuth ambiguity component are suppressed by the unnecessary component suppression unit 4. The target detecting unit 5 constitutes a target detecting unit.

In the example of FIG. 1, each of the registration unit 1, the signal component estimation unit 2, and the target detection unit 5 that are components of the radar signal processing device has dedicated hardware (for example, a semiconductor integrated circuit in which a CPU is mounted, Alternatively, a one-chip microcomputer or the like is assumed, but the radar signal processing apparatus may be configured with a computer.
When the radar signal processing apparatus is configured by a computer, a program describing the processing contents of the registration unit 1, the signal component estimation unit 2, and the target detection unit 5 is stored in the memory of the computer, and the CPU of the computer stores the memory. The program stored in the program may be executed.
FIG. 2 is a flowchart showing the processing contents of the radar signal processing apparatus according to Embodiment 1 of the present invention.

Next, the operation will be described.
In this first embodiment, two SAR images (1) and (2) are received by receiving a reflected wave of the pulse, which is repeatedly radiated from one transmitting antenna and reflected by two receiving antennas. And the moving target is detected from the two SAR images (1) and (2).
At this time, the distance between the two receiving antennas is expressed by 2 mV _p / PRF + 2LV _p / PRF with respect to the platform speed V _p . PFR (Pulse Repetition Frequency) is a pulse repetition frequency, m integer, L is a real number, and -1 <L <0 or 0 <L <1.
In the case of L = 0, as shown in FIG. 11, the DPCA condition in which the transmission / reception antenna phase centers overlap is satisfied, but in this embodiment 1, since L ≠ 0, as shown in FIG. In addition, the DPCA condition is not satisfied.

When the SAR image (1) and the SAR image (2) having the same observation area and different observation times are given, the registration unit 1 registers the position of the SAR image (2) with the position of the SAR image (1). (Step ST1 in FIG. 2).
At this time, the registration unit 1 performs azimuth resampling in order to correct an image shift of (m + L) pixels in the azimuth direction. As a result, the received signal r ₁ (p, q) of the SAR image (1) in the p-th range bin and the q-th azimuth bin is described as follows.

In Equation (4), s (p, q) is a clutter component 111 as shown in FIG. 13, and a ₊ (p, q) is a primary azimuth ambiguity component 112 as shown in FIG. a ₋ (p, q) is a primary azimuth ambiguity component 113 as shown in FIG. 13.
The primary azimuth ambiguity component is generated by the sidelobe pattern of the antenna, and the secondary and subsequent components are assumed to be negligibly small.
n ₁ (p, q) is a noise component in the SAR image (1), and t ₁ (p, q) is a moving target component in the SAR image (1).
Note that there are P range bins and Q azimuth bins.

式（５）において、ｎ_２（ｐ，ｑ）はＳＡＲ画像（２）内の雑音成分、ｔ_２（ｐ，ｑ）はＳＡＲ画像（２）内の移動目標成分である。 On the other hand, the received signal r ₂ (p, q) of the SAR image (2) in the p-th range bin and the q-th azimuth bin is described as follows.

In Expression (5), n ₂ (p, q) is a noise component in the SAR image (2), and t ₂ (p, q) is a moving target component in the SAR image (2).

When the signal component estimation unit 2 receives the SAR image (1) and the SAR image (2) that have been aligned by the registration unit 1, the signal component estimation unit 2 is included in the SAR image (1) and the SAR image (2). Estimate clutter component and azimuth ambiguity component, suppress clutter component and azimuth ambiguity component, and output SAR image (1) and SAR image (2) with suppressed clutter component and azimuth ambiguity component To do.
Hereinafter, the processing content of the signal component estimation part 2 is demonstrated concretely.

Two types of azimuth ambiguity components are mixed in the reception signal r ₁ (p, q) of the SAR image (1) and the reception signal r ₂ (p, q) of the SAR image (2) after registration. Yes.
In order to simplify this problem, the signal component classification unit 3 of the signal component estimation unit 2 classifies the signal component in the Doppler frequency domain in the SAR image (1) into a positive region and a negative region, and the SAR image (2). The signal component in the Doppler frequency region at is separated into a positive region and a negative region (step ST2).

The signal components of 2P Doppler frequencies in the SAR image (1) and the SAR image (2) after registration are expressed as shown in the lower diagram of FIG.
When this signal component is copied into two, one signal component is given a zero sequence in the negative region of the Doppler frequency region, and the other signal component is given a zero sequence in the positive region of the Doppler frequency region, the former The spectrum of FIG. 3 is as shown in FIG. 3, and the latter spectrum is as shown in FIG.
When the above spectrum is returned to the time domain, the received signal r ₁₊ (p, q) of the SAR image (1) in which the zero sequence is given to the negative region of the Doppler frequency domain and the SAR image after registration (2) The received signal r ₂₊ (p, q) of the SAR image (1) is expressed by the following formulas (6) and (7), and the zero sequence is given to the positive region of the Doppler frequency region. r _1- (p, q) and the received signal r _2- (p, q) of the SAR image (2) after registration are expressed by the following equations (8) and (9).

  In the expressions (6) to (9), the subscript + added to the clutter component, the moving target component, and the noise component indicates that the 0 series is given to the negative region of the Doppler frequency region as in the case of the received signal. The subscript-indicates that the signal is obtained by giving a zero sequence to the positive region of the Doppler frequency region. By performing such conversion, the estimation problem can be simplified.

The unnecessary component suppressing unit 4 uses the received signals r ₁₊ (p, q), r ₂₊ (p, q) to prevent the clutter component s ₊ (p, q) and azimuth from being included as much as possible. In addition to estimating the ambiguity component a ₊ (p, q) and using the received signals r ₁₋ (p, q), r ₂₋ (p, q), a moving target component is included as much as possible. The clutter component s ₋ (p, q) and the azimuth ambiguity component a ₋ (p, q) are estimated (step ST3).
Hereinafter, a process of estimating the clutter component s ₊ (p, q) and the azimuth ambiguity component a ₊ (p, q) using the received signal r ₁₊ (p, q), r ₂₊ (p, q). explain.

First, the formulation of the received signals r ₁₊ (p, q), r ₂₊ (p, q) is organized.
The formula of the received signal r ₁₊ (p, q), r ₂₊ (p, q) is expressed as a matrix and a vector as follows.

A least square method is known as one of signal estimation methods formulated as in Expression (10) (for example, Non-Patent Document 2).
[Non-Patent Document 2]
MK Steven, “Fundamentals of Statistical Signal Processing: Estimation Theory”, Prentice Hall PTR Prentice-Hall, Inc., 1993.
A specific estimation method based on the least square method is expressed by an equation (16) below.

式（１６）において、Ｓの文字の上に付されている“−”は、推定結果であることを表している。以下、他の文字の上に“−”が付されている場合も、同様に推定結果であることを表している。
また、式（１６）において、^−１の添え字は逆行列を表し、^Ｈの添え字は共役転置を表している。

In Expression (16), “-” added on the letter S represents an estimation result. Hereinafter, the case where “-” is added on other characters also indicates the estimation result.
In the equation (16), the subscript ⁻¹ represents an inverse matrix, and the subscript ^H represents a conjugate transpose.

However, when estimating by the least square method, noise components n ₁₊ (p, q), n ₂₊ (p, q) other than the clutter component s ₊ (p, q) and the azimuth ambiguity component a ₊ (p, q) ) And the moving target component t ₁₊ (p, q) are not assumed.
Therefore, for example, in order to suppress the clutter component s ₊ (p, q) and the azimuth ambiguity component a ₊ (p, q) included in the received signal r ₁₊ (p, q), the clutter component s _{By using +} (p, q) and the azimuth ambiguity component a ₊ (p, q), r ₁₊ (p, q) −s ₊ (p, q) −a ₊ (p, q) is calculated Is also 0, and the moving target component t ₁₊ (p, q) does not remain.
Therefore, the method of estimating by the least square method cannot be applied, and an estimation method that assumes the existence of signals other than the clutter component s ₊ (p, q) and the azimuth ambiguity component a ₊ (p, q) is assumed. It is necessary to use it.
Therefore, in the first embodiment, it is assumed that a signal is estimated using a Wiener filter.

ただし、Ｅ（ａ_＋／ｓ_＋）はアジマスアンビギュイティ成分ａ_＋（ｐ，ｑ）とクラッタ成分ｓ_＋（ｐ，ｑ）の電力比の期待値であり、Ｅ（ｎ_＋／ｓ_＋）は雑音成分ｎ_＋（ｐ，ｑ）とクラッタ成分ｓ_＋（ｐ，ｑ）の電力比の期待値である。 The estimation method using the Wiener filter is expressed by the following equation (17).

However, E (a ₊ / s ₊ ) is an expected value of the power ratio between the azimuth ambiguity component a ₊ (p, q) and the clutter component s ₊ (p, q), and E (n ₊ / s ₊ ) Is an expected value of the power ratio between the noise component n ₊ (p, q) and the clutter component s ₊ (p, q).

This estimation method assumes the presence of signals other than the clutter component s ₊ (p, q) and the azimuth ambiguity component a ₊ (p, q), and the presence of these signals is determined by the clutter component s ₊ (p, q). This is a method in which the amount contained in the estimation of q) and the azimuth ambiguity component a ₊ (p, q) is statistically minimized.
Therefore, when r ₁ + (p, q) −s ₊ (p, q) −a ₊ (p, q) is calculated, the moving target component t ₁₊ (p, q) remains.

Similarly, the received signals r _1- (p, q) and r _2- (p, q) are given as follows when expressed in matrix and vector.

An estimation method using a Wiener filter is as shown in the following equation (26).

The unnecessary component suppression unit 4 uses a Wiener filter to perform the clutter components s ₊ (p, q), s ₋ (p, q) and the azimuth ambiguity components a ₊ (p, q), a ₋ (p, q ) Is used to suppress the clutter component and the azimuth ambiguity component included in the received signal r ₁ (p, q) of the SAR image (1) using the estimation results, and to reduce the SAR image (2 The clutter component and the azimuth ambiguity component included in the received signal r ₂ (p, q) are suppressed (step ST4).
There are several possible variations of the clutter component and azimuth ambiguity component suppression methods.
The simplest suppression method is to perform the following calculation.

It is possible to detect a moving target by using the calculation result of Expression (29) or the calculation result of Expression (30). However, as shown in the following Expression (31), if two-channel components are synthesized, Thus, the detection performance of the moving target can be improved.

Note that the clutter components s ₊ (p, q), s ₋ (p, q) and the azimuth ambiguity components a ₊ (p, q), a ₋ (p, q) estimated by the unnecessary component suppression unit 4 are used. Contains the estimation error.
In particular, it is considered that the clutter component estimation error greatly contributes to the degradation of the detection performance of the moving target because the power of the clutter component is large.
Therefore, the clutter is removed without using the clutter components s ₊ (p, q) and s ₋ (p, q) estimated by the unnecessary component suppression unit 4, and the remaining azimuth ambiguity components a ₊ (p, q ), A ₋ (p, q) can be considered as follows.

Although the clutter component and the azimuth ambiguity component suppression by the unnecessary component suppression unit 4 is performed in the amplitude dimension, as shown below, it is performed in the power dimension and then returned to the amplitude dimension. It is also possible to improve the detection performance of the moving target.

FIG. 5 is an explanatory diagram showing an image having a signal component in which the clutter component and the azimuth ambiguity component are suppressed by the unnecessary component suppression unit 4.
In the example of FIG. 5, four moving target components are represented, and the clutter component and the azimuth ambiguity component are suppressed.
The target detection unit 5 detects the moving target from an image having a signal component in which the clutter component and the azimuth ambiguity component are suppressed by the unnecessary component suppression unit 4 (hereinafter referred to as “post-suppression image”) (step ST5). .
As the moving target detection process, a threshold determination process or the like can be considered. However, the threshold determination process may be performed after the suppressed image is converted into a power component. In order to reduce noise, a threshold determination process may be performed after applying a moving average filter or the like.

式（３７）において、＊は共役を示す記号である。 It can also be used for ATI (Along Track Interferometry) that estimates the speed of a moving target by causing the SAR image (1) and the SAR image (2) to interfere with each other.
Specifically, a phase map with the following declination is calculated.

In the formula (37), * is a symbol indicating conjugation.

  As is apparent from the above, according to the first embodiment, the signal components in the Doppler frequency region in the SAR images (1) and (2) that have been aligned by the registration unit 1 are positive and negative regions. Assuming components other than the clutter component and the azimuth ambiguity component included in the signal component sorted into the positive region by the signal component sorting unit 3, the clutter component and By estimating the azimuth ambiguity component, the clutter component and the azimuth ambiguity component included in the signal component classified into the positive region are suppressed, and the signal component separation unit 3 separates the azimuth ambiguity component into the negative region. Assuming components other than the clutter component and azimuth ambiguity component included in the received signal component, the clutter component and azimuth And the unnecessary component suppressing unit 4 for suppressing the clutter component and the azimuth ambiguity component included in the signal component classified into the negative region by estimating the nbiguity component. Even when the phase centers do not overlap, it is possible to suppress both the clutter component and the azimuth ambiguity component and to detect the moving target.

In the first embodiment, as shown in the above equations (18) and (19), the expected value E of the power ratio between the azimuth ambiguity component a ₊ (p, q) and the clutter component s ₊ (p, q) Using the (a ₊ / s ₊ ) and the expected value E (n ₊ / s ₊ ) of the power ratio of the noise component n ₊ (p, q) and the clutter component s ₊ (p, q), the Wiener filter is used. As shown in equations (27) and (28), the expected value E (a ₋ / s) of the power ratio between the azimuth ambiguity component a ₋ (p, q) and the clutter component s ₋ (p, q) _- a), the noise component _n - (p, q) and clutter components _s - (p, the expected value E of the power ratio q) _(n - / s _-) and was used to produce a Wiener filter, their using Wiener filters, clutter component _{s + (p, q),} s - (p, q) and Ajimasua Bigyuiti components _{a + (p, q),} a - (p, q) showed to estimate, and the generation process of the estimation process and the Wiener filter repeatedly performed recursively to enhance the estimation accuracy It may be.

In other words, the unnecessary component suppression unit 4 uses the Wiener filter generated using the expected values E (a ₊ / s ₊ ) and E (n ₊ / s ₊ ) to generate the clutter component s ₊ (p, q) and azimuth. ambiguity components _a + (p, q) After estimating the azimuth ambiguity components _a + (p, q) obtained by the estimated clutter component _s + (p, q) power ratio _{(a +} / _s + ) And the estimated power ratio (n ₊ / s ₊ ) of the clutter component s ₊ (p, q) and the noise component n ₊ (p, q) to regenerate the Wiener filter The clutter component s ₊ (p, q) and the azimuth ambiguity component a ₊ (p, q) are estimated using the Wiener filter.
Similarly, using a Wiener filter generated using the expected values E (a ₋ / s ₋ ) and E (n ₋ / s ₋ ), the clutter component s ₋ (p, q) and the azimuth ambiguity component a ₋ After estimating (p, q), the estimated power ratio (a ₋ / s ₋ ) between the estimated azimuth ambiguity component a ₋ (p, q) and clutter component s ₋ (p, q) is estimated. Using the power ratio (n ₋ / s ₋ ) of the clutter component s ₋ (p, q) and the noise component n ₋ (p, q), the Wiener filter is regenerated, and the regenerated Wiener filter is used. The clutter component s ₋ (p, q) and the azimuth ambiguity component a ₋ (p, q) are estimated.

When the unnecessary component suppression unit 4 recursively repeats the estimation processing of the azimuth ambiguity component and the like and the Wiener filter generation processing, for example, whether or not the estimated clutter component and azimuth ambiguity component have converged is determined. The estimation process and the generation process may be repeated until the estimated clutter component and the azimuth ambiguity component converge, or the estimation process and the generation process may be repeated a preset number of times. It may be.
In addition, as the convergence determination of the clutter component and the azimuth ambiguity component, for example, the previous estimation result and the current estimation result are compared, and if the difference between these estimation results is within a preset threshold, the convergence is achieved. It can be considered to determine that it is.

In the first embodiment, the unnecessary component suppressing unit 4 generates the Wiener filter by using the expected values E (a ₊ / s ₊ ) and E (n ₊ / s ₊ ). If the signal power of the clutter component is dominant (if the ratio of the signal power of the clutter component is greater than or equal to a preset threshold value), the signal power of the clutter component is used to calculate the signal power of the clutter component. A filter may be generated.
That is, expected values E (a ₊ / s ₊ ) and E (n ₊ / s ₊ ) in the above equations (18) and (19), and expected values E (a ₋ / s _{− in} equations (27) and (28). ), E (n ₋ / s ₋ ), a winner filter may be generated by substituting a value inversely proportional to the signal power of the clutter component.
Also in this case, the estimation accuracy of the azimuth ambiguity component and the like and the Wiener filter generation processing may be recursively repeated to improve the estimation accuracy.

That is, the unnecessary component suppression unit 4 estimates the clutter component s ₊ (p, q) and the azimuth ambiguity component a ₊ (p, q) using the Wiener filter generated using the signal power of the clutter component. After that, the power ratio (a ₊ / s ₊ ) of the estimated azimuth ambiguity component a ₊ (p, q) and the clutter component s ₊ (p, q) and the estimated clutter component s ₊ (p, q ) And the power ratio (n ₊ / s ₊ ) of the noise component n ₊ (p, q) to regenerate the Wiener filter, and use the regenerated Wiener filter to generate the clutter component s ₊ (p, q) and the azimuth ambiguity component a ₊ (p, q) are estimated.
Similarly, the clutter component s ₋ (p, q) and the azimuth ambiguity component a ₋ (p, q) are estimated using the Wiener filter generated using the signal power of the clutter component, and then the estimated azimuth is estimated. The power ratio (a ₋ / s ₋ ) between the ambiguity component a ₋ (p, q) and the clutter component s ₋ (p, q), the estimated clutter component s ₋ (p, q), and the noise component n ₋ The Wiener filter is regenerated using the power ratio (n ₋ / s ₋ ) of (p, q), and the clutter component s ₋ (p, q) and the azimuth ambiguity are regenerated using the regenerated Wiener filter. The tee component a ₋ (p, q) is estimated.
When the unnecessary component suppression unit 4 recursively repeats the estimation processing of the azimuth ambiguity component and the like and the Wiener filter generation processing, for example, whether or not the estimated clutter component and azimuth ambiguity component have converged is determined. The estimation process and the generation process may be repeated until the estimated clutter component and the azimuth ambiguity component converge, or the estimation process and the generation process may be repeated a preset number of times. It may be.

式（３８）において、Ｉは単位行列、λは任意の実数である。 In the first embodiment, the unnecessary component suppressing unit 4 uses a Wiener filter to perform the clutter components s ₊ (p, q), s ₋ (p, q) and the azimuth ambiguity component a ₊ (p, q). , A ₋ (p, q) is shown, but the clutter components s ₊ (p, q), s ₋ (p, q) and the azimuth ambiguity are obtained using a filter that performs the regularized least squares method. The tee components a ₊ (p, q), a ₋ (p, q) may be estimated. A filter that performs the regularized least squares method is represented by the following equation (38).

In Equation (38), I is a unit matrix, and λ is an arbitrary real number.

Here, λ is an arbitrary real number, but instead of λ, an expected value E (n ₊ ) of the power ratio between the noise component n ₊ (p, q) and the clutter component s ₊ (p, q) is shown. / S ₊ ) (or the expected value E (n ₋ / s ₋ ) of the power ratio between the noise component n ₋ (p, q) and the clutter component s ₋ (p, q)) is substituted into the equation (38). A filter that performs the regularized least squares method may be generated.
In addition, when the signal power of the pixel component in the signal component is calculated and the power of the clutter component is dominant (when the ratio of the signal power of the clutter component is equal to or larger than a preset threshold value), the clutter is substituted for λ. A filter for performing the regularized least square method may be generated by substituting the signal power of the component into the equation (38).

In the first embodiment, the unnecessary component suppression unit 4 uses a Wiener filter or a filter that performs a regularized least squares method to generate clutter components s ₊ (p, q), s ₋ (p, q) and azimuth ambiguity. Although the estimation of the tee components a ₊ (p, q), a ₋ (p, q) is shown, the present invention is not limited to this, and a filter or estimation method that assumes the presence of other signal components of the component to be estimated It is possible to apply.
In the first embodiment, an example in which two reception antennas are used has been described. However, a case in which a plurality of transmission antennas are used can also be handled.

Embodiment 2. FIG.
In the second embodiment, pulses are repeatedly radiated from one transmitting antenna, and three receiving antennas receive the reflected waves of the pulse reflected by the target, so that three SAR images (1) and (2) are received. A radar signal processing apparatus that acquires (3) and detects a moving target from three SAR images (1), (2), and (3) will be described.
At this time, the distance between the receiving antenna corresponding to the SAR image (1) and the receiving antenna corresponding to the SAR image (2) is 2m ₂ V _p / PRF + 2L ₂ V _p / PRF with respect to the platform speed V _p . Shall be represented. In addition, the distance between the reception antenna corresponding to the SAR image (1) and the reception antenna corresponding to the SAR image (3) is expressed by 2m ₃ V _p / PRF + 2L ₃ V _p / PRF.
PFR is a pulse repetition frequency, m ₂ and m ₃ are integers, L ₂ and L ₃ are real numbers, and 0 <L ₂ and L ₃ <1.

FIG. 6 is a block diagram showing a radar signal processing apparatus according to Embodiment 2 of the present invention.
In FIG. 6, when the SAR image (1) and the SAR image (2) having the same observation area and different observation times are given, the registration unit 11a changes the position of the SAR image (2) to the position of the SAR image (1). Perform registration to match.
When the SAR image (1) and the SAR image (3) having the same observation area and different observation times are given, the registration unit 11b performs registration for aligning the position of the SAR image (3) with the position of the SAR image (1). carry out. Note that the registration units 11a and 11b constitute positioning means.

The signal component estimation unit 12 includes a signal component classification unit 13 and an unnecessary component suppression unit 14, and performs a process of outputting SAR images (1), (2), and (3) in which clutter components and azimuth ambiguity components are suppressed. carry out.
The signal component classification unit 13 performs a process of classifying the signal components in the Doppler frequency region in the SAR images (1), (2), and (3) that have been aligned by the registration units 11a and 11b into a positive region and a negative region. carry out. The signal component classification unit 13 constitutes signal component classification means.

The unnecessary component suppression unit 14 assumes a component other than the clutter component and the azimuth ambiguity component included in the signal component classified into the positive region by the signal component classification unit 13, and the clutter component and the azimuth ambiguity are assumed. By estimating the component, the clutter component and the azimuth ambiguity component included in the signal component sorted into the positive region are suppressed, and the signal component sorted into the negative region by the signal component sorting unit 13 is suppressed. Assuming components other than the included clutter component and azimuth ambiguity component, and estimating the clutter component and azimuth ambiguity component, the clutter included in the signal component classified into the negative region The process which suppresses a component and an azimuth ambiguity component is implemented. The unnecessary component suppression unit 14 constitutes an unnecessary component suppression unit.
The target detection unit 15 performs a process of detecting a moving target from an image having a signal component in which the clutter component and the azimuth ambiguity component are suppressed by the unnecessary component suppression unit 14. The target detection unit 15 constitutes a target detection unit.

In the example of FIG. 6, each of the registration units 11 a and 11 b, the signal component estimation unit 12, and the target detection unit 15, which are components of the radar signal processing device, is dedicated hardware (for example, a semiconductor integrated with a CPU mounted). It is assumed that the circuit is configured by a circuit or a one-chip microcomputer, but the radar signal processing apparatus may be configured by a computer.
When the radar signal processing apparatus is configured by a computer, a program describing the processing contents of the registration units 11a and 11b, the signal component estimation unit 12 and the target detection unit 15 is stored in the memory of the computer, and the CPU of the computer The program stored in the memory may be executed.

Next, the operation will be described.
In the second embodiment, when three SAR images (1), (2), and (3) having the same observation area and different observation times are given, the following processing is performed based on the SAR image (1). To do.
When the SAR image (1) and the SAR image (2) are given, the registration unit 11a performs registration that matches the position of the SAR image (2) with the position of the SAR image (1).
In addition, when the SAR image (1) and the SAR image (3) are given, the registration unit 11a performs registration that matches the position of the SAR image (3) with the position of the SAR image (1).
At this time, the registration unit 11a performs azimuth resampling in order to correct an image shift of (m ₂ + L ₂ ) pixels in the azimuth direction.
Further, the registration unit 11b performs azimuth resampling in order to correct an image shift of (m ₃ + L ₃ ) pixels in the azimuth direction.
As a result, the received signal r ₁ (p, q) of the SAR image (1) in the p-th range bin and the q-th azimuth bin is described as follows.

In the equation (39), s (p, q) is a clutter component 111 as shown in FIG. 13, and a ₊ (p, q) is a primary azimuth ambiguity component 112 as shown in FIG. a ₋ (p, q) is a primary azimuth ambiguity component 113 as shown in FIG. 13.
The primary azimuth ambiguity component is generated by the sidelobe pattern of the antenna, and the secondary and subsequent components are assumed to be negligibly small.
n ₁ (p, q) is a noise component in the SAR image (1), and t ₁ (p, q) is a moving target component in the SAR image (1).
Note that there are P range bins and Q azimuth bins.

式（４０）において、ｎ_２（ｐ，ｑ）はＳＡＲ画像（２）内の雑音成分、ｔ_２（ｐ，ｑ）はＳＡＲ画像（２）内の移動目標成分である。 On the other hand, the received signal r ₂ (p, q) of the SAR image (2) in the p-th range bin and the q-th azimuth bin is described as follows.

In Expression (40), n ₂ (p, q) is a noise component in the SAR image (2), and t ₂ (p, q) is a moving target component in the SAR image (2).

式（４１）において、ｎ_３（ｐ，ｑ）はＳＡＲ画像（３）内の雑音成分、ｔ_３（ｐ，ｑ）はＳＡＲ画像（３）内の移動目標成分である。 The received signal r ₃ (p, q) of the SAR image (3) in the p-th range bin and the q-th azimuth bin is described as follows.

In Expression (41), n ₃ (p, q) is a noise component in the SAR image (3), and t ₃ (p, q) is a moving target component in the SAR image (3).

When the signal component estimation unit 12 receives the SAR images (1), (2), and (3) that have been aligned by the registration units 11a and 11b, the signal component estimation unit 12 is included in the SAR images (1), (2), and (3). The clutter component and the azimuth ambiguity component are estimated, the clutter component and the azimuth ambiguity component are suppressed, and the clutter component and the azimuth ambiguity component are suppressed (1) (2) (3) Is output.
Hereinafter, the processing content of the signal component estimation part 12 is demonstrated concretely.

Received signal r ₁ (p, q) of SAR image (1), received signal r ₂ (p, q) of SAR image (2) after registration, and received signal r ₃ of SAR image (3) after registration Two types of azimuth ambiguity components are mixed in (p, q).
In order to simplify this problem, the signal component classification unit 13 of the signal component estimation unit 12 classifies the signal components in the Doppler frequency region in the SAR image (1) into a positive region and a negative region, and the SAR image (2). The signal component in the Doppler frequency domain at is divided into a positive area and a negative area, and the signal component in the Doppler frequency area in the SAR image (3) is classified into a positive area and a negative area.

For example, signal components of 2P Doppler frequencies in the SAR image (1) and the registered SAR image (2) are expressed as shown in the lower diagram of FIG.
When this signal component is copied into two, one signal component is given a zero sequence in the negative region of the Doppler frequency region, and the other signal component is given a zero sequence in the positive region of the Doppler frequency region, the former The spectrum of FIG. 3 is as shown in FIG. 3, and the latter spectrum is as shown in FIG.
When the above spectrum is returned to the time domain, the received signal r ₁₊ (p, q) of the SAR image (1) in which the zero sequence is given to the negative region of the Doppler frequency region, the SAR image (2) after registration The received signal r ₂₊ (p, q) and the received signal r ₃₊ (p, q) of the registered SAR image (3) are expressed by the following equations (42) to (44), and the Doppler frequency The received signal r _1- (p, q) of the SAR image (1) in which a zero sequence is given to the positive area of the area, and the received signal r _2- (p, q) of the SAR image (2) after registration And the received signal r ₃₋ (p, q) of the SAR image (3) after registration is expressed as the following equations (45) to (47).

  In the expressions (42) to (47), the subscript + added to the clutter component, the moving target component, and the noise component indicates that the 0 series is given to the negative region of the Doppler frequency region, similarly to the received signal. The subscript-indicates that the signal is obtained by giving a zero sequence to the positive region of the Doppler frequency region. By performing such conversion, the estimation problem can be simplified.

The unnecessary component suppression unit 14 uses the received signals r ₁₊ (p, q), r ₂₊ (p, q), r ₃₊ (p, q) to prevent the clutter component s from being included as much as possible. ₊ (P, q) and the azimuth ambiguity component a ₊ (p, q) are estimated, and the received signals r _1- (p, q), r _2- (p, q), r _3- (p, The clutter component s ₋ (p, q) and the azimuth ambiguity component a ₋ (p, q) are estimated using q) so that the moving target component is not included as much as possible.
Hereinafter, the clutter component s ₊ (p, q) and the azimuth ambiguity component a ₊ (p) are obtained using the received signals r ₁₊ (p, q), r ₂₊ (p, q), r ₃₊ (p, q). , Q) will be described.

First, the formulation of the received signals r ₁₊ (p, q), r ₂₊ (p, q), r ₃₊ (p, q) is organized.
When the _{formulas of the} received signals r ₁₊ (p, q), r ₂₊ (p, q), r ₃₊ (p, q) are expressed by matrices and vectors, they are given as follows.

ただし、Ｅ（ａ_＋／ｓ_＋）はアジマスアンビギュイティ成分ａ_＋（ｐ，ｑ）とクラッタ成分ｓ_＋（ｐ，ｑ）の電力比の期待値であり、Ｅ（ｎ_＋／ｓ_＋）は雑音成分ｎ_＋（ｐ，ｑ）とクラッタ成分ｓ_＋（ｐ，ｑ）の電力比の期待値である。 In the second embodiment, for the same reason as in the first embodiment, a signal is estimated using a Wiener filter.
The estimation method using the Wiener filter is represented by the following equation (54).

However, E (a ₊ / s ₊ ) is an expected value of the power ratio between the azimuth ambiguity component a ₊ (p, q) and the clutter component s ₊ (p, q), and E (n ₊ / s ₊ ) Is an expected value of the power ratio between the noise component n ₊ (p, q) and the clutter component s ₊ (p, q).

This estimation method assumes the presence of signals other than the clutter component s ₊ (p, q) and the azimuth ambiguity component a ₊ (p, q), and the presence of these signals is determined by the clutter component s ₊ (p, q). This is a method in which the amount contained in the estimation of q) and the azimuth ambiguity component a ₊ (p, q) is statistically minimized.
Therefore, when r ₁₊ (p, q) −s ₊ (p, q) −a ₊ (p, q) is calculated, the moving target component t ₁₊ (p, q) remains.

Similarly, the received signals r _1- (p, q), r _2- (p, q), r _3- (p, q) are given as follows when expressed in matrix and vector.

The estimation method using the Wiener filter is as shown in the following equation (63).

The unnecessary component suppression unit 14 uses a Wiener filter to perform the clutter components s ₊ (p, q), s ₋ (p, q) and the azimuth ambiguity components a ₊ (p, q), a ₋ (p, q ) Is used to suppress the clutter component and the azimuth ambiguity component included in the received signal r ₁ (p, q) of the SAR image (1) using the estimation results, and to reduce the SAR image (2 The clutter component and the azimuth ambiguity component contained in the received signal r ₂ (p, q) are suppressed.
Moreover, the clutter component and the azimuth ambiguity component contained in the received signal r ₃ (p, q) of the SAR image (3) are suppressed.
The simplest suppression method is to perform the following calculation.

The movement target can be detected by using the calculation result of Expression (66), the calculation result of Expression (67), or the calculation result of Expression (68), but as shown in Expression (69) below, If the three-channel components are combined, the detection performance of the moving target can be further enhanced.

Note that the clutter components s ₊ (p, q), s ₋ (p, q) and the azimuth ambiguity components a ₊ (p, q), a ₋ (p, q) estimated by the unnecessary component suppression unit 14 are used. Contains the estimation error.
In particular, it is considered that the clutter component estimation error greatly contributes to the degradation of the detection performance of the moving target because the power of the clutter component is large.
Therefore, the clutter is removed without using the clutter components s ₊ (p, q) and s ₋ (p, q) estimated by the unnecessary component suppression unit 14, and the remaining azimuth ambiguity components a ₊ (p, q ), A ₋ (p, q) can be considered as follows.

Although the clutter component and the azimuth ambiguity component suppression by the unnecessary component suppression unit 14 is performed in the amplitude dimension, as shown below, the suppression is performed in the power dimension and then back to the amplitude dimension. It is also possible to improve the detection performance of the moving target.

The target detection unit 15 detects a moving target from the post-suppression image that is an image having a signal component in which the clutter component and the azimuth ambiguity component are suppressed by the unnecessary component suppression unit 14.
As the moving target detection process, a threshold determination process or the like can be considered. However, the threshold determination process may be performed after the suppressed image is converted into a power component. In order to reduce noise, a threshold determination process may be performed after applying a moving average filter or the like.

式（７６）〜（７８）において、＊は共役を示す記号である。 Also, the ATI that estimates the speed of the moving target by causing the SAR image (1) and the SAR image (2), the SAR image (1) and the SAR image (3), and the SAR image (2) and the SAR image (3) to interfere with each other. Can also be used.
Specifically, a phase map with the following declination is calculated.

In the formulas (76) to (78), * is a symbol indicating conjugation.

As is apparent from the above, according to the second embodiment, the signal components in the Doppler frequency domain in the SAR images (1), (2), and (3) that have been aligned by the registration units 11a and 11b are positive. Assuming components other than the clutter component and the azimuth ambiguity component included in the signal component sorted into the positive region by the signal component sorting unit 13 and the signal component sorting unit 13 that separates the region into the negative region By estimating the clutter component and the azimuth ambiguity component, the clutter component and the azimuth ambiguity component contained in the signal component classified into the positive region are suppressed, and the signal component classification unit 13 Assuming components other than the clutter component and azimuth ambiguity component included in the signal components classified into An unnecessary component suppression unit 14 configured to suppress the clutter component and the azimuth ambiguity component included in the signal component classified into the negative region by estimating the latter component and the azimuth ambiguity component is provided. Therefore, even when the phase centers of a plurality of antennas do not overlap, there is an effect that both the clutter component and the azimuth ambiguity component are suppressed and the moving target can be detected.
In addition, since the number of channels is larger than that in the first embodiment, the speed estimation range can be expanded by solving the phase unwrapping.

In the second embodiment, as shown in the above formulas (55) and (56), the expected value E of the power ratio between the azimuth ambiguity component a ₊ (p, q) and the clutter component s ₊ (p, q) Using the (a ₊ / s ₊ ) and the expected value E (n ₊ / s ₊ ) of the power ratio of the noise component n ₊ (p, q) and the clutter component s ₊ (p, q), the Wiener filter is used. And an expected value E (a ₋ / s) of the power ratio between the azimuth ambiguity component a ₋ (p, q) and the clutter component s ₋ (p, q) as shown in the equations (64) and (65). _- a), the noise component _n - (p, q) and clutter components _s - (p, the expected value E of the power ratio q) _(n - / s _-) and was used to produce a Wiener filter, their using Wiener filters, clutter component _{s + (p, q),} s - (p, q) and Ajimasua Bigyuiti components _{a + (p, q),} a - (p, q) showed to estimate, and the generation process of the estimation process and the Wiener filter repeatedly performed recursively to enhance the estimation accuracy It may be.

In other words, the unnecessary component suppression unit 14 uses the Wiener filter generated using the expected values E (a ₊ / s ₊ ) and E (n ₊ / s ₊ ) to generate the clutter component s ₊ (p, q) and azimuth. ambiguity components _a + (p, q) After estimating the azimuth ambiguity components _a + (p, q) obtained by the estimated clutter component _s + (p, q) power ratio _{(a +} / _s + ) And the estimated power ratio (n ₊ / s ₊ ) of the clutter component s ₊ (p, q) and the noise component n ₊ (p, q) to regenerate the Wiener filter The clutter component s ₊ (p, q) and the azimuth ambiguity component a ₊ (p, q) are estimated using the Wiener filter.
Similarly, using a Wiener filter generated using the expected values E (a ₋ / s ₋ ) and E (n ₋ / s ₋ ), the clutter component s ₋ (p, q) and the azimuth ambiguity component a ₋ After estimating (p, q), the estimated power ratio (a ₋ / s ₋ ) between the estimated azimuth ambiguity component a ₋ (p, q) and clutter component s ₋ (p, q) is estimated. Using the power ratio (n ₋ / s ₋ ) of the clutter component s ₋ (p, q) and the noise component n ₋ (p, q), the Wiener filter is regenerated, and the regenerated Wiener filter is used. The clutter component s ₋ (p, q) and the azimuth ambiguity component a ₋ (p, q) are estimated.

When the unnecessary component suppression unit 14 recursively repeats the estimation processing of the azimuth ambiguity component and the Wiener filter generation processing, for example, whether or not the estimated clutter component and azimuth ambiguity component have converged. The estimation process and the generation process may be repeated until the estimated clutter component and the azimuth ambiguity component converge, or the estimation process and the generation process may be repeated a preset number of times. It may be.
In addition, as the convergence determination of the clutter component and the azimuth ambiguity component, for example, the previous estimation result and the current estimation result are compared, and if the difference between these estimation results is within a preset threshold, the convergence is achieved. It can be considered to determine that it is.

In the second embodiment, the unnecessary component suppression unit 14 generates the Wiener filter using the expected values E (a ₊ / s ₊ ) and E (n ₊ / s ₊ ). If the signal power of the clutter component is dominant (if the ratio of the signal power of the clutter component is greater than or equal to a preset threshold value), the signal power of the clutter component is used to calculate the signal power of the clutter component. A filter may be generated.
That is, the expected values E (a ₊ / s ₊ ), E (n ₊ / s ₊ ) in the above equations (55) and (56), and the expected value E (a ₋ / s _{− in} equations (64) and (65). ), E (n ₋ / s ₋ ), a winner filter may be generated by substituting a value inversely proportional to the signal power of the clutter component.
Also in this case, the estimation accuracy of the azimuth ambiguity component and the like and the Wiener filter generation processing may be recursively repeated to improve the estimation accuracy.

That is, the unnecessary component suppression unit 14 estimates the clutter component s ₊ (p, q) and the azimuth ambiguity component a ₊ (p, q) using the Wiener filter generated using the signal power of the clutter component. After that, the power ratio (a ₊ / s ₊ ) of the estimated azimuth ambiguity component a ₊ (p, q) and the clutter component s ₊ (p, q) and the estimated clutter component s ₊ (p, q ) And the power ratio (n ₊ / s ₊ ) of the noise component n ₊ (p, q) to regenerate the Wiener filter, and use the regenerated Wiener filter to generate the clutter component s ₊ (p, q) and the azimuth ambiguity component a ₊ (p, q) are estimated.
Similarly, the clutter component s ₋ (p, q) and the azimuth ambiguity component a ₋ (p, q) are estimated using the Wiener filter generated using the signal power of the clutter component, and then the estimated azimuth is estimated. The power ratio (a ₋ / s ₋ ) between the ambiguity component a ₋ (p, q) and the clutter component s ₋ (p, q), the estimated clutter component s ₋ (p, q), and the noise component n ₋ The Wiener filter is regenerated using the power ratio (n ₋ / s ₋ ) of (p, q), and the clutter component s ₋ (p, q) and the azimuth ambiguity are regenerated using the regenerated Wiener filter. The tee component a ₋ (p, q) is estimated.
When the unnecessary component suppression unit 14 recursively repeats the estimation processing of the azimuth ambiguity component and the Wiener filter generation processing, for example, whether or not the estimated clutter component and azimuth ambiguity component have converged. The estimation process and the generation process may be repeated until the estimated clutter component and the azimuth ambiguity component converge, or the estimation process and the generation process may be repeated a preset number of times. It may be.

式（７９）において、Ｉは単位行列、λは任意の実数である。 In the second embodiment, the unnecessary component suppression unit 14 uses a Wiener filter to perform the clutter components s ₊ (p, q), s ₋ (p, q) and the azimuth ambiguity component a ₊ (p, q). , A ₋ (p, q) is shown, but the clutter components s ₊ (p, q), s ₋ (p, q) and the azimuth ambiguity are obtained using a filter that performs the regularized least squares method. The tee components a ₊ (p, q), a ₋ (p, q) may be estimated. A filter that performs the regularized least squares method is represented by the following equation (79).

In Expression (79), I is a unit matrix, and λ is an arbitrary real number.

Here, λ is an arbitrary real number, but instead of λ, an expected value E (n ₊ ) of the power ratio between the noise component n ₊ (p, q) and the clutter component s ₊ (p, q) is shown. / S ₊ ) (or the expected value E (n ₋ / s ₋ ) of the power ratio between the noise component n ₋ (p, q) and the clutter component s ₋ (p, q)) is substituted into the equation (79). A filter that performs the regularized least squares method may be generated.
In addition, when the signal power of the pixel component in the signal component is calculated and the power of the clutter component is dominant (when the ratio of the signal power of the clutter component is equal to or larger than a preset threshold value), the clutter is substituted for λ. A filter for performing the regularized least square method may be generated by substituting the signal power of the component into the equation (79).

In the second embodiment, the unnecessary component suppression unit 14 uses a Wiener filter or a filter that performs the regularized least squares method to generate clutter components s ₊ (p, q), s ₋ (p, q) and azimuth ambiguity. Although the estimation of the tee components a ₊ (p, q), a ₋ (p, q) is shown, the present invention is not limited to this, and a filter or estimation method that assumes the presence of other signal components of the component to be estimated It is possible to apply.
In the second embodiment, an example in which three receiving antennas are used has been described. However, the present invention can also be applied to a case in which four or more receiving antennas are used. Further, it is possible to cope with a case where a plurality of transmission antennas are used or a case where a plurality of transmission antennas and a plurality of reception antennas are used.

Embodiment 3 FIG.
In this third embodiment, as in the second embodiment, an example using three receiving antennas will be described. However, as shown in FIG. 7, the influence of the azimuth ambiguity component in the Doppler frequency region is large, Even if the signal components are separated in the Doppler frequency region, two types of azimuth ambiguity components are mixed.
In the third embodiment, as shown in FIG. 7, even when two types of azimuth ambiguity components are mixed, both the clutter component and the azimuth ambiguity component are suppressed and the moving target is detected. A radar signal processing apparatus that can be used will be described.

FIG. 8 is a block diagram showing a radar signal processing apparatus according to Embodiment 3 of the present invention. In the figure, the same reference numerals as those in FIG.
The unnecessary component suppression unit 20 includes clutter components and azimuths included in signal components in the positive region of the Doppler frequency region in the SAR images (1), (2), and (3) that have been aligned by the registration units 11a and 11b. Assuming the components other than the ambiguity component, the clutter component included in the signal component in the positive region is estimated by estimating the clutter component and the azimuth ambiguity component included in the signal component in the positive region. And azimuth ambiguity component are suppressed, and components other than the clutter component and azimuth ambiguity component included in the negative signal component of the Doppler frequency region are assumed to be the negative signal component. By including the clutter component and azimuth ambiguity component included, it is included in the signal component in the negative region. And processed to implement suppressing the clutter component and the azimuth ambiguities components are. The unnecessary component suppression unit 20 constitutes an unnecessary component suppression unit.

In the third embodiment, pulses are repeatedly radiated from one transmitting antenna, and three receiving antennas receive the reflected waves of the pulse reflected by the target, so that three SAR images (1) and (2) are received. (3) is acquired, and the moving target is detected from the three SAR images (1), (2), and (3).
At this time, the distance between the receiving antenna corresponding to the SAR image (1) and the receiving antenna corresponding to the SAR image (2) is 2m ₂ V _p / PRF + 2L ₂ V _p / PRF with respect to the platform speed V _p . Shall be represented. In addition, the distance between the reception antenna corresponding to the SAR image (1) and the reception antenna corresponding to the SAR image (3) is expressed by 2m ₃ V _p / PRF + 2L ₃ V _p / PRF.
PFR is a pulse repetition frequency, m ₂ and m ₃ are integers, L ₂ and L ₃ are real numbers, and 0 <L ₂ and L ₃ <1.

Next, the operation will be described.
In the third embodiment, when three SAR images (1), (2), and (3) having the same observation area and different observation times are given, the following processing is performed based on the SAR image (1). To do.
When the SAR image (1) and the SAR image (2) are given, the registration unit 11a performs registration that matches the position of the SAR image (2) with the position of the SAR image (1).
In addition, when the SAR image (1) and the SAR image (3) are given, the registration unit 11a performs registration that matches the position of the SAR image (3) with the position of the SAR image (1).
At this time, the registration unit 11a performs azimuth resampling in order to correct an image shift of (m ₂ + L ₂ ) pixels in the azimuth direction.
Further, the registration unit 11b performs azimuth resampling in order to correct an image shift of (m ₃ + L ₃ ) pixels in the azimuth direction.
As a result, the received signal r ₁ (p, q) of the SAR image (1) in the p-th range bin and the q-th azimuth bin is described as follows.

In the equation (80), s (p, q) is a clutter component 111 as shown in FIG. 13, and a ₊ (p, q) is a primary azimuth ambiguity component 112 as shown in FIG. a ₋ (p, q) is a primary azimuth ambiguity component 113 as shown in FIG. 13.
The primary azimuth ambiguity component is generated by the sidelobe pattern of the antenna, and the secondary and subsequent components are assumed to be negligibly small.
n ₁ (p, q) is a noise component in the SAR image (1), and t ₁ (p, q) is a moving target component in the SAR image (1).
Note that there are P range bins and Q azimuth bins.

式（８１）において、ｎ_２（ｐ，ｑ）はＳＡＲ画像（２）内の雑音成分、ｔ_２（ｐ，ｑ）はＳＡＲ画像（２）内の移動目標成分である。 On the other hand, the received signal r ₂ (p, q) of the SAR image (2) in the p-th range bin and the q-th azimuth bin is described as follows.

In Expression (81), n ₂ (p, q) is a noise component in the SAR image (2), and t ₂ (p, q) is a moving target component in the SAR image (2).

式（８２）において、ｎ_３（ｐ，ｑ）はＳＡＲ画像（３）内の雑音成分、ｔ_３（ｐ，ｑ）はＳＡＲ画像（３）内の移動目標成分である。 The received signal r ₃ (p, q) of the SAR image (3) in the p-th range bin and the q-th azimuth bin is described as follows.

In Expression (82), n ₃ (p, q) is a noise component in the SAR image (3), and t ₃ (p, q) is a moving target component in the SAR image (3).

When the unnecessary component suppression unit 20 of the signal component estimation unit 12 receives the SAR images (1), (2), and (3) that have been aligned by the registration units 11a and 11b, the SAR image (1) (2). The SAR image (1) in which the clutter component and the azimuth ambiguity component included in (3) are estimated, the clutter component and the azimuth ambiguity component are suppressed, and the clutter component and the azimuth ambiguity component are suppressed. (2) Output (3).
Received signal r ₁ (p, q) of SAR image (1), received signal r ₂ (p, q) of SAR image (2) after registration, and received signal r ₃ of SAR image (3) after registration In (p, q), two types of azimuth ambiguity components as shown in FIG. 7 are mixed.

The unnecessary component suppression unit 20 uses the received signals r ₁ (p, q), r ₂ (p, q), r ₃ (p, q) to prevent the clutter component s from being included as much as possible. (P, q) and the azimuth ambiguity components a ₊ (p, q), a ₋ (p, q) are estimated.
Hereinafter, using the received signals r ₁ (p, q), r ₂ (p, q), r ₃ (p, q), the clutter component s (p, q) and the azimuth ambiguity component a ₊ (p, q A process for estimating q), a ₋ (p, q) will be described.

First, the formulation of the received signals r ₁ (p, q), r ₂ (p, q), r ₃ (p, q) is organized.
When the formulas of the received signals r ₁ (p, q), r ₂ (p, q), r ₃ (p, q) are expressed by matrices and vectors, they are given as follows.

ただし、Ｅ（ａ_＋／ｓ）はアジマスアンビギュイティ成分ａ_＋（ｐ，ｑ）とクラッタ成分ｓ（ｐ，ｑ）の電力比の期待値であり、Ｅ（ａ₋／ｓ）はアジマスアンビギュイティ成分ａ₋（ｐ，ｑ）とクラッタ成分ｓ（ｐ，ｑ）の電力比の期待値である。
また、Ｅ（ｎ／ｓ）は雑音成分ｎ（ｐ，ｑ）とクラッタ成分ｓ（ｐ，ｑ）の電力比の期待値である。 In the third embodiment, for the same reason as in the first embodiment, a signal is estimated using a Wiener filter.
The estimation method using the Wiener filter is expressed by the following equation (89).

Where E (a ₊ / s) is the expected value of the power ratio of the azimuth ambiguity component a ₊ (p, q) and the clutter component s (p, q), and E (a ₋ / s) is the azimuth ambition. This is the expected value of the power ratio between the Guyty component a ₋ (p, q) and the clutter component s (p, q).
E (n / s) is an expected value of the power ratio between the noise component n (p, q) and the clutter component s (p, q).

This estimation method assumes the existence of signals other than the clutter component s (p, q) and the azimuth ambiguity components a ₊ (p, q), a ₋ (p, q). This is a method in which the amounts included in the estimation of the clutter component s (p, q) and the azimuth ambiguity components a ₊ (p, q), a ₋ (p, q) are statistically minimized.
_{Thus, r 1 (p, q)} -s (p, q) -a + (p, q) -a - (p, q) when calculating the the movement target component _t 1 (p, q) is To remain.

When the unnecessary component suppression unit 20 estimates the clutter component s (p, q) and the azimuth ambiguity components a ₊ (p, q), a ₋ (p, q), the estimation result is used to determine the SAR image. The clutter component and the azimuth ambiguity component included in the received signal r ₁ (p, q) of (1) are suppressed and included in the received signal r ₂ (p, q) of the SAR image (2). The clutter component and the azimuth ambiguity component are suppressed.
Moreover, the clutter component and the azimuth ambiguity component contained in the received signal r ₃ (p, q) of the SAR image (3) are suppressed.
The simplest suppression method is to perform the following calculation.

It is possible to detect the moving target using the calculation result of Expression (92), the calculation result of Expression (93), or the calculation result of Expression (94), but as shown in Expression (95) below, If the three-channel components are combined, the detection performance of the moving target can be further enhanced.

Note that the clutter component s (p, q) and the azimuth ambiguity components a ₊ (p, q), a ₋ (p, q) estimated by the unnecessary component suppression unit 20 include estimation errors.
In particular, it is considered that the clutter component estimation error greatly contributes to the degradation of the detection performance of the moving target because the power of the clutter component is large.
Therefore, the clutter is removed without using the clutter component s (p, q) estimated by the unnecessary component suppression unit 20, and the remaining azimuth ambiguity components a ₊ (p, q), a ₋ (p, q) The following processing for suppressing the above can be considered.

Although the clutter component and the azimuth ambiguity component suppression by the unnecessary component suppression unit 20 is performed in the amplitude dimension, as shown below, it is performed in the power dimension and then returned to the amplitude dimension. It is also possible to improve the detection performance of the moving target.

The target detection unit 15 detects the moving target from the post-suppression image that is an image having a signal component in which the clutter component and the azimuth ambiguity component are suppressed by the unnecessary component suppression unit 20.
As the moving target detection process, a threshold determination process or the like can be considered. However, the threshold determination process may be performed after the suppressed image is converted into a power component. In order to reduce noise, a threshold determination process may be performed after applying a moving average filter or the like.

式（１０２）〜（１０４）において、＊は共役を示す記号である。 Also, the ATI that estimates the speed of the moving target by causing the SAR image (1) and the SAR image (2), the SAR image (1) and the SAR image (3), and the SAR image (2) and the SAR image (3) to interfere with each other. Can also be used.
Specifically, a phase map with the following declination is calculated.

In the formulas (102) to (104), * is a symbol indicating conjugation.

  As apparent from the above, according to the third embodiment, even when two types of azimuth ambiguity components are mixed as shown in FIG. 7, both the clutter component and the azimuth ambiguity component are suppressed. As a result, the moving target can be detected.

In the third embodiment, as shown in the above formulas (90) and (91), the expected value E () of the power ratio between the azimuth ambiguity component a ₊ (p, q) and the clutter component s (p, q). a ₊ / s), the expected value E (a ₋ / s) of the power ratio of the azimuth ambiguity component a ₋ (p, q) and the clutter component s (p, q), and the noise component n (p, q) ) And the expected value E (n / s) of the power ratio of the clutter component s (p, q), a Wiener filter is generated, and the Clutter component s (p, q) and azimuth are generated using the Wiener filter. Although the ambiguity components a ₊ (p, q), a ₋ (p, q) are estimated, the estimation process and the Wiener filter generation process are recursively performed to reduce the estimation accuracy. You may make it raise.

In other words, the unnecessary component suppression unit 20 uses the Wiener filter generated using the expected values E (a ₊ / s), E (a ₋ / s), and E (n / s), and uses the clutter component s (p, q) and the azimuth ambiguity component a ₊ (p, q), a ₋ (p, q) are estimated, and then the estimated azimuth ambiguity component a ₊ (p, q) and the clutter component s (p, q q) power ratio (a ₊ / s), its estimated azimuth ambiguity component a ₋ (p, q) and clutter component s (p, q) power ratio (a ₋ / s), and its estimation The Wiener filter is regenerated using the power ratio (n / s) of the clutter component s (p, q) and the noise component n (p, q), and the clutter component is regenerated using the regenerated Wiener filter. s (p, q) and azimuth ambiguity components _a + (p, q), _- (p, q) so as to estimate.

When the unnecessary component suppression unit 20 recursively repeats the estimation processing of the azimuth ambiguity component and the like and the Wiener filter generation processing, for example, whether or not the estimated clutter component and azimuth ambiguity component have converged is determined. The estimation process and the generation process may be repeated until the estimated clutter component and the azimuth ambiguity component converge, or the estimation process and the generation process may be repeated a preset number of times. It may be.
In addition, as the convergence determination of the clutter component and the azimuth ambiguity component, for example, the previous estimation result and the current estimation result are compared, and if the difference between these estimation results is within a preset threshold, the convergence is achieved. It can be considered to determine that it is.

In the third embodiment, the unnecessary component suppression unit 20 generates a Wiener filter using the expected values E (a ₊ / s), E (a ₋ / s), and E (n / s). However, when the signal power of the pixel component in the signal component is calculated and the power of the clutter component is dominant (when the ratio of the signal power of the clutter component is equal to or higher than a preset threshold value), the signal power of the clutter component A Wiener filter may be generated using
That is, instead of the expected values E (a ₊ / s), E (a ₋ / s), and E (n / s) in the above equations (90) and (91), values that are inversely proportional to the signal power of the clutter component are obtained. It may be substituted to generate a winner filter.
Also in this case, the estimation accuracy of the azimuth ambiguity component and the like and the Wiener filter generation processing may be recursively repeated to improve the estimation accuracy.

That is, the unnecessary component suppression unit 20 uses the Wiener filter generated using the signal power of the clutter component, and the clutter component s (p, q) and the azimuth ambiguity component a ₊ (p, q), a ₋ ( After estimating p, q), the estimated azimuth ambiguity component a ₊ (p, q) and the clutter component s (p, q) power ratio (a ₊ / s) and the estimated azimuth ambiguity Power ratio (a ₋ / s) between the tee component a ₋ (p, q) and the clutter component s (p, q), and the estimated clutter component s (p, q) and noise component n (p, q) The Wiener filter is regenerated using the power ratio (n / s), and the clutter component s (p, q) and the azimuth ambiguity component a ₊ (p, q) are regenerated using the regenerated Wiener filter. , A ₋ (p, q) To do.
When the unnecessary component suppression unit 20 recursively repeats the estimation processing of the azimuth ambiguity component and the like and the Wiener filter generation processing, for example, whether or not the estimated clutter component and azimuth ambiguity component have converged is determined. The estimation process and the generation process may be repeated until the estimated clutter component and the azimuth ambiguity component converge, or the estimation process and the generation process may be repeated a preset number of times. It may be.

式（１０５）において、Ｉは単位行列、λは任意の実数である。 In Embodiment 3, the unnecessary component suppression unit 20 uses a Wiener filter to convert the clutter component s (p, q) and the azimuth ambiguity components a ₊ (p, q), a ₋ (p, q). Although what was estimated was shown, the clutter component s (p, q) and the azimuth ambiguity components a ₊ (p, q), a ₋ (p, q) are obtained by using a filter that performs regularized least squares. You may make it estimate. A filter that performs the regularized least squares method is represented by the following equation (105).

In Expression (105), I is a unit matrix, and λ is an arbitrary real number.

Here, λ is an arbitrary real number, but instead of λ, an expected value E (n / s) of the power ratio of the noise component n (p, q) and the clutter component s (p, q). May be substituted into equation (105) to generate a filter that performs the regularized least squares method.
In addition, when the signal power of the pixel component in the signal component is calculated and the power of the clutter component is dominant (when the ratio of the signal power of the clutter component is equal to or larger than a preset threshold value), the clutter is substituted for λ. A filter that performs the regularized least squares method may be generated by substituting the signal power of the component into Expression (105).

In the third embodiment, the unnecessary component suppression unit 20 uses a Wiener filter or a filter that performs a regularized least squares method, and uses a clutter component s (p, q) and an azimuth ambiguity component a ₊ (p, q). , A ₋ (p, q) is shown, but the present invention is not limited to this, and it is possible to apply a filter or an estimation method that assumes the presence of other signal components of the component to be estimated.
In the third embodiment, an example in which three receiving antennas are used has been described. However, the present invention can also be applied to a case in which four or more receiving antennas are used. Further, it is possible to cope with a case where a plurality of transmission antennas are used or a case where a plurality of transmission antennas and a plurality of reception antennas are used.

  In the present invention, within the scope of the invention, any combination of the embodiments, or any modification of any component in each embodiment, or omission of any component in each embodiment is possible. .

  DESCRIPTION OF SYMBOLS 1 Registration part (positioning means), 2 Signal component estimation part, 3 Signal component classification part (Signal component classification means), 4 Unnecessary component suppression part (Unnecessary component suppression means), 5 Target detection part (Target detection means), 11a, 11b Registration unit (positioning unit), 12 signal component estimation unit, 13 signal component classification unit (signal component classification unit), 14, 20 unnecessary component suppression unit (unnecessary component suppression unit), 15 target detection unit (target) Detection means), 101 registration unit, 102 difference image generation unit, 111 clutter component, 112, 113 azimuth ambiguity component.

Claims (11)

Alignment means for aligning a plurality of synthetic aperture radar images having the same observation area and different observation times;
Signal component classification means for classifying signal components in the Doppler frequency domain in a plurality of synthetic aperture radar images that have been aligned by the alignment module into positive and negative areas;
The clutter component and the azimuth ambiguity component included in the signal component classified into the positive area by the signal component classification means are estimated, and the clutter component included in the signal component classified into the positive area And the azimuth ambiguity component are suppressed, and the clutter component and the azimuth ambiguity component included in the signal component classified into the negative region by the signal component classification unit are estimated and separated into the negative region. Unnecessary component suppression means for suppressing the clutter component and the azimuth ambiguity component included in the received signal component;
A radar signal processing apparatus comprising: target detection means for detecting a moving target from an image having signal components in which clutter components and azimuth ambiguity components are suppressed by the unnecessary component suppression means. Alignment means for aligning a plurality of synthetic aperture radar images having the same observation area and different observation times;
Assuming components other than the clutter component and the azimuth ambiguity component included in the signal component of the positive region of the Doppler frequency region in the plurality of synthetic aperture radar images aligned by the alignment unit, By estimating the clutter component and the azimuth ambiguity component included in the signal component of the positive region, the clutter component and the azimuth ambiguity component included in the signal component of the positive region are suppressed, Assuming components other than the clutter component and the azimuth ambiguity component included in the signal component in the negative region of the Doppler frequency region, the clutter component and the azimuth ambiguity included in the signal component in the negative region are assumed. By estimating the tee component, the clutter component and the And the unnecessary component suppressing means for suppressing azimuth ambiguities component,
A radar signal processing apparatus comprising: target detection means for detecting a moving target from an image having signal components in which clutter components and azimuth ambiguity components are suppressed by the unnecessary component suppression means.   The unnecessary component suppressing means performs estimation processing of clutter components and azimuth ambiguity components included in signal components of the positive region and the negative region using a Wiener filter. The radar signal processing apparatus according to claim 1 or 2.   The unnecessary component suppressing means includes an expected value of the power ratio of the azimuth ambiguity component and the clutter component included in the signal component of the positive region and the negative region, and the positive region and the negative region. 4. The radar signal processing apparatus according to claim 3, wherein the Wiener filter is generated using an expected value of a power ratio between a noise component and a clutter component included in the signal component.   The unnecessary component suppression means uses a Wiener filter generated by using the expected value of the power ratio to remove a clutter component and an azimuth ambiguity component included in the signal component of the positive region and the negative region. After the estimation, the Wiener filter is regenerated using the estimated power ratio of the azimuth ambiguity component and the clutter component, and the estimated clutter component and the noise component power ratio, and the regenerated Wiener The radar signal processing apparatus according to claim 4, wherein a clutter component and an azimuth ambiguity component included in the signal components of the positive region and the negative region are re-estimated using a filter.   When the unnecessary component suppression means estimates the clutter component and the azimuth ambiguity component included in the signal components of the positive region and the negative region, the estimated clutter component and azimuth ambiguity component converge. And regenerating the Wiener filter until the estimated clutter component and azimuth ambiguity component converge, and using the regenerated Wiener filter, the positive region and the negative 6. The radar signal processing apparatus according to claim 5, wherein a process of estimating a clutter component and an azimuth ambiguity component included in the signal component of the region is repeatedly performed.   The unnecessary component suppression means, when the ratio of the signal power of the clutter component in the signal power of the pixel in the signal component of the positive region and the negative region is greater than or equal to a preset threshold value, the clutter component The radar signal processing apparatus according to claim 3, wherein the Wiener filter is generated using the signal power of the signal.   The unnecessary component suppression means uses a Wiener filter generated by using the signal power of the clutter component to remove the clutter component and the azimuth ambiguity component included in the signal component of the positive region and the negative region. After the estimation, the Wiener filter is regenerated using the estimated power ratio of the azimuth ambiguity component and the clutter component, and the positive region and the negative region are regenerated using the regenerated Wiener filter. The radar signal processing apparatus according to claim 7, wherein clutter components and azimuth ambiguity components included in the signal components are re-estimated.   The unnecessary component suppression means performs estimation processing of clutter components and azimuth ambiguity components included in the signal components of the positive region and the negative region, using a filter that performs a regularized least square method. 3. The radar signal processing apparatus according to claim 1, wherein the radar signal processing apparatus is a radar signal processing apparatus.   The unnecessary component suppression unit generates a filter that performs the regularized least squares method using an expected value of a power ratio between a noise component and a clutter component included in the signal components of the positive region and the negative region. The radar signal processing apparatus according to claim 9.   The unnecessary component suppression means, when the ratio of the signal power of the clutter component in the signal power of the pixel in the signal component of the positive region and the negative region is greater than or equal to a preset threshold value, the clutter component The radar signal processing apparatus according to claim 9, wherein a filter that performs the regularized least squares method is generated using the signal power of the signal.

Images