JP2014235155A - Radar image processing apparatus and radar image processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To discriminate changes in a target in an observation area by using a uniform threshold over the entire observation area.SOLUTION: A radar image processing apparatus comprises an index calculation processing section 4 that calculates a coherence γ showing a correlation between two radar images whose misalignment are compensated by an alignment processing section 2, and calculates average scattering intensity of each radar image for calculating an average normalized scattering intensity differential value C which is a normalized value of a differential value in the two average scattering intensity. A target change discrimination section 5 discriminates changes in a target in an observation area by using the coherence γ and the average normalized scattering intensity differential value C calculated by the index calculation processing section 4.

Description

  The present invention relates to a radar image processing apparatus and a radar image processing method for acquiring two radar images at different observation times for the same observation area and discriminating changes in a target in the observation area from the two radar images. .

A radar image processing apparatus that detects a change in a target in an observation region is disclosed in Patent Document 1 below.
In this radar image processing apparatus, two radar images at different observation times for the same observation area are acquired, and coherence indicating a correlation between the two radar images is calculated.
This coherence is “1” when two radar images completely match, and has a property of approaching “0” as the difference between the two radar images increases.
Therefore, this radar image processing apparatus compares the coherence with a predetermined threshold value, and if the coherence is smaller than the threshold value, detects a temporal change in the observation region occurring between different acquisition times of the two radar images. To do.
Further, this radar image processing apparatus calculates a scattering intensity difference between two radar images and a standard deviation of the scattering intensity difference, and combines the scattering intensity difference and the standard deviation with coherence to change temporal changes in the observation region. To detect.

JP 2008-46107 A (paragraph numbers [0031] to [0035])

Since the conventional radar image processing apparatus is configured as described above, the temporal change in the observation region is detected using the scattering intensity difference between the two radar images and the standard deviation of the scattering intensity difference. In general, since the radar image has a large dynamic range, the difference in scattering intensity and the standard deviation value may fluctuate greatly for each observation region. For this reason, there has been a problem that it is necessary to acquire geographical information, optical image information, and the like of the observation area, and to set different threshold values for each observation area using the information.
In addition, even if the temporal change of the observation area can be detected, there is a problem that the appearance of a new target, the disappearance of the target, the minute movement of the target, and the like cannot be distinguished.
In addition, when the signal-to-noise power ratio (SNR) decreases and the correlation between two radar images decreases, there is a problem that the detection accuracy of temporal changes decreases.

The present invention has been made to solve the above-described problems, and a radar image processing apparatus and a radar capable of discriminating a change in a target in an observation region using a uniform threshold value over the entire observation region. An object is to obtain an image processing method.
It is another object of the present invention to provide a radar image processing apparatus that can discriminate target changes with high accuracy even when the signal-to-noise power ratio is lowered.

  A radar image processing apparatus according to the present invention acquires two radar images at different observation times for the same observation area, and compensates for a positional deviation between the two radar images, and a positional deviation compensation means. The coherence indicating the correlation between the two radar images whose positional deviation is compensated by calculating the average scatter intensity of each radar image and calculating the average of the difference values of the two average scatter intensity A target change-related index calculating means for calculating a normalized scattering intensity difference value, and the target change discriminating means uses the coherence calculated by the target change-related index calculating means and the average normalized scattering intensity difference value to It is intended to discriminate changes in the target.

  According to the present invention, the coherence indicating the correlation between the two radar images whose positional deviation has been compensated for by the positional deviation compensation means is calculated, and the average scattering intensity of each radar image is calculated to obtain the two average scattering intensities. The target change related index calculating means for calculating the average normalized scattering intensity difference value, which is the normalized value of the difference value, is provided, and the target change discriminating means includes the coherence and average normalized scattering calculated by the target change related index calculating means. Since it is configured to discriminate changes in the target in the observation region using the intensity difference value, there is an effect that it is possible to discriminate the change in the target in the observation region using a uniform threshold value over the entire observation region. is there.

BRIEF DESCRIPTION OF THE DRAWINGS It is a block diagram which shows the radar image processing apparatus by Embodiment 1 of this invention. It is a flowchart which shows the processing content (radar image processing method) of the radar image processing apparatus by Embodiment 1 of this invention. It is a block diagram which shows the target change discrimination part 5 of the radar image processing apparatus by Embodiment 1 of this invention.

Embodiment 1 FIG.
1 is a block diagram showing a radar image processing apparatus according to Embodiment 1 of the present invention.
In FIG. 1, a radar image storage unit 1 is constituted by a storage device such as a RAM or a hard disk, and a plurality of radar images (radar images at different observation times for the same observation region) generated by a radar device (not shown). Is stored.
The type of the radar device is not particularly limited, and a sound wave radar device that transmits and receives sound waves, a light wave radar device that transmits and receives light waves, and the like are conceivable.

The alignment processing unit 2 is composed of, for example, a semiconductor integrated circuit on which a CPU is mounted, a one-chip microcomputer, or the like. Two radar images at different observation times for the same observation area are obtained from the radar image storage unit 1. A process for acquiring and compensating for the positional deviation between the two radar images is performed. The alignment processing unit 2 constitutes a misalignment compensation unit.
The preprocessing unit 3 is composed of, for example, a semiconductor integrated circuit on which a CPU is mounted, a one-chip microcomputer, or the like, and a Doppler frequency difference between two radar images whose positional deviation is compensated by the alignment processing unit 2. And pre-processing to compensate for geometric phase difference. The preprocessing unit 3 constitutes preprocessing means.

The index calculation processing unit 4 is composed of, for example, a semiconductor integrated circuit on which a CPU is mounted, a one-chip microcomputer or the like, and a coherence γ indicating a correlation between two radar images after preprocessing by the preprocessing unit 3. Is calculated, and the average scattering intensity of each radar image is calculated, and an average normalized scattering intensity difference value C that is a normalized value of the difference value of the two average scattering intensity is calculated. The index calculation processing unit 4 constitutes a target change related index calculation unit.
The target change discriminating unit 5 is composed of, for example, a semiconductor integrated circuit on which a CPU is mounted or a one-chip microcomputer, and the coherence γ and the average normalized scattering intensity difference value C calculated by the index calculation processing unit 4 are used. To perform a process of discriminating changes in the target in the observation region. The target change discriminating unit 5 constitutes a target change discriminating means.
The output storage unit 6 includes a storage device such as a RAM or a hard disk, and stores the discrimination result of the target change discrimination unit 5.

In the example of FIG. 1, each of the radar image storage unit 1, the alignment processing unit 2, the preprocessing unit 3, the index calculation processing unit 4, the target change discriminating unit 5, and the output storage unit 6 that are components of the radar image processing apparatus. Is assumed to be configured with dedicated hardware, but the radar image processing apparatus may be configured with a computer.
When the radar image processing apparatus is configured by a computer, the radar image storage unit 1 and the output storage unit 6 are configured on an internal memory or an external memory of the computer, and an alignment processing unit 2, a preprocessing unit 3, an index calculation A program describing the processing contents of the processing unit 4 and the target change discriminating unit 5 may be stored in a memory of a computer, and the CPU of the computer may execute the program stored in the memory.
FIG. 2 is a flowchart showing the processing contents (radar image processing method) of the radar image processing apparatus according to the first embodiment of the present invention.

FIG. 3 is a block diagram showing the target change discriminating unit 5 of the radar image processing apparatus according to Embodiment 1 of the present invention.
In FIG. 3, the amplitude determination unit 11 includes absolute values A ₁ and A ₂ of amplitudes of local regions in two radar images after the preprocessing by the preprocessing unit 3 and a preset amplitude threshold T ₀ (to noise power). (Corresponding threshold values) are compared, and processing for determining whether or not the absolute values A ₁ and A ₂ of the amplitudes are larger than the amplitude threshold value T ₀ is performed.

If the determination result of the amplitude determination unit 11 indicates that at least one of the amplitude absolute values A ₁ and A ₂ is greater than the amplitude threshold T _0, the first difference value determination unit 12 is an index calculation processing unit. The average normalized scattering intensity difference value C calculated in step 4 is compared with a preset difference value threshold value T ₁ (upper limit side threshold value), and the average normalized scattering intensity difference value C is determined as the difference value threshold value T _1. A process for determining whether or not the value is larger is performed.
If the determination result of the first difference value determination unit 12 indicates that the average normalized scattering intensity difference value C is greater than the difference value threshold T _{1, the} target disappearance recognition unit 13 performs a process for determining the disappearance of the target. carry out.

When the determination result of the amplitude determination unit 11 indicates that at least one of the absolute values A ₁ and A ₂ of the amplitude is greater than the amplitude threshold T _0, the second difference value determination unit 14 When the determination result of the difference value determination unit 12 indicates that the average normalized scattering intensity difference value C is smaller than the difference value threshold T ₁ , the average normalized scattering intensity difference value C calculated by the index calculation processing unit 4 And a preset difference value threshold value T ₂ (lower limit side threshold value) are compared to determine whether or not the average normalized scattering intensity difference value C is smaller than the difference value threshold value T ₂ .
If the determination result of the second difference value determination unit 14 indicates that the average normalized scattering intensity difference value C is smaller than the difference value threshold T _{2, the} target appearance recognition unit 15 recognizes the appearance of a new target. Perform the process.

The coherence determination unit 16 is calculated by the index calculation processing unit 4 when the average normalized scattering intensity difference value C calculated by the index calculation processing unit 4 is larger than the difference value threshold T ₂ and smaller than the difference value threshold T _1. The coherence γ is compared with a preset coherence threshold T ₃ to determine whether or not the coherence γ is greater than the coherence threshold T ₃ .
If the determination result of the coherence determination unit 16 indicates that the coherence γ is smaller than the coherence threshold T ₃ , the minute movement authorization unit 17 performs a process of authorizing the target minute movement.
When the determination result of the coherence determination unit 16 indicates that the coherence γ is greater than the coherence threshold T ₃ , or the determination result of the amplitude determination unit 11 indicates that the immobility recognition unit 18 indicates that the absolute values of amplitudes A ₁ and A ₂ If is showing the amplitude threshold T ₀ is less than fact, it carries out a process to certify the stationary target.

Next, the operation will be described.
The alignment processing unit 2 acquires two radar images at different observation times for the same observation area from the plurality of radar images stored in the radar image storage unit 1, and obtains a value between the two radar images. The displacement is compensated (step ST1 in FIG. 2).
As a method for compensating for the positional deviation between two radar images, for example, the amount of positional deviation between the two radar images is calculated from the geometric positional relationship between the radar apparatus and the observation area when each radar image is generated. In addition, there is a method of compensating for a positional deviation between two radar images by geometrically compensating at least one of the radar images (for example, geometric compensation using an affine transformation or a mapping function) according to the positional deviation amount. .

However, if the observation area is the ground surface, the radar image may be affected by the effects of distance deviation, sampling deviation, and chirp-like complex signals due to geometrical effects depending on the platform on which the radar device is installed. The complex sine wave in the frequency space may not be uniform due to the non-linearity of the image shift for each region.
For this reason, a complex sine wave accompanying a positional shift of a local region between two radar images (for example, a small section near a point with high scattering intensity) is calculated, and the complex sine wave is subjected to two-dimensional inverse Fourier transform. Thus, a positional deviation amount is calculated for each small section of the radar image, and at least one radar image is geometrically compensated according to the positional deviation amount (for example, bilinear, bicubic, sinc convolution using an affine transformation or a mapping function) By performing various interpolation processes in accordance with applications such as interpolation, the positional deviation between the two radar images is compensated.

When the registration processing unit 2 compensates for the positional deviation between the two radar images, the preprocessing unit 3 performs preprocessing for compensating for the Doppler frequency difference and the geometric phase difference between the two radar images after the positional deviation compensation. Implement (step ST2).
However, if the trajectory of the radar apparatus viewed from the observation region and the scanning of the antenna beam in the radar apparatus are exactly the same, the preprocessing by the preprocessing unit 3 may be omitted.

In order to compensate for the Doppler frequency difference between the two radar images, it is necessary to apply a so-called spectral filter. To apply the spectral filter, it is necessary to obtain the Doppler center frequency for each coordinate of the radar image.
The Doppler center frequency may be calculated in advance geometrically from the rate of change of the radar distance with respect to the position vector between the observation region and the antenna phase center and the radar transmission wavelength.
Alternatively, the radar image is subjected to a Deramp process for multiplying a chirp signal having a chirp rate α in the azimuth direction to minimize the bandwidth of the azimuth Doppler frequency or to solve an optimization problem based on the amplitude of the same frequency. You may make it estimate.

If the synthetic aperture processing time is relatively long and the azimuth Doppler center frequency can be obtained with a second or higher order polynomial, the optimization problem of the chirp rate α, which is a first order azimuth Doppler center frequency polynomial coefficient, is obtained. The optimal solution can be obtained by sequentially substituting.
Needless to say, the polynomial coefficient of the range Doppler center frequency generated in the range direction by the squint observation can be derived in the same manner as the estimation of the azimuth Doppler center frequency.
Using the obtained Doppler center frequency as an instantaneous frequency, the Doppler center frequency is multiplied by 2π and time-integrated, the integration result is multiplied by an imaginary unit and “−1”, and the multiplication result is expressed as Napier. The Deramp function in the radar image time is obtained as the exponent of the exponential function with the number as a base.

The Deramp process is executed by multiplying the radar image by the Deramp function.
As a result, the center frequency of the radar image after the two-dimensional Fourier transform can be collected at one point, so that a common radar frequency band can be extracted here, a non-common band can be multiplied by 0, Spectral filter processing is performed by replacing it with 0 above.
Since the radar image is generally multiplied by the window function, the window function may be compensated by multiplying the inverse of the window function after the Delamp process, or the window function may be re-applied after the spectrum filter process. You may make it multiply.
After the spectrum filtering process, a Reramp process for multiplying any radar image by a complex conjugate of the Deramp function is performed to compensate for a Doppler frequency difference between the two radar images.

The geometric phase difference between the two radar images corresponds to a so-called plane phase, and causes a deterioration in coherence, so it is necessary to remove the geometric phase difference.
The planar phase (phase image) that is the geometric phase difference can be calculated from the geometric distance between the radar apparatus and the observation area when each radar image is generated. That is, the phase of the two radar images after the positional deviation compensation can be estimated by performing a two-dimensional Fourier transform.
In addition, when a DEM (Digital Elevation Map) that is the same topographic information of two radar images after positional deviation compensation is obtained, a planar phase (phase image) may be simulated from the DEM. Moreover, you may obtain a plane phase (phase image) from the interference result of the same topographic information.
When a planar phase (phase image) that is a geometric phase difference is calculated, any radar image is multiplied by the complex conjugate of the phase image to compensate for the geometric phase difference between the two radar images.
The plane phase, which is the geometric phase difference, may be geometrically removed with the tangent plane where the observation area is located as the ground surface reference, or the phase center of the radar observation with the earth ellipsoid as the ground surface reference. You may remove geometrically using coordinates.

The index calculation processing unit 4 inputs the two radar images after the preprocessing by the preprocessing unit 3, and calculates the correlation between the two radar images as an index related to the change of the target existing in the observation region. The coherence γ shown is calculated, and the average scattering intensity of each radar image is calculated, and an average normalized scattering intensity difference value C, which is a normalized value of the difference value between the two average scattering intensities, is calculated.
Hereinafter, the processing content of the index calculation processing unit 4 will be specifically described.

式（１）において、ｓ₁は２枚のレーダ画像のうち、観測時刻ｔ₁のレーダ画像Ｇ１の局所領域内の画素値であり、ｓ₂は観測時刻ｔ₂のレーダ画像Ｇ２の局所領域内の画素値である。なお、本明細書において画素値と呼称する場合、レーダ画像特有の複素数値である場合を含むものとする。一方で、画素値の輝度値または絶対値と呼称した場合、それは画素値の絶対値をとった値と同一である。また、＊は複素共役である。この実施の形態１では、説明の便宜上、観測時刻ｔ₁は、観測時刻ｔ₂より早い時刻であるとする。
式（１）の分子にある「ｓ₁ｓ₂ ^*」は、レーダ画像Ｇ１の局所領域内の画素値ｓ₁と、レーダ画像Ｇ２の局所領域内の画素値ｓ₂との内積を表している。 The index calculation processing unit 4 calculates the coherence γ by substituting the pixel values s ₁ and s ₂ in the local area in the two radar images into the following equation (1) (step ST3).

In Expression (1), s ₁ is a pixel value in the local region of the radar image G1 at the observation time t ₁ out of the two radar images, and s ₂ is in the local region of the radar image G2 at the observation time t ₂ . Pixel value. In this specification, the term “pixel value” includes the case of a complex value unique to a radar image. On the other hand, when referred to as the luminance value or absolute value of the pixel value, it is the same as the value obtained by taking the absolute value of the pixel value. * Is a complex conjugate. In the first embodiment, for the convenience of explanation, it is assumed that the observation time t ₁ is earlier than the observation time t ₂ .
"S ₁ s ₂ ^*" on the molecule of formula (1) includes a pixel value s ₁ in the local region of the radar image G1, represents the inner product of the pixel values s ₂ in the local region of the radar image G2 .

However, when calculating the coherence γ, if the local region includes a region with a low signal-to-noise power ratio, the value of the true coherence γ is close to zero. At this time, it is known that the calculation accuracy of the coherence γ deteriorates and a bias is given to a high value.
For example, in Equation (1), when the number of samples L = 1 without performing spatial averaging, γ = 1 is always established.
Therefore, the index calculation processing unit 4 is a local region in the radar image in which the number of pixels in which the luminance value of the pixel value s is greater than a predetermined luminance value threshold is equal to or greater than a predetermined number in order to prevent deterioration in the calculation accuracy of the coherence γ. And the coherence γ is calculated using the pixel value s in the local region.

That is, when the number of samples in the coherence estimation window (the number of pixels in the local region) is L, the index calculation processing unit 4 determines that the number N of pixels whose luminance value s is higher than the luminance value threshold value T _L is the following equation (2 ) Is searched for a local area satisfying
N> ξL (2)
In equation (2), ξ is a preset coefficient (0 <ξ ≦ 1).
When the index calculation processing unit 4 searches for a local region that satisfies the equation (2), the index calculation processing unit 4 substitutes the pixel values s ₁ and s ₂ in the local region into the equation (1) to calculate the coherence γ.

式（３）では、２次元ノルムの二乗を用いて、平均規格化散乱強度差分値Ｃを算出しているが、ｎ次元ノルムを用いて（ｎ≧３）、平均規格化散乱強度差分値Ｃを算出するようにしてもよい。
式（３）では、平均規格化散乱強度差分値Ｃは、−１〜＋１の値をとるため、例えば、１−｜Ｃ｜として、コヒーレンスγと同じ値域を持つようにしてもよい。 Further, the index calculation processing unit 4 calculates the average normalized scattering intensity difference value C by substituting the pixel values s ₁ and s ₂ in the local region in the two radar images into the following equation (3). (Step ST4).

In Expression (3), the average normalized scattering intensity difference value C is calculated using the square of the two-dimensional norm, but the average normalized scattering intensity difference value C is calculated using the n-dimensional norm (n ≧ 3). May be calculated.
In Expression (3), the average normalized scattering intensity difference value C takes a value of −1 to +1, and may have the same value range as the coherence γ, for example, as 1− | C |.

When the index calculation processing unit 4 calculates the coherence γ and the average normalized scattering intensity difference value C, the target change discriminating unit 5 uses the coherence γ and the average normalized scattering intensity difference value C to change the target in the observation region. Discriminate.
Hereinafter, the processing content of the target change discrimination part 5 is demonstrated concretely.

The amplitude determining unit 11 of the target change discriminating unit 5 is the absolute value A ₁ of the amplitude of the local region in the radar image G1 at the observation time t ₁ out of the two radar images G1 and G2 after the preprocessing by the preprocessing unit 3. Is calculated.
As the absolute value A ₁ of the amplitude of the local region, for example, an average value of luminance values of the pixel value s ₁ in the local region is calculated.
The local region here may be the same region as the local region searched by the index calculation processing unit 4 or may be a different region.
When the amplitude determination unit 11 calculates the absolute value A ₁ of the local region amplitude in the radar image G _{1 at} the observation time t ₁ , the amplitude absolute value A ₁ and a preset amplitude threshold T ₀ (corresponding to noise power) To determine whether the absolute value A ₁ of the amplitude is greater than the amplitude threshold T ₀ (step ST5).
For example, when the luminance value of the pixel value s constituting the radar image takes a value from 0 to 255, the amplitude threshold T ₀ is set to any value from ₀ to 255.

The amplitude judging section 11 calculates an absolute value A ₂ of the amplitude of the local area in the radar image G2 of observation time t _2.
When the amplitude determination unit 11 calculates the absolute value A ₂ of the amplitude of the local region in the radar image G2 at the observation time t ₂ , the amplitude determination unit 11 compares the absolute value A ₂ of the amplitude with the amplitude threshold value T ₀ and calculates the absolute value of the amplitude. It is determined whether A ₂ is larger than the amplitude threshold T ₀ (step ST6).

If the determination result of the amplitude determination unit 11 indicates that at least one of the absolute values A ₁ and A ₂ of the amplitude is greater than the amplitude threshold T _0, the first difference value determination unit 12 performs index calculation processing. Part 4 by comparing the difference value thresholds T ₁ which is set in advance as the average normalized scattering intensity difference value C calculated by, whether the average normalized scattering intensity difference value C is greater than the difference value thresholds T ₁ Is determined (step ST7). The difference value threshold T ₁ is set in advance to any value from 0 to ₁ .
If the determination result of the first difference value determination unit 12 indicates that the average normalized scattering intensity difference value C is greater than the difference value threshold T ₁ , the target disappearance recognition unit 13 disappears from the observation region. (Step ST8).

When the determination result of the amplitude determination unit 11 indicates that at least one of the absolute values A ₁ and A ₂ of the amplitude is greater than the amplitude threshold T _0, the second difference value determination unit 14 If the determination result of the difference value determination unit 12 indicates that the average normalized scattering intensity difference value C is smaller than the difference value threshold T _1, the average normalized scattering intensity difference value calculated by the index calculation processing unit 4 C is compared with a preset difference value threshold T ₂ to determine whether or not the average normalized scattering intensity difference value C is smaller than the difference value threshold T ₂ (step ST9). The difference value threshold T ₂ is set in advance to any value in the range from −1 to 0.
If the determination result of the second difference value determination unit 14 indicates that the average normalized scattering intensity difference value C is smaller than the difference value threshold T ₂ , the target appearance recognition unit 15 creates a new target in the observation region. Is recognized (step ST10).

The coherence determination unit 16 determines that the average normalized scattering intensity difference value C calculated by the index calculation processing unit 4 is larger than the difference value threshold T ₂ and smaller than the difference value threshold T ₁ (T ₂ <C <T ₁ ), The coherence γ calculated by the index calculation processing unit 4 is compared with a preset coherence threshold T ₃ to determine whether the coherence γ is larger than the coherence threshold T ₃ (step ST11). The coherence threshold T ₃ is set in advance to any value from 0 to 1.
If the determination result of the coherence determination unit 16 indicates that the coherence γ is smaller than the coherence threshold T ₃ , the minute movement recognition unit 17 recognizes that the target in the observation region has moved or changed slightly. (Step ST12).

If the determination result of the coherence determination unit 16 indicates that the coherence γ is greater than the coherence threshold T ₃ , the immobility recognition unit 18 recognizes that the target in the observation region is not moving at all (step ST13).
The immobility recognition unit 18 also moves the target in the observation region at all even when the determination result of the amplitude determination unit 11 indicates that the absolute values A ₁ and A ₂ of the amplitude are smaller than the amplitude threshold T _0. If not, it is certified (step ST13).
The discrimination result of the target change discriminating unit 5 is stored by the output storage unit 6, and is output to an external output device such as a display as necessary.

  As is apparent from the above, according to the first embodiment, the coherence γ indicating the correlation between the two radar images whose positional deviation has been compensated for by the alignment processing unit 2 is calculated, and the average of each radar image is calculated. An index calculation processing unit 4 that calculates the scattering intensity and calculates an average normalized scattering intensity difference value C that is a normalized value of the difference value between the two average scattering intensities is provided, and the target change discriminating unit 5 performs the index calculation processing. Using the coherence γ calculated by the unit 4 and the average normalized scattering intensity difference value C, it is configured to discriminate target changes in the observation region, so that a uniform threshold is used over the entire observation region, There is an effect that the change of the target in the observation region can be discriminated.

  Further, according to the first embodiment, the index calculation processing unit 4 searches for a local region in the radar image in which the number of pixels in which the luminance value of the pixel value s is greater than the predetermined luminance value threshold is equal to or greater than the predetermined number. Since the coherence γ is calculated using the pixel value s in the local region, the coherence γ can be calculated excluding the region where the bias is generated, and as a result, Thus, there is an effect that it is possible to prevent deterioration in calculation accuracy of the coherence γ.

  Further, according to the first embodiment, the immobility recognition unit 18 is configured to recognize the target immobility when the amplitude of the local region in the two radar images is smaller than the noise power. It is possible to exclude the observation area that is mirror-like and difficult to change at each observation from the change discrimination target, and as a result, even if the signal-to-noise power ratio is reduced, the target change There is an effect that can be discriminated with high accuracy.

  In the present invention, any constituent element of the embodiment can be modified or any constituent element of the embodiment can be omitted within the scope of the invention.

  DESCRIPTION OF SYMBOLS 1 Radar image storage part, 2 Positioning process part (position shift compensation means), 3 Pre-processing part (Pre-processing means), 4 Index calculation process part (Target change related index calculation means), 5 Target change discrimination part (Target change) Discrimination means), 6 output storage unit, 11 amplitude determination unit, 12 first difference value determination unit, 13 target disappearance recognition unit, 14 second difference value determination unit, 15 target appearance determination unit, 16 coherence determination unit, 17 Micro movement certification part, 18 immobility certification part.

Claims (13)

Misalignment compensation means for acquiring two radar images at different observation times for the same observation region and compensating for misalignment between the two radar images;
The coherence indicating the correlation between the two radar images whose positional deviation has been compensated by the positional deviation compensation means is calculated, the average scattering intensity of each radar image is calculated, and the standard of the difference value between the two average scattering intensity is calculated. A target change related index calculating means for calculating an average normalized scattering intensity difference value which is a normalized value;
A radar image processing apparatus comprising: target change discriminating means for discriminating a change in a target in the observation region using the coherence calculated by the target change related index calculating means and the average normalized scattering intensity difference value.   The target change discriminating means recognizes the disappearance of the target when the average normalized scattering intensity difference value calculated by the target change related index calculating means is larger than a preset upper limit side threshold value. The radar image processing apparatus according to 1.   The target change discriminating means recognizes the appearance of a new target when the average normalized scattering intensity difference value calculated by the target change related index calculating means is smaller than a preset lower threshold value. The radar image processing apparatus according to claim 1 or 2.   The target change discriminating means has a case where the average normalized scattering intensity difference value calculated by the target change related index calculating means is larger than a preset lower limit side threshold and smaller than a preset upper limit side threshold. If the coherence calculated by the target change related index calculating means is smaller than a preset coherence threshold, the minute movement of the target is recognized, and if the coherence is larger than the coherence threshold, the target immobility is recognized. The radar image processing apparatus according to any one of claims 1 to 3, wherein   The target change discriminating means recognizes the immobility of the target when the amplitude of the local region in the two radar images is about noise power. Radar image processing device.   6. The position shift compensation means acquires two radar images from a radar device that generates a radar image using sound waves or light waves, according to any one of claims 1 to 5. Radar image processing device.   The positional deviation compensation means calculates the positional deviation amount between the two radar images from the geometric positional relationship between the radar apparatus and the observation area when generating each radar image, and at least one of them is determined according to the positional deviation amount. The radar image processing apparatus according to claim 6, wherein a positional deviation between the two radar images is compensated by geometrically compensating the radar image.   The misregistration compensation means calculates a misregistration amount between the two radar images from a complex sine wave accompanying the misregistration of the local region between the two radar images, and at least one of the radar images according to the misregistration amount. The radar image processing apparatus according to claim 6, wherein the positional deviation between the two radar images is compensated by geometrically compensating for.   When calculating the coherence indicating the correlation between the two radar images, the target change related index calculating means calculates a local region in the radar image in which the number of pixels whose luminance value is greater than a predetermined luminance value threshold is a predetermined number or more. The radar image processing apparatus according to any one of claims 1 to 8, wherein search is performed, and coherence is calculated using a pixel value in the local region.   Pre-processing for compensating for the Doppler frequency difference or geometric phase difference between the two radar images whose positional deviation has been compensated by the positional deviation compensating means is performed, and the two radar images after the pre-processing are converted into target change related index calculating means. The radar image processing apparatus according to claim 1, further comprising pre-processing means for outputting the signal to   The preprocessing means calculates a phase image from the geometric distance between the radar apparatus and the observation area when each radar image is generated, and multiplies one of the radar images by the complex conjugate of the phase image to The radar image processing apparatus according to claim 10, wherein a geometric phase difference between a plurality of radar images is compensated.   The pre-processing means is either a phase image showing an interference result of the same terrain information of two radar images whose positional deviation is compensated by the positional deviation compensation means, or a complex conjugate of the phase image calculated from the terrain information. The radar image processing apparatus according to claim 10, wherein a geometric phase difference between the two radar images is compensated by multiplying the radar image. A displacement compensation processing step in which the displacement compensation means acquires two radar images at different observation times for the same observation region, and compensates for the displacement between the two radar images;
The target change related index calculation means calculates coherence indicating the correlation between the two radar images whose positional deviation has been compensated in the positional deviation compensation processing step, calculates the average scattering intensity of each radar image, and calculates 2 A target change related index calculation processing step for calculating an average normalized scattering intensity difference value that is a normalized value of a difference value of two average scattering intensities;
The target change discriminating means comprises a target change discrimination processing step for discriminating changes in the target in the observation region using the coherence calculated in the target change related index calculation processing step and the average normalized scattering intensity difference value. Radar image processing method.

Images