JP5328491B2 - Radar image processing device - Google Patents

Description

  The present invention relates to a radar image processing apparatus, and more particularly, to a radar image processing apparatus for processing a radar image obtained by irradiating an observation target with radio waves.

  A mobile platform such as an aircraft or an artificial satellite is equipped with a radar such as a synthetic aperture radar or a real aperture radar, and a plurality of radar images obtained by the radar may be processed in a superimposed manner. This superposition processing is used for terrain undulation measurement, minute change extraction, cutting (tomography) image creation, and the like.

  The superimposition process is divided into two stages. In the first stage of the process, another image is superimposed on the reference image, and the size of the shift of the other image with respect to the reference image is estimated. In the second stage of processing, other images superimposed in accordance with the estimated deviation amount so that the observation target (such as a building on the ground) indicated by each pixel of the reference image and each pixel of the other image match. Shift and resample.

とする（例えば、非特許文献１参照）。図１において、レーダ画像Ａを基準画像とし、レーダ画像Ｂを、レーダ画像Ａに重ね合わせる他のレーダ画像とする。ずれの推定は、画像全体で一つのずれ量とみなして推定してもよいし、画像を小画像に分割し、小画像ごとにずれ量を推定してもよい。相互相関には、次式（１）および（２）で示される、相互相関係数Ｃ_ｃ（Δｘ，Δｙ）もしくは相互相関関数Ｃ_ｆ（Δｘ，Δｙ）を用いる。 For example, as shown in FIG. 1, the conventional estimation method of the magnitude of the deviation in the first stage of the process calculates the cross-correlation between two images and calculates the cross-correlation with respect to the deviation (Δx, Δy). Create a distribution map and estimate the position where the maximum cross-correlation occurred.

(For example, see Non-Patent Document 1). In FIG. 1, a radar image A is a reference image, and a radar image B is another radar image that is superimposed on the radar image A. The estimation of the shift may be performed by regarding the entire image as one shift amount, or the image may be divided into small images and the shift amount may be estimated for each small image. For the cross-correlation, a cross-correlation coefficient C _c (Δx, Δy) or a cross-correlation function C _f (Δx, Δy) represented by the following expressions (1) and (2) is used.

  Here, x and y are the vertical and horizontal axes of the radar image A and the radar image B, respectively, and A (x, y) and B (x, y) are the coordinates (x, y) of the radar image A and the radar image B, respectively. ). xmin, xmax, ymin, and ymax respectively indicate the x-axis minimum value, the x-axis maximum value, the y-axis minimum value, and the y-axis maximum value of the radar image for estimating the shift amount. * Indicates a complex conjugate.

に合わせて、レーダ画像Ｂをずらしてリサンプルする。 In the conventional resampling method in the second stage of processing, the estimated shift is made so that the observation target (such as a building on the ground surface) indicated by each pixel of the radar image B superimposed with each pixel of the radar image A matches. amount

The radar image B is shifted and resampled at the same time.

D. S. Schwartz, et al. “P-3 SAR motion compensation techniques”, Proc of SPIE, vol. 4050, pp. 219-230, 2000

In the conventional overlay process, there is a problem in the shift amount estimation process. In the acquisition of radar images, the trajectories of the moving platforms at the time of observing the images A and B do not match due to the influence of wind and the limit on the accuracy of trajectory control. Then, since the pixels of the radar image include a phase proportional to the distance between the trajectory and the observation target, in the image A and the image B, there is a difference in phase between pixels indicating the same observation target. Because of this phase difference, the correlation between images decreases even if there is no shift, so even if an attempt is made to estimate the shift amount from the cross-correlation coefficient C _c (Δx, Δy) or the cross-correlation function C _f (Δx, Δy). There is a possibility that the maximum value of the correlation is lowered at an accurate deviation amount, buried in noise, and maximized at other deviation amounts. For this reason, there is a problem that an error occurs in the shift amount.

  The present invention has been made to solve such a problem, and an object thereof is to obtain a radar image processing apparatus capable of accurately estimating a deviation amount by removing a phase difference caused by a trajectory difference. To do.

  The present invention provides a radar image storage unit that stores a plurality of radar images obtained by observing the same observation target from different trajectories, and a phase correction amount calculation that calculates a phase correction amount between radar images generated due to a trajectory difference between the trajectories. Based on the calculation result of the phase correction amount, a phase correction unit for correcting the phase of the radar image, a cross-correlation calculation unit for calculating a cross-correlation between the phase-corrected radar images, and the cross-correlation A radar comprising: a deviation amount estimation unit that estimates a deviation amount between radar images; and a resample unit that shifts and resamples the radar images so that the radar images overlap each other based on the estimation result of the deviation amount An image processing apparatus.

  The present invention provides a radar image storage unit that stores a plurality of radar images obtained by observing the same observation target from different trajectories, and a phase correction amount calculation that calculates a phase correction amount between radar images generated due to a trajectory difference between the trajectories. Based on the calculation result of the phase correction amount, a phase correction unit for correcting the phase of the radar image, a cross-correlation calculation unit for calculating a cross-correlation between the phase-corrected radar images, and the cross-correlation A radar comprising: a deviation amount estimation unit that estimates a deviation amount between radar images; and a resample unit that shifts and resamples the radar images so that the radar images overlap each other based on the estimation result of the deviation amount Since the image processing apparatus is used, it is possible to accurately estimate the shift amount by removing the phase difference caused by the trajectory difference.

It is explanatory drawing explaining the overlay process of a radar image. It is the block diagram which showed the structure of the radar image processing apparatus which concerns on Embodiment 1 of this invention. It is the flowchart which showed the flow of the process of the radar image processing apparatus which concerns on Embodiment 1 of this invention. It is the block diagram which showed the structure of the radar image processing apparatus which concerns on Embodiment 2 of this invention. It is the flowchart which showed the flow of the process of the radar image processing apparatus which concerns on Embodiment 2 of this invention. It is the block diagram which showed the structure of the radar image processing apparatus which concerns on Embodiment 3 of this invention. It is the flowchart which showed the flow of the process of the radar image processing apparatus which concerns on Embodiment 3 of this invention.

Embodiment 1 FIG.
A radar image processing apparatus according to an embodiment of the present invention will be described with reference to the drawings. The radar image processing apparatus of the present invention processes a radar image obtained by irradiating an observation target with radio waves from a radar, and performs a radar image overlay process. In the radar image superimposition process that is the object of the present invention, the same observation target such as the ground surface is observed multiple times to obtain a plurality of radar images, and one reference image is obtained from the plurality of radar images. Establish and superimpose other images on the reference image, estimate the magnitude of the deviation, and observe the observation target (such as a building on the ground) indicated by each pixel of the reference image and the same observation target in the superimposed image. The superimposed images are shifted in accordance with the estimated shift amount so as to match, and resampled. Hereinafter, the reference image is referred to as a radar image A, and the superimposed image is referred to as a radar image B. It is assumed that the radar for acquiring the radar image A and the radar image B is mounted on a mobile platform such as an aircraft or an artificial satellite. Radar image A and radar image B are images obtained by observing the same observation target from different trajectories of the mobile platform.

  FIG. 2 is a block diagram showing the configuration of the radar image processing apparatus according to Embodiment 1 of the present invention. In the following, the same or corresponding parts are denoted by the same reference numerals in the respective drawings. In FIG. 2, a radar image A storage unit 1110 stores a radar image A that serves as a reference image in the overlay process. As the reference image, the operator of this apparatus can set an arbitrary image from a plurality of radar images. The radar image B storage unit 1120 stores a radar image B of an image to be superimposed. The trajectory information A storage unit 1210 and the trajectory information B storage unit 1220 store trajectory information of the moving platform when the radar image A and the radar image B are acquired, respectively. Examples of the trajectory information include, but are not limited to, platform position information for each time obtained from an inertial navigation system or Global Positioning System (GPS), or information that can be estimated. The phase difference calculation unit 1230 based on the trajectory is based on the trajectory information stored in the trajectory information A storage unit 1210 and the trajectory information B storage unit 1220, and the phase between the radar image A and the radar image B generated by the trajectory difference. Calculate the correction amount. The phase correction unit 1300 corrects the phases of the radar image A and the radar image B based on the phase correction amount calculated by the phase difference calculation unit 1230 based on the trajectory. The cross-correlation calculation unit 1400 calculates a cross-correlation distribution map between the phase-corrected radar image A and radar image B. The deviation amount estimation unit 1500 estimates the deviation amount between the radar images A and B from the cross-correlation distribution diagram. The resample unit 1600 resamples the radar image B based on the estimated deviation amount. The output storage unit 1700 stores the resampled radar image B.

  Next, the operation will be described. FIG. 3 is a flowchart showing a processing flow of the radar image processing apparatus according to the first embodiment. The step number basically corresponds to the code of the block to be executed in the block diagram (the same applies hereinafter).

  First, in step ST1100, a radar image A and a radar image B obtained by a radar are input and stored in a radar image A storage unit 1110 and a radar image B storage unit 1120, respectively.

Next, in step ST1200, the trajectory-based phase difference calculation unit 1230 reads the trajectory information A and the trajectory information B stored in advance in the trajectory information A storage unit 1210 and the trajectory information B storage unit 1220, and the trajectory information. From A and the trajectory information B, the phase correction amount for each pixel of the radar image A and the radar image B is calculated. The phase correction amounts φ _A (x, y) and φ _B (x, y) of the radar image A and the radar image B can be calculated by the following equations (3) and (4).

ここで、λはレーダの波長、ｒ_Ａ（ｘ，ｙ）はレーダ画像Ａの座標（ｘ，ｙ）に位置する画素の示す観測対象と、その画素を観測した時刻におけるプラットフォームの位置との距離である。この観測した時刻におけるプラットフォームの位置は、画像再生の際に定義される位置である。ｒ_Ｂ（ｘ，ｙ）も同様に、レーダ画像Ｂの座標（ｘ，ｙ）に位置する画素の示す観測対象と、その画素を観測した時刻におけるプラットフォームの位置との距離である。

Here, λ is the radar wavelength, r _A (x, y) is the distance between the observation target indicated by the pixel located at the coordinates (x, y) of the radar image A and the platform position at the time when the pixel is observed. It is. The platform position at the observed time is a position defined at the time of image reproduction. Similarly, r _B (x, y) is the distance between the observation target indicated by the pixel located at the coordinates (x, y) of the radar image B and the platform position at the time when the pixel was observed.

  Next, in step ST1300, phase correction section 1300 corrects the phases of radar image A and radar image B based on the phase correction amount calculated in step ST1200. Pixels A ′ (x, y) and B ′ (x, y) in the corrected radar images A and B are expressed by the following equations (5) and (6), respectively.

  However, j shows an imaginary unit. Since this correction corrects the phase difference between the radar images A and B, for example, it is possible to correct only the radar image A by the following equation (7) and not the radar image B.

  In step ST1400, cross-correlation calculation section 1400 calculates the cross-correlation distribution of phase-corrected radar images A and B. For example, this calculation may be performed sequentially using the above-described equation (1) or equation (2) while changing the shift amount (Δx, Δy). Moreover, the calculation of Formula (2) can also be performed by the following Formula (8) using Fourier transform.

Here, F [] and F ⁻¹ [] represent a two-dimensional Fourier transform and a two-dimensional inverse Fourier transform, respectively.

とする。 In step ST1500, the deviation amount estimation unit 1500 obtains a deviation amount that maximizes the cross-correlation from the cross-correlation distribution obtained in step ST1400, and calculates the deviation amount estimated value.

And

に応じてレーダ画像Ｂをずらし、レーダ画像Ａの各画素と重なるようにレーダ画像Ｂの各画素を補間処理によりリサンプルする。この補間には、例えば、最近画素の画素を用いる方法や、一次関数または二次関数などの多項式により補間する方法などがある。なお、ステップＳＴ１６００でリサンプルされたレーダ画像Ｂは、出力格納部１７００に格納される。 Next, in step ST1600, the resampling unit 1600 determines the estimated deviation amount.

Accordingly, the radar image B is shifted, and each pixel of the radar image B is resampled by interpolation processing so as to overlap with each pixel of the radar image A. This interpolation includes, for example, a method using a pixel of the latest pixel and a method of interpolation using a polynomial such as a linear function or a quadratic function. The radar image B resampled in step ST1600 is stored in the output storage unit 1700.

  As described above, according to the first embodiment, by correcting the phase difference caused by the orbital difference before calculating the cross-correlation, the error in estimating the deviation amount caused by the reduction of the cross-correlation due to the phase difference can be reduced. It is possible to prevent and accurately estimate the deviation amount.

Embodiment 2. FIG.
In the first embodiment described above, the phase correction amount is calculated based on the information on the trajectory of the mobile platform at the time of radar observation. In the second embodiment, the phase correction amount is estimated from the interference fringes between the radar images A and B instead of using the trajectory information.

  FIG. 4 is a block diagram showing the configuration of the radar image processing apparatus according to Embodiment 2 of the present invention. In FIG. 4, the same or corresponding parts as those in the first embodiment are denoted by the same reference numerals, and the description thereof is omitted here. The difference from the configuration of the first embodiment is that the second embodiment does not have the trajectory information A storage unit 1210 and the trajectory information B storage unit 1220 shown in FIG. Instead, a rough overlapping unit 2210 is provided, and the phase difference calculation unit 1230 based on the trajectory of FIG. 2 shown in the first embodiment is not provided. The phase correction amount calculation unit 2220 is provided.

  The coarse overlay unit 2210 superimposes the radar image A and the radar image B with coarse accuracy. Further, the phase correction amount calculation unit 2220 based on the interference fringes calculates the phase correction amount from the interference fringes of the radar image B that is roughly registered with the radar image A.

  Next, the operation will be described. FIG. 5 is a flowchart showing a processing flow of the radar image processing apparatus according to the second embodiment. Step ST1100 is the same as that in the first embodiment.

  In step ST2210, the coarse overlay unit 2210 superimposes the radar image A and the radar image B with coarse accuracy. In general, when the phase difference between two radar images without any deviation is calculated, interference fringes are generated. As the deviation between the radar images increases, the interference fringes are buried in noise and cannot be confirmed. The rough overlay referred to in step ST2210 refers to a roughness that can confirm interference fringes. In addition, as a rough superposition method, in an image (amplitude image) consisting only of the amplitude components of the radar image A and the radar image B, one or a plurality of feature points (for example, a corner of a building or a utility pole) are included in the image. There are methods such as resampling so that they overlap.

Next, in step ST2220, the phase correction amount calculation unit 2220 based on the interference fringes estimates the phase correction amount based on the interference fringes. Interference fringes in radar images are phase changes that occur between two observations. This phase change is a change in appearance due to a change in the ground surface, a change in line of sight, and / or a phase difference due to a trajectory difference. The phase difference due to this orbital difference is the phase correction amount. When the distribution of the phase difference between the radar images A and B is seen, the change in the ground surface and the change in the appearance are random distributions, and the phase difference due to the orbital difference appears as a striped pattern having periodicity. For this reason, when the distribution of the phase difference between the radar images A and B is two-dimensional Fourier transformed, the frequency component of the phase difference due to the orbital difference has a greater power on the spectrum than the other two frequency components of the phase change. Therefore, the phase difference due to the orbital difference can be extracted by taking out the frequency component of this large power and performing the two-dimensional inverse Fourier transform. This phase difference corresponds to φ _A (x, y) −φ _B (x, y) in the equation (7).

  In step ST1300, the phase correction unit 1300 corrects the phases of the radar image A and the radar image B based on the calculated phase correction amount in the same procedure as in the first embodiment. However, in the case of the present embodiment, the phase correction amount is calculated in step ST2220.

  The subsequent processing from step ST1400 to step ST1600 is the same as that in the above-described first embodiment, and thus description thereof is omitted here.

  As described above, according to the second embodiment, the cross-correlation due to this phase difference is corrected by correcting the phase difference caused by the orbital difference before calculating the cross-correlation, as in the first embodiment. Therefore, it is possible to prevent an error in estimating the amount of deviation caused by the decrease in the number of errors and to estimate the amount of deviation accurately. Further, in the second embodiment, since the phase correction amount is estimated from the interference fringes between the radar images A and B and the trajectory information is not required, it is accurate without being affected by the measurement error of the trajectory information. It is possible to estimate the amount of deviation. Further, since no trajectory information is required, an accurate deviation amount can be estimated even with an apparatus having a limited data storage capacity. Further, an accurate deviation amount can be estimated even for a radar image without trajectory information.

Embodiment 3 FIG.
In the third embodiment, instead of the rough fitting of the above-described second embodiment, phase correction and cross-correlation calculation are performed every time the shift is made little by little, so that there is no feature point in the image and rough fitting is performed. Even if this is not possible, phase difference correction without using trajectory information is realized, and thus accurate deviation amount estimation is possible.

  FIG. 6 is a block diagram showing a configuration of a radar image processing apparatus according to Embodiment 3 of the present invention. In FIG. 6, the same or corresponding parts as those in the first embodiment or the second embodiment are denoted by the same reference numerals, and the description thereof is omitted here. The difference between the second embodiment and the second embodiment is that the third embodiment does not have the rough overlapping section 2210 shown in FIG. 4 and is provided with a deviation amount setting section. 3200 and a resample unit 3600 are provided, and the cross-correlation calculation unit 1400 of FIG. 4 shown in the second embodiment is not provided, and instead, a cross-correlation sequential calculation unit 3400 is provided. This is the point.

  The deviation amount setting unit 3200 sets a plurality of deviation amounts (Δx, Δy) in advance, and updates the radar image B data stored in the radar image B storage unit 1120 and the cross-correlation sequential calculation unit 3400. The set deviation amount (Δx, Δy) is sequentially output with the signal as an input. Note that the resample unit 3600 connected to the deviation amount setting unit 3200 resamples in accordance with the set deviation amount. The cross-correlation sequential calculation unit 3400 calculates a cross-correlation from the radar image A and the radar image B phase-corrected by the phase correction amount calculation unit 2220 based on interference fringes, and sets an update signal of the set deviation amount and the cross-correlation distribution. Output the figure.

  Next, the operation will be described. FIG. 7 is a flowchart showing a processing flow of the radar image processing apparatus according to the third embodiment. Step ST1100 is the same as in Embodiments 1 and 2, and thus description thereof is omitted here.

  In step ST3200, the shift amount setting unit 3200 sets a shift amount for shifting the radar image B. For the setting of the deviation amount, a plurality of deviation amounts (Δx, Δy) to be set are determined in advance, and these are sequentially set.

  In step ST3600, the resample unit 3600 shifts the radar image B according to the set amount of deviation, and resamples each pixel of the radar image B by interpolation processing so as to overlap each pixel of the radar image A. This interpolation includes, for example, a method using a pixel of the latest pixel and a method of interpolation using a polynomial such as a linear function or a quadratic function.

  Step ST2220 and step ST1300 are basically the same as those in the second embodiment. That is, in step ST2220, the phase correction amount calculation unit 2220 superimposes the resampled radar image B and the radar image A that is the basic image to interfere, and estimates the phase correction amount from the interference fringes. In step ST1300, as in the second embodiment, phase correction section 1300 corrects the phases of radar image A and radar image B based on the phase correction amount calculated in ST2220.

  In step ST3410, the cross-correlation sequential calculation unit 3400 calculates the cross-correlation between the phase-corrected radar image A and the phase-corrected radar image B after shifting by the shift amount set in step ST3200. Further, based on the calculated cross-correlation value and the deviation amount set at that time, a distribution map of the deviation amount and the cross-correlation is created.

  In step ST3420, all preset deviation amounts are sequentially set, and it is confirmed whether calculation of cross-correlation for all deviation amounts and creation of distribution are completed. If not completed, the process returns to the process of step ST3200, set to the value of the deviation that has not been set yet and the cross-correlation calculation or the like has not been completed, and the processes from step ST3600 to step ST3410 are performed. On the other hand, as a result of the determination in step ST3420, if calculation of cross-correlation for all the deviation amounts and creation of distribution are completed, the process proceeds to step ST1500, and the processes in steps ST1500 and ST1600 are performed. Since the processes in step ST1500 and step ST1600 are the same as those in the first and second embodiments, the description thereof is omitted here.

  As described above, according to the third embodiment, as in the first and second embodiments, the phase difference caused by the trajectory difference is corrected before the cross-correlation is calculated. It is possible to prevent a deviation amount estimation error caused by a decrease in cross-correlation due to a phase difference, and to accurately estimate a deviation amount. Also in the third embodiment, as in the second embodiment, since orbit information is not required, accurate deviation amount estimation is possible without being affected by the measurement error of the orbit information, and the data Even an apparatus with a limited storage capacity can accurately estimate the amount of deviation, and can also accurately estimate the amount of deviation for a radar image without trajectory information. In addition, in the third embodiment, instead of rough fitting, phase correction and cross-correlation calculation are performed every time the shift is performed little by little. Even when rough overlaying as shown in FIG. 2 cannot be performed, accurate deviation amount estimation is possible.

  1110 Radar image A storage unit, 1120 Radar image B storage unit, 1210 Orbit information A storage unit, 1220 Orbit information B storage unit, 1230 Phase difference calculation unit based on orbit, 1300 Phase correction unit, 1400 Cross correlation calculation unit, 1500 deviation Amount estimation unit, 1600 resample unit, 1700 output storage unit, 2210 coarse overlay unit, 2220 phase correction amount calculation unit based on interference fringes, 3200 deviation amount setting unit, 3400 cross-correlation sequential calculation unit, 3600 resample unit.

Claims (4)

A radar image storage unit for storing a plurality of radar images obtained by observing the same observation target from different orbits;
A phase correction amount calculation unit for calculating a phase correction amount between radar images generated due to a trajectory difference between the trajectories;
A phase correction unit for correcting the phase of the radar image based on the calculation result of the phase correction amount;
A cross-correlation calculator for calculating cross-correlation between the phase-corrected radar images;
A deviation amount estimation unit for estimating a deviation amount between the radar images from the cross-correlation;
A radar image processing apparatus comprising: a resample unit that shifts and resamples the radar images so that the radar images overlap with each other based on the estimation result of the shift amount. The phase correction amount calculation unit
A trajectory information storage unit for storing trajectory information of the radar image;
The radar image processing apparatus according to claim 1, further comprising: a correction amount calculation unit that receives the trajectory information and calculates a phase correction amount between radar images based on a trajectory difference. The phase correction amount calculation unit
A rough overlay unit that roughly overlays the radar image to the extent that the radar images interfere with each other and interference fringes can be confirmed;
The radar image processing apparatus according to claim 1, further comprising: a correction amount calculation unit that calculates a phase correction amount between radar images based on the interference fringes. A radar image storage unit for storing a plurality of radar images obtained by observing the same observation target from different orbits;
A shift amount setting unit for setting a shift amount for shifting another radar image to be superimposed on the reference image when one of the plurality of radar images is set as a reference image;
A first resampler for shifting and resampling the other radar image based on the set shift amount;
A phase correction amount calculation unit that calculates the phase correction amount between the other radar image and the reference image based on the interference fringes by causing the resampled other radar image and the reference image to overlap and interfere with each other. When,
A phase correction unit that corrects the phases of the other radar image and the reference image based on the calculation result of the phase correction amount;
A cross-correlation calculator that calculates a cross-correlation between the other radar image that has undergone phase correction and the reference image;
A deviation amount estimation unit for estimating a deviation amount between the other radar image and the reference image from the cross-correlation;
A radar image processing apparatus, comprising: a second resampler that resamples the other radar image and the reference image so as to overlap each other based on the estimation result of the shift amount.

Images