JP2008185375A - 3d shape calculation device of sar image, and distortion correction device of sar image - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To perform highly accurate shape estimation relative to an SAR (Synthetic Aperture Radar) image, and to perform correction of layover of the SAR image. <P>SOLUTION: A 3D shape of the SAR image is estimated by using as input images, two or more SAR images imaging the same spot, acquired by imaging with changed squint angles from the same direction. Foreshortening and optical falling are corrected based on the estimated 3D shape and the input images, and brightness of a pixel after correcting a distortion is calculated from correlation of corresponding points in a plurality of input images, to thereby correct the layover. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a technique for generating a highly visible image by correcting distortion generated in an imaging target for a Synthetic Aperture Radar (SAR) image captured from a platform such as an artificial satellite.

A conventional SAR image distortion correction method includes ortho correction. The ortho correction process is a process for correcting foreshortening of a target having an altitude by using altitude data DEM (Digital Elevation Model) of the observation target area and a ground control point GCP (Ground Control Point) (for example, Patent Documents). 1).
In addition, as processing for estimating the shape of the observation target based on the amount of foreshortening of a plurality of SAR images with different imaging conditions, elevation data from two SAR images with different off-nadir angles in imaging from the direction of Doppler zero is used. There is processing for obtaining DEM (SAR stereo DEM creation processing).
JP 2003-323611 A

In the conventional ortho correction, only the foreshortening is corrected, so that the layover cannot be corrected. Therefore, there is a problem that the visibility of a target having a vertical surface such as an artificial structure cannot be improved.
In the SAR stereo DEM creation process, two SAR images having different off-nadir angles are used. For this reason, the radar reflection cross-sectional area greatly depends on the incident angle of the radar wave. As a result, the correlation between the two SAR images is deteriorated. Therefore, the SAR stereo DEM creation process has a problem that the shape estimation accuracy is deteriorated.

  An object of the present invention is to perform highly accurate shape estimation using, for example, two or more SAR images obtained by imaging the same location. Another object of the present invention is to correct layover by combining information of a plurality of images, for example.

The SAR image 3D shape calculation apparatus according to the present invention includes, for example, a SAR image input unit that inputs a plurality of SAR (Synthetic Aperture Radar) images obtained by imaging the same location with different synthetic aperture radars, and stores them in a storage device;
A distortion amount calculation unit that calculates a difference in distortion amount between a plurality of SAR images input by the SAR image input unit by a processing device;
A height conversion coefficient calculation unit that calculates a height conversion coefficient for converting a difference in distortion amount into height based on the plurality of SAR images and GCP (Ground Control Point) data representing the shape of the ground;
Based on the difference between the distortion amounts calculated by the distortion amount calculation unit and the height conversion coefficient calculated by the height conversion coefficient calculation unit, the height of the portion imaged in the SAR image is calculated, And a 3D shape calculation unit that obtains 3D (Dimensional) shape data of a location captured in the SAR image and stores the data in a storage device.

  According to the 3D shape calculation apparatus for a SAR image according to the present invention, 3D shape data can be obtained based on a difference in distortion amount and a height conversion coefficient of a plurality of SAR images. In addition, based on the 3D shape data, it is possible to estimate the shape of the SAR image with high accuracy and to correct the layover.

Embodiment 1 FIG.
In the first embodiment, a method for obtaining 3D shape data of a location captured in a SAR image based on a plurality of SAR images and GCP data will be described. A method for correcting the distortion of the SAR image based on the 3D shape data will be described.

First, SAR image distortion and distortion correction processing will be described.
SAR measures and images the distance (slant range) between a satellite antenna and a ground target. Therefore, a target with a high altitude is imaged at a position close to the satellite side. This phenomenon is called foreshortening. In a normal map, targets that are higher than the ground surface should be in the position of the target. However, in the SAR image, the target at a position higher than the ground surface is imaged at a position having the same slant range, that is, shifted toward the front (falling down).

  In addition, when a target with a high altitude is shifted and displayed on the ground surface in front by foreshortening, there may be another target on the ground surface. In this case, the high altitude target and the other target are displayed in an overlapping manner in the SAR image. This phenomenon is called layover. It is difficult to separate the high-altitude target displayed on the SAR image and the other target.

  Here, the distortion correction processing is processing for estimating the shape of the observation target based on the amount of foreshortening of a plurality of images with different imaging conditions and correcting foreshortening and layover.

  Next, an outline of the SAR image distortion correction system 100 according to the first embodiment will be described with reference to FIG. FIG. 1 is an overall configuration diagram of a SAR image distortion correction system 100 (ortho correction processing apparatus) of a SAR image according to the first embodiment.

The SAR image distortion correction system 100 includes a shape estimation processing unit 5 and a distortion correction processing unit 6. First, the SAR image 1, the SAR image 2, and the GCP data 3 are input to the shape estimation processing unit 5. And the shape estimation process part 5 outputs the 3D shape data 4 of the location imaged by the input SAR image. Further, the SAR image 1, the SAR image 2, and the 3D shape data 4 are input to the distortion correction processing unit 6. Then, the distortion correction processing unit 6 outputs a distortion corrected image 7.
Here, three or more SAR images may be input to the shape estimation processing unit 5 and the distortion correction processing unit 6. However, for the sake of simplicity of explanation, it is assumed here that two SAR images have been input.
The shape estimation processing unit 5 is a SAR image 3D shape calculation device. The distortion correction processing unit 6 is a SAR image distortion correction apparatus.
Here, the SAR image 1 includes the position information of the synthetic aperture, and map information (latitude and longitude or map coordinates) of each pixel can be acquired. The SAR image 2 is a SAR image obtained by capturing the same part as the SAR image 1 using another synthetic aperture. Similar to the SAR image 1, the SAR image 2 includes the position information of the synthetic aperture, and map information of each pixel can be acquired. The GCP data 3 is position and altitude data acquired from map data or the like. The 3D shape data 4 is data indicating the 3D shape of a portion captured in the input SAR image. The distortion corrected image 7 is an image after correcting foreshortening and layover.

Next, foreshortening and layover will be described with reference to FIG. FIG. 2 is a conceptual diagram showing the concept of SAR foreshortening using a satellite-mounted antenna.
As described above, the SAR measures and images the distance (slant range) between the synthetic aperture and the ground target. For this reason, a phenomenon called foreshortening occurs in which a high altitude target is imaged at a position close to the satellite side.
In FIG. 2, target A with altitude should be at position B on the map. However, the SAR image is imaged at a position C where the slant range is equal (distance SA = distance SC). This is called foreshortening. Further, the position D where the distance S-C is equal to the distance from the satellite is imaged to the position C in the same manner. That is, both the position A and the position D are imaged so as to overlap the position C. This is called layover.

Next, the function of the shape estimation process part 5 is demonstrated based on FIG. FIG. 3 is a functional block diagram illustrating functions of the shape estimation processing unit 5.
The shape estimation processing unit 5 includes a SAR image input unit 51, a correlation region division unit 52, a distortion amount calculation unit 53, a height conversion coefficient calculation unit 54, a 3D shape calculation unit 55, and a GCP data input unit 56.
The SAR image input unit 51 inputs a plurality of SAR images obtained by capturing the same portion with different synthetic aperture radars and stores them in the storage device.
The correlation area dividing unit 52 divides each of the plurality of SAR images into a plurality of minute areas by the processing device. That is, the correlation area dividing unit 52 receives the SAR image 1 and the SAR image 2 as input, divides each SAR image, and outputs a minute area.
The distortion amount calculation unit 53 calculates the distortion amount difference between the plurality of SAR images input by the SAR image input unit 51 by the processing device. That is, the distortion amount calculation unit 53 obtains a correlation between each of the plurality of SAR images in each of the plurality of minute regions divided by the correlation region dividing unit 52. Then, a relative positional shift between the plurality of SAR images is obtained, and a distortion amount difference between the plurality of SAR images is calculated. That is, the distortion amount calculation unit 53 receives the divided minute area and the SAR image 1 and the SAR image 2 as inputs and outputs a difference in distortion amount between the minute areas.
Based on a plurality of SAR images and GCP data representing the shape of the ground, the height conversion coefficient calculation unit 54 calculates a height conversion coefficient for converting a difference in distortion amount into a height by a processing device. That is, the height conversion coefficient calculation unit 54 receives the SAR image 1, the SAR image 2, and the GCP data, and outputs a height conversion coefficient.
The 3D shape calculation unit 55 calculates the height of the part imaged in the SAR image based on the difference between the distortion amounts calculated by the distortion amount calculation unit 53 and the height conversion coefficient calculated by the height conversion coefficient calculation unit 54. calculate. Then, the 3D shape calculation unit 55 obtains 3D shape data of locations captured by the plurality of SAR images and stores them in the storage device. That is, the 3D shape calculation unit 55 outputs the 3D shape data 4 based on the difference in distortion amount of each minute region and the height conversion coefficient.
The GCP data input unit 56 inputs GCP data from an input device.

Next, the function of the distortion correction processing unit 6 will be described with reference to FIG. FIG. 4 is a functional block diagram illustrating functions of the distortion correction processing unit 6.
The distortion correction processing unit 6 includes a SAR image input unit 61, a 3D shape data input unit 62, a coordinate calculation unit 63, an image corresponding point calculation unit 64, a coordinate luminance calculation unit 65, and a distortion correction image generation unit 66.
The SAR image input unit 61 inputs a plurality of SAR images obtained by capturing the same portion with different synthetic aperture radars and stores them in the storage device.
The 3D shape data input unit 62 inputs 3D shape data of locations captured in the plurality of SAR images input by the SAR image input unit 61 and stores them in the storage device. That is, the 3D shape data input unit 62 inputs the 3D shape data 4 calculated by the 3D shape calculation unit 55.
The coordinate calculation unit 63 calculates the coordinates of each point after distortion correction of a plurality of SAR images based on the 3D shape data input by the 3D shape data input unit. That is, the coordinate calculation unit 63 inputs the SAR image 1, the SAR image 2, and the 3D shape data 4, and outputs the coordinates of each point (grid point) of the distortion corrected image.
The image corresponding point calculation unit 64 calculates the pixels in the SAR image corresponding to each point after distortion correction for which the coordinate calculation unit 63 has calculated the coordinates by the processing device. That is, the image corresponding point calculation unit 64 inputs the coordinates of the lattice points of the distortion correction image, the SAR image 1, the SAR image 2, and the 3D shape data 4, and calculates the coordinates of the pixels in each SAR image corresponding to each lattice point. Output.
The coordinate luminance calculation unit 65 calculates the luminance of each point after distortion correction by the processing device based on the luminance of each pixel in the plurality of SAR images corresponding to each point after distortion correction calculated by the image corresponding point calculation unit 64. To do. That is, the coordinate luminance calculation unit 65 inputs the coordinates of the pixels in each SAR image corresponding to each grid point, the SAR image 1, the SAR image 2, and the 3D shape data 4, and calculates the luminance of each point after distortion correction. To do.
The distortion correction image generation unit 66 generates a distortion correction image in which the distortion of the SAR image is corrected based on the coordinates of each point after distortion correction and the luminance of each point after distortion correction calculated by the coordinate luminance calculation unit 65. To do. That is, the distortion corrected image generation unit 66 outputs the distortion corrected image 7 with the coordinates of each point after distortion correction and the luminance of each point after distortion correction as inputs.

  Next, the operation of the SAR image distortion correction system according to the first embodiment will be described with reference to FIG. FIG. 5 is a flowchart showing an operation in the SAR image orthocorrection processing according to the first embodiment.

As shown in FIG. 5, the SAR image ortho-correction processing according to the first embodiment mainly includes two STEPs.
First, in STEP 1 (shape estimation processing), the shape estimation processing unit 5 calculates (estimates) 3D shape data based on the SAR image 1, the SAR image 2, and the GCP data.
Next, in STEP 2 (distortion correction processing), the distortion correction processing unit 6 outputs a distortion correction image 7 based on the SAR image 1, the SAR image 2, and the 3D shape data 4.

  Next, an outline of STEP 1 (shape estimation processing) will be described with reference to FIGS. The purpose of STEP 1 is to obtain the three-dimensional shape of the target object of distortion of the SAR image before performing STEP 2.

FIG. 6 is a conceptual diagram showing target distortion appearing in the SAR image. FIG. 7 is a conceptual diagram showing foreshortening, layover, and shadow in a SAR image.
In FIG. 6, h is the target height, and θ is the depression angle when the target is viewed from the synthetic aperture. As described above, the distance from the SAR position S to the target position A is the same as the distance from the SAR position S to the collapsed position C. Here, in general, h is sufficiently smaller than the distance between the position S (synthetic aperture) of the SAR and the target position A. Therefore, it can be approximated that the straight line connecting the position S of the SAR and the target position A and the straight line connecting the collapsed position C and the target position A are orthogonal. In this figure, for the sake of simplicity, it is assumed that the squint angle is 0, that is, the synthetic aperture is in the direction of Doppler zero.
In this case, in the SAR image, the foreshortening occurs on the side surface in front of the target (SAR side) by h · tan θ. Further, a layover occurs between the side surface in front of the target and the ground in front. Also, the rear side of the target with the height (the side opposite to the SAR) cannot be imaged from the SAR. For this reason, shadows (portions where imaging cannot be performed) occur by h / tan θ behind the target. When the SAR is imaged in the state of FIG. 6, foreshortening, layover, and shadow are generated as shown in FIG.
STEP 1 (shape estimation processing) calculates a difference in the amount of foreshortening between each image from a plurality of SAR images, and estimates a target shape.

FIG. 8 is a conceptual diagram when the position of the synthetic aperture is generalized with respect to the relational expression between the foreshortening amount and the synthetic aperture.
In FIG. 8, the antenna forming the synthetic aperture is indicated by a satellite. In FIG. 8, h is the target height vector, R is the displacement vector when the target is viewed from the synthetic aperture, V is the velocity vector of the antenna forming the synthetic aperture, and L is the extension of the velocity vector from the synthetic aperture position. , R ₀ is a displacement vector when the target is viewed from the foot S of a perpendicular line dropped from the target to L, and x is a vector representing distortion due to foreshortening.
Here, distortion x due to foreshortening occurs in a plane orthogonal to V.

  Next, the processing of STEP 1 (shape estimation processing) will be described in detail with reference to FIG. FIG. 9 is a flowchart showing the processing of STEP 1 (shape estimation processing).

  First, in STEP 1-1 (correlation region division processing), the SAR image input unit 51 inputs two or more SAR images obtained by capturing the same part with different synthetic apertures. Then, the correlation area dividing unit 52 divides each image into minute areas having characteristic patterns. Here, the micro region having a characteristic pattern is, for example, a small region including a point with high luminance.

Next, in STEP 1-2 (distortion amount calculation processing), the distortion amount calculation unit 53 receives the minute region divided in STEP 1-1 as an input, and obtains a correlation between the minute region and the SAR image 1 or the SAR image 2. Then, the distortion amount calculation unit 53 obtains a relative positional shift between the SAR image 1 and the SAR image 2 for a minute area in the SAR image, and calculates a relative distortion amount difference in the minute area. .
Here, the distortion amount calculation unit 53 calculates the difference in relative distortion amounts for all the minute regions.

Next, in STEP 1-3 (height conversion coefficient calculation processing), the height conversion coefficient calculation unit 54 calculates the relative position of a plurality of SAR images based on the position information of the synthetic aperture included in the SAR image and the GCP data. A coefficient for converting the difference in distortion amount into height is calculated.
First, the height conversion coefficient calculation unit 54 obtains a displacement vector _RGCP when the GCP is viewed from the synthetic aperture from the GCP position and the synthetic aperture position. Assuming that the GCP is on the same SAR image as the target and that the synthetic aperture is sufficiently away from the target, the displacement vector R when the target is viewed from the synthetic aperture may be treated as almost the same as _RGCP .
That means
R >> R-R _GCP
Therefore,
R ≒ R _GCP
Is established.
Furthermore, since R ₀ is a displacement vector when the target is viewed from the foot S of the perpendicular line that has been lowered from the target to L, R ₀ can be obtained from R and V. That is, R ₀ can be obtained from R _GCP and V.

In FIG. 8, the angle corresponding to the depression angle θ when the target is viewed from the synthetic aperture is an angle formed by R ₀ × V and h. Further, the distortion x due to foreshortening is perpendicular to h in a plane stretched by R ₀ × V and h. Furthermore, the distortion x due to foreshortening is
| X | ≒ h · tanθ
It is.
in this case,
sin θ = | (R ₀ × V) × h | / | R ₀ × V | · | h |,
cos θ = (R ₀ × V) · h / | R ₀ × V | · | h |
It is.
Therefore, the distortion x due to foreshortening is
x≈ (h × R ₀ × V × h) / ((R ₀ × V) · h) (Formula 3)
x: distortion due to foreshortening,
h: target height vector,
R ₀ : vector from Doppler zero to target,
V: moving speed vector,
×: outer product,
・ ： Inner product or product.

  Here, since the geometry is generalized as described above, Equation 3 can also be applied to the case where the target is viewed from the synthetic aperture. In other words, a plurality of SAR images obtained by performing a plurality of imaging by changing the squint angle from the same direction can be used as an input image. A plurality of SAR images having different amounts of distortion can be obtained while maintaining the correlation between the images by changing the squint angle from the same direction. Therefore, by changing the squint angle from the same direction and using a plurality of SAR images obtained by performing a plurality of imaging as input images, the accuracy of shape estimation can be improved.

Here, if the target size is sufficiently smaller than R, the distortion amount x is the same in all locations in an image under the same imaging condition, and the size is proportional to | h |. . That is, as shown in Expression 3, when R ₀ and V are equal, the distortion amount x is in the same direction everywhere in the image, and the magnitude is proportional to | h |. Therefore, when the distortion at the point of the SAR image 1 is x ₁ and the distortion at the point of the SAR image 2 corresponding to the point of the SAR image 1 is x ₂ , the following Expression 4 is established.
x ₁ = a ₁ · | h | · x ₁₀ ,
x ₂ = a ₂ · | h | · x ₂₀
(A ₁ and a ₂ are numerical values and constants, and x ₁₀ and x ₂₀ are vectors and constants) (Formula 4)

Therefore, between the SAR image 1 and the SAR image 2, the following expression 5 is established for the difference Δx of the collapse vector of the corresponding points.
Δx = x ₂ −x ₁ = | h | (a ₂ · x ₂₀ −a ₁ · x ₁₀ ) (Formula 5)

Here, a ₂ · x ₂₀ -a ₁ · x ₁₀ is a constant vector when the image is fixed.
Therefore, the difference in collapse of the corresponding points between the two SAR images is the same direction and the size is proportional to | h | Further, it can be seen from Equation 5 that the coefficient for converting the distortion amount into the height is obtained as the reciprocal of the above constant. That means
| H | = | Δx | / | (a ₂ · x ₂₀ −a ₁ · x ₁₀ ) | (Formula 6)
It is.

That is, the height conversion coefficient calculation unit 54 calculates the above a ₂ · x ₂₀ -a ₁ · x ₁₀ to obtain a coefficient for converting a relative distortion amount difference between a plurality of SAR images into a height. .

  In STEP1-4 (3D shape calculation processing), the 3D shape calculation unit 55 obtains the height of each point of the SAR image based on the height conversion coefficient and the amount of distortion of each minute region. Then, the 3D shape calculation unit 55 outputs 3D shape data representing a three-dimensional distribution of scattered points from the coordinates and height of each point of the SAR image. Here, the 3D shape data is, for example, data (file) including three-dimensional position information of all scattering points in the target region.

  That is, in STEP 1 (shape estimation process), data including three-dimensional position information of all scattering points in the target region is output.

  Next, an outline of STEP2 (distortion correction processing) will be described with reference to FIGS. In STEP 2, the distortion amounts of the SAR image 1 and the SAR image 2 are calculated based on the height information of the 3D shape data 4. Then, based on the calculated distortion amount, a distortion corrected image 7 in which the distortion is corrected is output.

First, in the distortion correction process, the position of each point in the SAR image is corrected. That is, it is calculated which point of the corrected distortion corrected image 7 each point of the SAR image is.
FIG. 10 is a conceptual diagram showing an operation for correcting distortion in the SAR image and restoring it to an image with high visibility. In the case shown in FIG. 10, the distortion amount due to (1) foreshortening of each point of the image is corrected. Thereafter, the distortion amount corresponding to (2) optical collapse of each point of the image is further corrected. As a result of executing (1) and (2), the shadow portion in the SAR image is covered. In other words, by executing (1) and (2), an image that has been collapsed to the SAR side due to foreshortening has fallen to the opposite side of the imaging device (SAR), like an optical photograph. Is displayed. The upper surface of the restored image in FIG. 10 is moved by the amount corresponding to the optical collapse after returning the foreshortening.

Further, in the distortion correction process, the luminance in the corrected image of the point where the layover has occurred due to the foreshortening is obtained. That is, in (1) and (2) above, the position of each point after correction can be obtained, but the luminance after correction of the point that has caused layover due to foreshortening cannot be obtained. That is, for a portion (pixel) where a layover has occurred due to foreshortening, it is necessary to obtain the luminance of the pixel after correction. Therefore, for a portion (pixel) where a layover has occurred due to foreshortening, an appropriate value (pixel luminance) is calculated from the correlation of corresponding points in a plurality of SAR images, and the layover is corrected.
FIG. 11 is a diagram illustrating the concept of layover. Based on the principle of the SAR image, a point that is equidistant from S on a plane that includes S in FIG. 8 and that is perpendicular to V forms an image at a point on the same image. That is, A1, B1, and C1 that are equidistant R1 from S1 of the SAR image 1 in the figure are imaged at the same location, and A2, B2, and C2 that are equidistant R2 from S2 of the SAR image 2 in the figure are the same. The image is formed at That is, in the SAR image, A1, B1, and C1 are displayed as one pixel, and when the A1, B1, and C1 are displayed separately, the luminance of each of the pixels A1, B1, and C1 is unknown. The same applies to A2, B2 and C2. Therefore, as described above, the luminance of each pixel of A1, B1, and C1 is calculated from the correlation of corresponding points of the plurality of SAR images, and the layover is corrected.

  Next, the processing of STEP2 (distortion correction processing) will be described in detail with reference to FIG. FIG. 12 is a flowchart showing the processing of STEP2 (distortion correction processing).

  First, in STEP 2-1 (coordinate calculation process), the coordinate calculation unit 63 calculates the coordinates of each point (lattice point) after distortion correction based on the SAR image and 3D shape data. That is, the coordinate calculation unit 63 calculates the coordinates of the lattice points that are the pixels of the distortion corrected image 7 that is the final output of STEP2. In STEP 2-1, only the coordinate value of the lattice point of the distortion corrected image 7 is obtained, and the brightness of the lattice point is not obtained. That is, in STEP 2-1, information on which position the lattice point is in the image after correcting the distortion is obtained. In STEP 2-1, the coordinate calculation unit 63 mechanically divides the range captured in the SAR image into equal parts (in blocks), and sets the corners of the equally divided areas as lattice points. That is, the coordinate calculation unit 63 calculates the coordinates of the lattice points by, for example, equally dividing the image and calculating the coordinates.

Next, in STEP 2-2 (image corresponding point calculation processing), the image corresponding point calculation unit 64 calculates the distortion amount for each SAR image of the lattice points calculated in STEP 2-1. That is, the image corresponding point calculation unit 64 calculates the distortion amount corresponding to each SAR image based on the coordinates of the lattice points. By obtaining the amount of distortion for each SAR image of the lattice point, the coordinates of the pixel of each SAR image corresponding to the lattice point are obtained. Here, when the distortion amount generated when the relationship between the position of the synthetic aperture radar (platform) and the target position is shown in FIG. 8 is formulated, the distortion amount due to foreshortening is expressed by Equation 5 described above. Further, since the distortion amount y corresponding to the optical collapse occurs in the plane stretched by R and h shown in FIG. .
y≈h × (h × R) / R · h (Formula 7)

STEP 2-2 is performed for all grid points. That is, the coordinates of the pixels of each corresponding SAR image are obtained for all grid points.
That is, in STEP 2-2, it is possible to determine which pixel of each SAR image corresponds to each lattice point of the distortion corrected image 7.

  Next, in STEP 2-3 (coordinate luminance calculation processing), the coordinate luminance calculator 65 calculates the luminance (pixel value) of each corresponding SAR image pixel from each SAR image pixel corresponding to each grid point. . Then, the coordinate luminance calculation unit 65 calculates (estimates) the luminance of the pixel of the distortion corrected image 7 from the luminance of the pixel of each corresponding SAR image. That is, the coordinate luminance calculation unit 65 calculates the luminance of the pixel after correcting the pixel that has been laid over.

As a method for estimating the luminance of the pixel of the distortion corrected image 7, several methods are conceivable.
As a first method, a pixel having the minimum luminance is selected from the pixels of each SAR image corresponding to the pixel for calculating the luminance of the distortion correction image 7, and the luminance is used as the luminance of the pixel of the distortion correction image 7. There is a method to adopt. That is, the coordinate luminance calculation unit 65 sets the minimum luminance among the luminances of the pixels in each of the plurality of SAR images corresponding to each point after distortion correction as the luminance of the point after the distortion correction.
In this method, when a pixel whose luminance is to be calculated does not lay over in any one of a plurality of SAR images, a pixel in the SAR image that does not lay over is selected in principle. Further, in all SAR images, when a pixel whose luminance is to be calculated is laid over, a pixel having the least influence of layover is selected. That is, by selecting the pixel with the lowest luminance from the pixels of each SAR image corresponding to the pixel for calculating the luminance, the luminance closest to the original luminance is obtained from the corresponding pixels in the plurality of SAR images. A pixel can be selected. However, with this method, in all SAR images, accurate luminance cannot be calculated when the pixel whose luminance is to be calculated lays over.

As a second method, all the coordinates of the distortion corrected image 7 superimposed by layover are calculated for each pixel of each SAR image, and the correspondence between the luminance of the pixel of the SAR image and the calculated coordinates of the distortion corrected image 7 is calculated. There is a method of solving simultaneous equations obtained from
That is, the pixel value of the SAR image is considered to be the sum of the luminances of the coordinate points superimposed by layover. Therefore, the coordinate luminance calculation unit 65 derives simultaneous equations based on the correspondence between each point after distortion correction, the pixel in each of the plurality of SAR images, and the luminance of the pixel. The coordinate luminance calculation unit 65 can calculate the luminance of each point after distortion correction by solving the simultaneous equations. That is, all the coordinate points of the distortion corrected image 7 superimposed on a certain pixel of a certain SAR image by layover are calculated. Then, the formula that the sum of the calculated luminances of all the coordinate points is the luminance of the pixel of the SAR image is established. For all the calculated coordinate points, the luminance of each target coordinate point can be obtained by obtaining an equation based on another SAR image and solving the simultaneous equations.
In this method, unlike the first method, in all SAR images, it is possible to calculate an accurate luminance even when a pixel whose luminance is to be calculated is laid over.

  Then, the distortion corrected image generation unit 66 uses the coordinates of each point after distortion correction calculated by the coordinate calculation unit 63 and the luminance of each point after distortion correction calculated by the coordinate luminance calculation unit 65 of the SAR image. A distortion corrected image in which distortion is corrected is generated.

  As described above, since the conventional correction method uses two SAR images having different off-nadir angles, there is a problem that the correlation between the images is deteriorated, and as a result, the shape estimation accuracy is deteriorated. Further, since only foreshortening is corrected, there is a problem that layover cannot be corrected. However, in the distortion correction system according to the present embodiment, since the geometry is generalized, while maintaining the correlation between images obtained by performing multiple imaging by changing the squint angle from the same direction. Correction can be performed using a plurality of SAR images having different distortion amounts. Further, in the distortion correction system according to the present embodiment, it is possible to correct a portion where a layover has occurred by combining information of a plurality of images.

Here, the above embodiment is summarized as follows.
The SAR image distortion correction system according to the above-described embodiment receives, as an input, two or more SAR images obtained by capturing the same part with different synthetic apertures, and correlates each of the images into minute regions for correlation. A distortion amount calculation unit that obtains a difference in relative distortion amounts among a plurality of SAR images by inputting the divided minute area and the SAR image and taking a correlation; and position information of the synthetic aperture included in the SAR image A height conversion coefficient calculation unit that calculates a coefficient for converting the distortion amount into height based on the GCP data, and 3D that outputs 3D shape data based on the height conversion coefficient and the distortion amount of each minute region Based on the shape estimation unit, the lattice point calculation unit that calculates the coordinates of the lattice points of the distortion correction image based on the SAR image and the 3D shape data, the coordinates of the lattice points of the distortion correction image, the SAR image image, and the 3D shape data And each SA corresponding to each grid point An image corresponding point calculation unit that calculates pixels in the image, a calculation result of the pixels in each SAR image corresponding to each grid point, a luminance of each grid point based on the SAR image and 3D shape data, and a distortion corrected image And a grid point luminance estimation unit for outputting

  Further, the distortion correction method using the SAR image distortion correction system generalizes the equation for calculating the amount of distortion so that it can be applied even when the target is viewed from the synthetic aperture. It is characterized in that a plurality of SAR images having different distortion amounts can be obtained while maintaining the correlation between the images by changing the angle and performing a plurality of images.

  Further, the distortion correction method using the distortion correction system for the SAR image obtains the pixel value on the corresponding SAR image from the coordinates of the pixel in each SAR image corresponding to each grid point, and calculates the pixel value from the pixel value. When estimating and outputting the pixel value of the distortion corrected image, the smallest one of the corresponding pixel values is adopted as the pixel value of the distortion corrected image.

  Further, the distortion correction method using the distortion correction system for the SAR image obtains the pixel value on the corresponding SAR image from the coordinates of the pixel in each SAR image corresponding to each grid point, and those pixels. When estimating and outputting the pixel value of the distortion-corrected image from the value, all the coordinate points superimposed by layover for each pixel are calculated, and the pixel value on the other SAR image is obtained for all the coordinates, It is characterized in that the luminance of each target coordinate point is obtained by solving simultaneous equations.

Next, a hardware configuration of the 3D shape processing apparatus and the distortion correction apparatus according to the above embodiment will be described with reference to FIG. FIG. 13 is a diagram illustrating an example of a hardware configuration of the 3D shape processing apparatus and the distortion correction apparatus according to the embodiment.
In FIG. 13, the 3D shape processing apparatus and the distortion correction apparatus include a CPU 911 (also referred to as a central processing unit, a central processing unit, a processing unit, an arithmetic unit, a microprocessor, a microcomputer, and a processor) that executes a program. . The CPU 911 is connected to the ROM 913, RAM 914, keyboard 902, communication board 915, and magnetic disk device 920 via the bus 912, and controls these hardware devices. The 3D shape processing apparatus and the distortion correction apparatus may further include an LCD 901 or the like.

The RAM 914 is an example of a volatile memory. The storage medium of the ROM 913 and the magnetic disk device 920 is an example of a nonvolatile memory. These are examples of the storage device.
The communication board 915, the keyboard 902, and the like are examples of input devices.

The communication board 915 is an example of a communication device.
An operating system 921 (OS), a window system 922, a program group 923, and a file group 924 are stored in the magnetic disk device 920 or the ROM 913. The programs in the program group 923 are executed by the CPU 911, the operating system 921, and the window system 922.

The program group 923 stores programs for executing the functions of the 3D shape processing apparatus and the distortion correction apparatus in the embodiment described above. The program is read and executed by the CPU 911.
In the file group 924, the information, data, signal values, variable values, and parameters described as “˜data” and “˜image” in the above-described embodiment are the items of “file” and “database”. Is remembered as The “file” and “database” are stored in a recording medium such as a disk or a memory. Information, data, signal values, variable values, and parameters stored in a storage medium such as a disk or memory are read out to the main memory or cache memory by the CPU 911 via a read / write circuit, and extracted, searched, referenced, compared, It is used for operations of the CPU 911 such as calculation / calculation / processing / output / printing / display. Information, data, signal values, variable values, and parameters are temporarily stored in the main memory, cache memory, and buffer memory during the operation of the CPU 911 for extraction, search, reference, comparison, calculation, calculation, processing, output, printing, and display. Is remembered.
In the embodiment described above, the arrows in the processing flow and flowchart mainly indicate input / output of data and signals, and the data and signal values are the memory of the RAM 914, the flexible disk of the FDD904, the compact disk, and the magnetic disk. The data is recorded on a recording medium such as a magnetic disk, other optical disk, mini disk, or DVD (Digital Versatile Disc) in the apparatus 920. Data and signals are transmitted online via a bus 912, signal lines, cables, or other transmission media.

  In the embodiment described above, what is described as “to part” may be “to circuit”, “to device”, “to device”, “to means”, and “to step”. ”,“ ˜procedure ”, or“ ˜processing ”. That is, what is described as “˜unit” may be realized by firmware stored in the ROM 913. Alternatively, it may be implemented only by software, or only by hardware such as elements, devices, substrates, and wirings, by a combination of software and hardware, or by a combination of firmware. Firmware and software are stored as programs in a recording medium such as a magnetic disk, a flexible disk, an optical disk, a compact disk, a mini disk, and a DVD. The program is read by the CPU 911 and executed by the CPU 911. That is, the program causes the computer to function as “to part” in the embodiment described above. Alternatively, in the embodiment described above, the computer executes the procedure and method of “to part”.

1 is an overall configuration diagram of a SAR image distortion correction system 100 (ortho correction processing apparatus) of a SAR image according to Embodiment 1. FIG. The conceptual diagram which shows the concept of the SAR foreshortening using a satellite-mounted antenna. The functional block diagram which shows the function of the shape estimation process part 5. FIG. FIG. 3 is a functional block diagram showing functions of a distortion correction processing unit 6. 5 is a flowchart showing an operation in an SAR image orthocorrection process according to the first embodiment. The conceptual diagram which shows the distortion of the target which appears in a SAR image. The conceptual diagram which shows the foreshortening, layover, and shadow in a SAR image. The conceptual diagram at the time of generalizing the position of a synthetic aperture regarding the relational expression of the said foreshortening amount and a synthetic aperture. The flowchart which shows the process of STEP1 (shape estimation process). The conceptual diagram which shows operation which correct | amends the distortion in a SAR image, and decompress | restores in an image with high visibility. The figure which shows the concept of layover. 7 is a flowchart showing a STEP 2 (distortion correction process) process. The figure which shows an example of the hardware constitutions of the 3D shape processing apparatus which concerns on embodiment, and a distortion correction apparatus.

Explanation of symbols

  1, 2 SAR image, 3 GCP data, 4 3D shape data, 5 shape estimation processing unit, 51, 61 SAR image input unit, 52 correlation region division unit, 53 distortion amount calculation unit, 54 height conversion coefficient calculation unit, 55 3D shape calculation unit, 56 GCP data input unit, 6 distortion correction processing unit, 62 3D shape data input unit, 63 coordinate calculation unit, 64 image corresponding point calculation unit, 65 coordinate luminance calculation unit, 66 distortion correction image generation unit, 7 Distortion correction image, 100 SAR image distortion correction system, 902 K / B, 911 CPU, 912 bus, 913 ROM, 914 RAM, 915 communication board, 920 magnetic disk device, 921 OS, 922 window system, 923 program group, 924 Files.

Claims (7)

A SAR image input unit that inputs a plurality of SAR (Synthetic Aperture Radar) images obtained by imaging the same portion with different synthetic apertures and stores them in a storage device;
A distortion amount calculation unit that calculates a difference in distortion amount between a plurality of SAR images input by the SAR image input unit by a processing device;
A height conversion coefficient calculation unit that calculates a height conversion coefficient for converting a difference in distortion amount into height based on the plurality of SAR images and GCP (Ground Control Point) data representing the shape of the ground;
Based on the difference between the distortion amounts calculated by the distortion amount calculation unit and the height conversion coefficient calculated by the height conversion coefficient calculation unit, the height of the portion imaged in the SAR image is calculated, A 3D shape calculation device for a SAR image, comprising: a 3D shape calculation unit that obtains 3D (Dimensional) shape data of a location imaged in the SAR image and stores the data in a storage device. The 3D shape calculation device further includes:
A correlation area dividing unit that divides each of the plurality of SAR images into a plurality of minute areas;
The distortion amount calculation unit obtains a correlation between each of the plurality of SAR images in each micro region of the plurality of micro regions divided by the correlation region dividing unit, so that a relative relationship between the plurality of SAR images is obtained. The apparatus for calculating a 3D shape of a SAR image according to claim 1, wherein a difference in distortion amount between the plurality of SAR images is calculated by obtaining a positional shift. The SAR image input unit inputs a plurality of SAR images captured by changing the squint angle from the same direction,
The height conversion coefficient calculation unit calculates a height conversion coefficient for converting a difference in distortion amount into a height based on a plurality of SAR images captured by changing the squint angle from the same direction and the GCP data. The SAR image 3D shape calculation apparatus according to claim 1, wherein 4. The height conversion coefficient calculating unit calculates a height conversion coefficient for converting a distortion amount into a height based on an expression 2 obtained based on an expression 1 representing the distortion amount. SAR image 3D shape calculation device.
x≈ (h × R ₀ × V × h) / ((R ₀ × V) · h) (Formula 1)
| H | = Δx / (a ₂ · x ₂₀ -a ₁ · x ₁₀ ) (Formula 2)
x: distortion due to foreshortening,
h: target height vector,
R ₀ : vector from Doppler zero to target,
V: moving speed vector,
Δx: difference of collapse vector of corresponding points between two SAR images,
a ₁ , x ₁₀ , a ₂ , x ₂₀ : constant (a ₁ , a ₂ : numerical value, x ₁₀ , x ₂₀ : vector)
×: outer product,
-: Inner product or product A SAR image input unit that inputs a plurality of SAR (Synthetic Aperture Radar) images obtained by imaging the same portion with different synthetic apertures and stores them in a storage device;
A 3D shape data input unit that inputs 3D (Dimensional) shape data of a portion captured in a plurality of SAR images input by the SAR image input unit and stores the data in a storage device;
A coordinate calculation unit that calculates the coordinates of each point after distortion correction of the plurality of SAR images based on the 3D shape data input by the 3D shape data input unit;
An image corresponding point calculation unit that calculates a pixel in the SAR image corresponding to each point after distortion correction for which the coordinate calculation unit has calculated the coordinates by a processing device;
A coordinate luminance calculation unit that calculates the luminance of each point after the distortion correction by the processing device based on the luminance of the pixel in each of the plurality of SAR images corresponding to each point after the distortion correction calculated by the image corresponding point calculation unit. When,
A distortion-corrected image generation unit that generates a distortion-corrected image in which the distortion of the SAR image is corrected based on the coordinates of each point after the distortion correction and the luminance of each point after the distortion correction calculated by the coordinate luminance calculation unit. A distortion correction apparatus for a SAR image, comprising: The coordinate luminance calculation unit is characterized in that the minimum luminance among the luminances of the pixels in each of the plurality of SAR images corresponding to each point after the distortion correction is set as the luminance of the point after the distortion correction. The SAR image distortion correction apparatus according to claim 5. The coordinate luminance calculation unit solves the distortion-corrected points by solving simultaneous equations derived based on the correspondence between each point after the distortion correction, the pixel in each of the plurality of SAR images, and the luminance of the pixel. The SAR image distortion correction apparatus according to claim 5, wherein the luminance of the SAR image is calculated.

Images