JPS58186068A - Focusing system of synthetic aperture radar picture - Google Patents

Abstract

PURPOSE:To form a synthetic aperture radar picture having a high picture quality, by deciding an error from a picture position shift between looks corresponding to a peak of a mutual correlation function of two look picture data, and compensating an initial Doppler variation factor. CONSTITUTION:A correlation calculation of two look picture data is executed by shifting a picture position by one picture element each, by a two look picture correlation calculating means 15 of an azimuth direction compressing system 4, and a position shift quantity and a correlation quantity are supplied to a correlation maximum position searching means 16. Subsequently, a position shift which becomes a peak of a mutual correlation function is measured by the means 16, an initial Doppler variation factor error based on the position shift from the means 16 is decided by a Doppler variation factor correcting means 17, and also an error of an initial Doppler variation factor is compensated. According to a focusing system for compensating the initial Doppler variation factor which exert big influence on deterioration of its picture quality, to an exact value, a synthetic aperture radar picture having a high picture quality can be formed.

Description

DETAILED DESCRIPTION OF THE INVENTION Object of the Invention The present invention relates to a synthetic aperture radar (Syntheti, C Aperture) mounted on an artificial satellite or an aircraft.
The present invention relates to an image focusing method in a digital processing system for reproducing human-understandable images from imaging data obtained by Radar (hereinafter referred to as "5ARJ").

Conventional technology Satellites, aircraft, etc., or public funds for MoB sensing, are used as sensors to image the ground surface, and SA is capable of obtaining high-resolution images in the microwave band that penetrates clouds.
R is attracting attention.

Figure 1 shows the entire SAR system. A BAR having a radar sensor R and an antenna An is mounted on an artificial satellite or the like and images the ground surface while moving in the direction of an arrow A on a flight path Fp. Imaging data from the BAR is received by the ground station L8 and processed by the data processor Dp to create a photographic film 1, a magnetic tape MT for data storage, etc.

In addition, 0 is the decomposition cell, Ra is the range direction on the ground surface of the data collected by BAR, and the same axis Y direction is
A indicates the antenna beam, and OW indicates the cutting width.

An overview of the processing of data collected by SAR will be described below. For details, see, for example, J. RoBenne.
tt at ano, “A DigitalJ Prooe
Saor for th@Produ-tion o
f 5eaaat 5ynthsio Apertu
re R&MUR engineering mager7”197Gl, engineering taz
ytt#I want to be illuminated.

In the BAR received image, one point on the original image is distributed with the spread of a point spread pattern h (z, y),
Humans cannot understand this. Here, X indicates the range direction, and y indicates the azimuth direction. The information spread in the received image is first compressed in the range direction and then in the azimuth direction. The compression processing is performed by correlation processing with point image pattern data for each line of image data. However, if Correligiron processing is executed as it is, it will take a significant amount of processing time, so Fast 7-lier transform (hereinafter referred to as "F7T"),
Complex multiplication, fast inverse Fourier transform (hereinafter referred to as ``PFT'')
”. ) to speed up the process.

- The point image pattern h(y) used in the compression process in the azimuth direction described above is a linear frequency modulation signal, and h(y)
−αQO8(T Ky 1βy)1・I@・(1). Here, α+K+'l is a constant, of which K is the rate of change in frequency, which is called the Doppler rate of change since the frequency change is due to Doppler shift.

In order to improve the image quality of the image data after the compression processing, it is important to accurately obtain point image pattern data. The Dotsupura change rate includes orbit data of artificial satellites,
It is determined based on attitude data, earth rotation data, etc., but if there is an error in the data used here, the Doppler change rate K will also have an error from the true value, which will appear as a blur pattern in the output image. Figure 2 shows an example.
A1 is the original image, A2 is the point image pattern h(x, y) of two points on A1, and A is the point image pattern h(x, y) of A2.
+y) is compressed in the range direction and the azimuth direction, which schematically shows a blur pattern caused by an error in the Doppler change rate.

In addition to this, multi-look processing is required for SAR data processing. Multi-look processing aims to reduce the speckle noise characteristic of coherent systems such as BAR processing and obtain smooth images.
When azimuthal compressing 11 images, the band is divided in the frequency space immediately before the FFT, a compressed image is obtained for each divided band, and the intensities of these images are added. As a result, since the frequency band becomes narrower, the resolution of the output image decreases to some extent, but it is possible to improve the S/M ratio (in proportion to the number of divisions Σi).

Conventionally, a trial-and-error method has been known as a focusing method for removing an image blur pattern caused by the error in the Doppler change rate in the SAR output image.

This method adds or subtracts a certain value to the Doppler change rate to create a point image pattern h (y), performs azimuth compression on the point image pattern h (y), and evaluates the image quality of the output image obtained. The purpose is to find the Doppler change rate that gives the best image quality.

However, this method had the following drawbacks.

The first is that since it is a trial and error method, it takes an enormous amount of calculation time to find the Doppler change rate that provides the best image quality. Even if the image output for image quality evaluation is reduced to a partial image to reduce the amount of processing, the point image pattern h
The amount of processing required to repeat the calculation of b> and the azimuth compression process the same number of times by trial and error is large.

The second drawback is that it is difficult to evaluate the image quality of the compressed image. Even if it is possible for humans to compare two BAR images and determine which one has better image quality with less blur, it is difficult to do this automatically using computer processing. In order to automatically search an image for a point image whose blur amount is easy to measure, image compression processing must be performed over the entire screen. It is not impossible to store the position coordinates of a point image in the system in advance and estimate the position in the image, but this is difficult in the case of a system that processes received images of all parts of the earth. .

As mentioned above, the trial and error method has serious drawbacks both in terms of throughput and in terms of image quality evaluation.

Purpose of the Invention The present invention has been made in view of the above circumstances, and its purpose is to eliminate the above-mentioned drawbacks of the conventional focusing method, and to solve the cause of the deterioration of the image quality of the output image in the 8AR image processing system. An object of the present invention is to provide a BkR image focusing method that uniquely determines the error in the Doppler change rate from the partial output image, corrects the Doppler change rate, and outputs a high-quality SAR image.

The above-mentioned object of the present invention is to provide an image processing system for reproducing an image from imaging data obtained by SAR, by providing a means for partially obtaining multi-look image data, and by providing a means for partially obtaining multi-look image data, and by providing a means for partially obtaining multi-look image data. The positional shift of the II image between looks is measured by determining the peak position of , and the error in the initial Doppler change rate is determined from this measurement result to correct the initial Doppler change rate.
A. A cup achieved by a one-image focusing method.

Embodiments of the Invention Prior to describing embodiments, the present invention will be explained in detail. Third
The figure is a diagram for explaining the present invention in detail. Let the true Doppler change rate be X&, and let the initial Doppler change rate be KIL10aK. δ is the error in the Doppler rate of change. The mechanical FFT processing in the azimuth direction compression processing is shown in Fig. 3 (
For the frequency band ω in A), if there is no error in the Doppler rate of change, the information of one point spread over the band ω is
As a result of the colliding, the points are concentrated at one point and a point image P is reproduced.

However, if there is an error in the Doppler rate of change during the azimuth compression process, the direction of one point expands in the azimuth direction and becomes like y. FIG. 3(B) is an enlarged view of this, and b indicates the amount of blur of the point image.

To illustrate this more specifically, as shown in FIG. (including the entire shaded area).

Since there is a relationship between the amount of blur indicated by the dotted line and the Doppler change rate Ka, the same error δX, and the width of the bandwidth ω, it is possible to find the error δK in the Doppler change rate from the amount of blur. The amount of blur cannot be measured with a general image that is not a point image, as shown in FIG. 3 (0).

Therefore, in the present invention, as shown in FIG. 3(D),
Images R1R' are created for two different bands, and the amount of positional deviation d between these two reproduced images is measured.

Figure 3 (E) shows this enlarged, and Figure 3 (F) shows this more specifically. The original pattern Q on the ground surface (the part shaded in two directions) ) is S
After the AR compression process, a blur pattern Q' (the entire hatched area including Q) and a blur pattern Q with the same type and position shift as Q' shown by the broken line are obtained. Both of the blur patterns Q' and Q' include blur, but since they have the same blur position and can be treated as including blur, this does not pose a problem in determining the amount of shift.

For this reason, the amount of deviation d can be easily determined for general image patterns from the peak value of the mutual phase M number of the images (d'' in FIG. 4). Positional deviation amount d and frequency distance between two frequency bands! The Doppler change rate error δK can be determined by the following formula based on , and the correct Doppler change rate KIL can be determined from the result.

K j lower, 1 ・・・・・・(3)a
a The above is the principle of the present invention, and the correct Doppler change rate Ka
By using this, it is possible to obtain a high-quality, well-focused EIAR compressed image without blur.

Embodiments of the present invention will be described below based on the drawings. Fifth
The figure is a diagram showing an overview of the SA1 image processing system, in which the system 2 compresses received SAR data 1 in the range direction.
Device 3I for vertically and horizontally transposing compressed two-dimensional data in range direction
The system 4 compresses @ position data in the azimuth direction, and the storage device δ stores compressed image data. Further, FIG. 6 is a diagram showing details of the azimuth direction compression system 4. As shown in FIG. Azimuth direction compression system 4c, FFT means 6, range curvature [E means 7.
Dotsupura change rate storage means 81 points image fighting number creation means 9,
IIT means 10. 10 stages of multiplication 111 frequency band data extraction means 12. ■FFT means 13° Complex data norming means 14.2 Look image competition calculation means 15. Phase opening maximum position search means 16. Dotsupura change rate correction means 17
．． It is composed of m-image data addition means 18 and addition data quantization means 19.

The operation of the apparatus of this embodiment configured as described above will be explained below.

The range compressed data is converted into frequency space by the FFT means 6, and corrected by the range curvature correction means 7, which corrects blur caused by changes in the distance between the artificial satellite and the observation point. The initial Doppler change rate &'-X0δX determined from the orbit data of the artificial satellite, etc. is read out from the storage means 8 and input into the point image number creation means 9, and is calculated by the above formula (〇).
Point image data is created according to the following. The point image data is 1!
After being Fourier transformed by the FT means 10, the complex multiplication means 11 multiplies the image data on the frequency space.

The result of this multiplication, that is, the frequency space data of the compressed image data is processed by the frequency band data extraction means 12.
As shown in FIG. 3(D), the band data is extracted as two band data separated by a wave number distance l, and the IFF'l' and norming are performed using the means 13°1.
4, resulting in one line of compressed image data.

The above two-look compression process is performed on part of the image data, and two partial images as shown in FIG. 3 are obtained. The two-look image competition calculation means 15 calculates these two
Correlation calculation between the two partial images is performed while shifting the positions of the two images one pixel at a time, and the positional shift amount and correlation value are output to the maximum correlation position searching means 16.

The maximum correlation position search means 16 determines the positional deviation amount d (see FIG. 4) that gives the maximum correlation value. The Doppler change rate correction means 17 corrects this positional deviation d.

The true Doppler change rate KIL is calculated from the 2-look band distance @1 and the initial Doppler change rate & by the following formula.

x, -t/ (d+x; )...
...←) After finding the true Doppler change rate from the partial output image as described above, using this value, conventional azimuth compression processing is performed on all image data. That is, (the compressed images corresponding to the number of effective looks are subjected to look addition by the image data addition means 18 to improve the S/N ratio, and then converted into integer type image data by the interpolation means 19.

Note that by correcting the Doppler change rate, the blur pattern of the compressed image is removed, and positional deviations between look images are also removed. Another advantage is that no processing is required.

The two-look image correlation calculation means 15, the maximum correlation position search means 16, and the maximum correlation position search element ff shown in the above embodiment.
17 is constructed by a combination of WK logic circuits, but it goes without saying that these processes can also be programmed by a microcomputer.

Effects of the Invention As described above, according to the present invention, in an image processing system that reproduces an image from image data captured by BAR, a means for partially obtaining multi-look image data is provided, and two looks obtained by the means are provided. By determining the peak position of the mutual phase vA function of the image data, the positional shift of the image between looks is measured, and from this measurement result, the error in the initial Doppler change rate is determined and the initial Doppler change rate is corrected. This has the remarkable effect of making it possible to quickly obtain an image free of blur patterns.

[Brief explanation of the drawings] Fig. 1 shows the entire SAR system, Fig. 2 shows an example of an original image, a received image, a compressed image, and a compressed image. Figure 4 is a graph showing an example of the correlation function of two-look images; @5S is a diagram showing an overview of the SAR[i image processing system; FIG. 1 is a configuration diagram of an azimuth direction compression processing system showing one embodiment. 4: Azimuth direction compression system, 6; IFFT means, 7: Range curvature correction means, 8: ) Tsufura change rate storage means, 9: Point spread function creation means, 10 Near 7T means, 111 Complex multiplication means, 12: Frequency In-band data extraction means, 13: 17 FT means, 14: Complex data norming means, 15: 2-look image phase flash means, 1
6: Maximum correlation position search means, 17: Doppler change rate correction means, 188 image data addition means, IGM addition data quantization means. (To) Figure 1 Figure 2 Figure 3 Prisoner Figure 3 1 Figure 4 Soyuki IA Figure 5 6 Figure Continued from page 1 0 Inventor Hiroshi Ihara - 1099 Ozenji Co., Ltd., Tama-ku, Kawasaki City Inside Hitachi, Ltd.'s System Development Laboratory Inventor Yutaka Kubo 5-2-1 Hitachi University Mika-cho Inside Inumiga Factory, Hitachi, Ltd.

Claims (1)

[Claims] In an image processing system that reproduces an image from imaging data obtained by a synthetic aperture radar, a means for partially obtaining multi-look image data is provided, and a peak position of a correlation at of two look image data obtained by the means is determined. A method for focusing a synthetic aperture radar image, characterized in that the positional shift of the image between looks is measured by this method, and the error in the initial Doppler change rate is determined from the measurement result to correct the initial Doppler change rate.