JPH0750503B2 - Synthetic Aperture Radar Distance Direction Deblurring Method - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital signal processing method for reproducing an image from a received signal of a synthetic aperture radar, and more particularly to a geometrical feature called a range carbacher included in the received signal. The present invention relates to a signal processing method for removing a blur in a distance direction of a reproduced image due to distortion to obtain a reproduced image of high quality.

[Background of the Invention]

The synthetic aperture radar image is reproduced from the received signal by signal processing that correlates not the received signal itself but a function called a reference function with the received signal. The received signal data is two-dimensional sample data. One of the two coordinate axes indicates the emission direction of the radar pulse and is called the distance direction. The other is the azimuth direction, which refers to the traveling direction of a flying object equipped with a synthetic aperture radar. In the synthetic aperture radar, the observation information of any one point on the ground surface is a two-dimensional image so-called a blurred image that extends over 10 ⁶ sample data on the received signal. In the signal processing, first, a one-dimensional correlation calculation in the distance direction called distance compression is performed, and then a one-dimensional correlation calculation in the azimuth direction called direction compression is performed. By the distance compression, the plane image becomes a linear image along the azimuth direction, so to speak, and by the azimuth compression, it becomes a point image, and becomes an observation image of the ground surface. At this time, the line image has range migration distortion that is not parallel to the azimuth direction but is distorted in a parabolic shape.
In azimuth compression, received signal resampling (requantization)
To correct this distortion. The slope of the range migration distortion at the center of the line image is called range walk distortion, and the distortion obtained by removing the range walk distortion from the range migration distortion is called range coverage. In digital signal processing, the correlation is usually performed in the frequency domain in order to speed up the series of compressions. The same applies to distance compression, but in compression, the line image is first Fourier-transformed in the azimuth direction. As a result, only the azimuth direction changes to the frequency axis.
At this time, the range walk distortion causes blurring in the line image in the distance direction. When this blur is represented by A, its Fourier transform is represented by the following equation.

_b = radar pulse frequency bandwidth K _D = Doppler FM ratio _D = Doppler frequency λ = radar pulse carrier wavelength C = speed of light Regarding conventional techniques for suppressing the derivation and occurrence of this blur, International Geography and Remote Sensing Symposium ( 1983, IEEE Geography and Mote Sensing Society)
(International Geoscience and Remote Sensing Symp
osium (1983, IEEE Geoscience and Remote Sensing S
[Synthesis Aperture Radar Correlation Method-Algorithm for Data Processing with Large Range Walk] by M. Jin et al. (“SAR Correlatio
n Technique-An Algorithm for Processing Data with
Large Range Walk "). In this technique, first, the Fourier transform _R of the distance direction blur restoration function is obtained from the equation (1).

Here, * indicates a complex conjugate.

Next, this is multiplied by the Fourier transform of the distance compression reference function R (t), and _R is used as a new distance compression reference function. As a result, it is possible to prevent blurring in the distance direction that may occur due to azimuth compression.

A principle diagram of the prior art is shown in FIG. Reference numeral 1 denotes a received signal file which stores received signal data. Reference numeral 2 denotes a telemetry file, which stores measurement data of the position and velocity of a flying object equipped with a synthetic aperture radar at the time of observation and a radar constant of the synthetic aperture radar. A parameter rough estimation unit 3 calculates the values of the two parameters K _D and _{D in} the equation (1). In this calculation, the telemetry file data is used as an initial condition by using a mathematical model that captures the relative motion between the three considerations of the earth-flying body-synthetic aperture radar. Reference numeral 4 denotes a distance direction blur restoration function generation unit that generates the function of Expression (2). Reference numeral 5 is a distance compression unit. If the Fourier transform of the received signal is written, the distance compression in the distance compression unit 5 is
It is performed by the equation (3).

g (t) = F ^-1 [. {. _R } ^* ] (3) where F ^-1 is the inverse Fourier transform, * is the complex conjugate, g
(T) shows an output signal of distance compression.

By the way, in the signal processing of the synthetic aperture radar, a look division reproduction method is adopted in which an image (referred to as a look) of the same scene is reproduced from each signal by dividing the received signal into several bands having different Doppler frequencies. This is because one image obtained by adding and averaging a plurality of reproduced looks has lower noise than each look. In the case of this method, focusing on each look, the band center frequency _D differs depending on the look. Therefore, the range walk distortion at _D for each look is different due to the range curvature. This indicates that it is desirable for high image quality reproduction to take different blur suppression measures for each look against the blur in the distance direction due to range walk distortion. However, when the correlation in the distance compression is performed in the frequency domain, the value of _{D in} the equation (2) can be only one for all looks, and cannot be changed depending on the look and the distance in the distance direction. As a result, different blur appears in the reproduced image depending on the look and the distance. This is a drawback of the prior art.

[Object of the Invention]

SUMMARY OF THE INVENTION It is therefore an object of the present invention to provide a method for distance blur removal which compensates for the drawbacks of the prior art described above and reduces distance blur for all looks.

[Outline of Invention]

In order to achieve the above-mentioned object, most of the distance direction blurring of each look, which cannot be suppressed during the distance compression in the present invention,
In azimuth compression, it is characterized in that it is removed by convolutional integration performed in time distance to correct range migration distortion.

The principle of the present invention will be described below. Now, the band center frequency _D of a certain Doppler frequency is represented by the following equation. _D = f _DC + Δ _D (4) Here, _DC is the Doppler band center frequency at the scene center, and Δ _D is the deviation of _D from f _DC . _{Since D} is a function of not only the look but also the distance d in the distance direction, Δ _D is also a function of the distance d in the distance direction. Also,
Since the Doppler FM ratio K _D in the equation (1) is also a function of d, it is expressed as follows like _D.

_{K D = K DC + ΔK D} (5) where, [Delta] K _D in the K _D, a deviation from K _DC, this and K _DC
Is a look-independent quantity. By substituting the equations (4) and (5) into the equation (1) and rearranging, the following equation is obtained.

Therefore, the equation (3) is expressed as the following equation.

g (t) F ^-1 [・ {・_R } ^* ] ＝ F ^-1 [・ (・_R ) ^* ] ・ Δ ▲ ^* _R ▼] (7) where By further modifying equation (7), equation (12) is obtained.

Equation (12) is expressed as follows: _D = _{D for} compression that minimizes the blur in the distance direction at any look and at any distance in the distance direction.
Output signal of distance compression with _DC , K _D = K _DC and F ^-1 [Δ ▲ ^*
It is shown that this can be realized by the convolution integral of _R ▼].
This equation gives the principle of the invention.

According to the present invention, it is possible to obtain a reconstructed image in which the distance direction blur is removed by adjusting the interpolation kernel at any look and at any distance in the distance direction.

Examples of the Invention Hereinafter, the present invention will be described in detail with reference to Examples. FIG. 2 is an embodiment diagram. 1 to 5 are the same as the blocks of corresponding numbers in FIG. The parameter precision estimation unit 6 analyzes the power spectrum and the phase in the azimuth direction of the output signal of the distance compression unit 5 to accurately determine the error contained in the values of K _D and _D in the output of the parameter rough estimation unit 3. Then, based on this result, K _D and _D are corrected and output.
For details on the parameter accurate estimation process, refer to 15th International Symposium on Remote Sensing of Environment, Ann Alber, MI, May 1985 (The Fifteenth Internatin
al Symposium On Remote Sensing Of Emironment, Ann A
Rbor, MI, May, 1981) “The future of generalized digital synthetic aperture radar processors” by JR Bennett et al.
eralized Digital Synthetic Aperture Radar Processo
r ").

Reference numeral 7 is a look division information file. The offset value DF _i necessary for determining the band center frequency _Di of each look i is stored. _Di is defined by _D + DF _i , where _D is the Doppler band center frequency of the received signal and is obtained as the output of 6.

Reference numeral 8 is an interpolation kernel generation unit. Here, from the output K _D , K _DC , _D , _{DC of 6} and DF _i given by 7, for each look i, the correction kernel in equation (12) is obtained for each sample point taken in the distance direction. Calculate F ^-1 [Δ ▲ ^* _R ▼]. In this calculation, delta _{D is} the _{D Di}
In other words, according to the definition formula (4), Δ _D = _Di − _DC = _D
+ DF _i − _DC . The above-mentioned function to be obtained is written as W (t).

W (t) = F ^-1 (Δ ▲ ^* _R ▼) (13) In the azimuth compression section 9, first, the output signal compressed by the distance is Fourier-transformed in the azimuth direction, and then the range migration distortion included in the signal is calculated. to correct. Since this distortion is geometric distortion of the signal, distortion correction involves resampling to requantize the signal. Now, if a signal is represented by (d _i , _i ) _i = 1, ..., N _d ; j = 1, ..., N _{f, then} at a given distance d obtained by one-dimensional resampling in the distance direction of the signal, New sample data value '(d, _j )
Is expressed by the following equation.

Here, W (Δd) is an interpolation kernel and has the following relationship with the equation (13).

W (Δd) = W (2Δd / C + T ₀ ) (15) where T ₀ : constant depending on how the origin of the sampling time in the range direction is taken C: speed of light because sampling time t in the range direction in equation (13) And the distance direction coordinate d in the reproduced image has a relationship of t = 2d / C + T ₀ (16). The first term on the right side of the equation (16) represents the time required for the radar signal to make a round trip in the section of the distance d.

In order to reduce the calculation amount of the equation (14), W (Δ
d) is approximated as follows.

This is because the sum of products calculation of the equation (14) is limited to the range of | Δd | ≦ D.

Similar to the distance compression, the signal with corrected range curve distortion is multiplied by the complex conjugate function of the reference function for azimuth compression and then inverse Fourier transformed to become the output signal of azimuth compression, that is, the reproduced image. , Stored in the reproduced image file 10.

〔The invention's effect〕

According to the present invention, most of the blur in the distance direction that can occur in each look of the reproduced image and the peripheral portion in the distance direction can be removed in the resampling process of the azimuth compression because the suppression is insufficient only by the conventional technique. , With all looks, clear images can be obtained up to the peripheral area in the distance direction.

[Brief description of drawings]

FIG. 1 is a block diagram of a prior art, and FIG. 2 is a block diagram of an embodiment of the present invention. Explanation of code 1 …… Received signal file 2 …… Telemetry file 3 …… Parameter rough estimation unit 4 …… Distance direction blur restoration function generation unit 5 …… Distance compression unit 6 …… Parameter precise estimation unit 7 …… Look division information File 8 …… Interpolation kernel generator 9 …… Direction compressor 10 …… Reproduced image file

Claims (1)

[Claims] 1. A synthetic aperture radar image reproduction processing method for dividing a received signal into a plurality of looks in the azimuth direction and reproducing an image from each of the looks, which is caused by geometric distortion of a range curvature included in the received signal. Then, the distance-direction blurring which appears differently in the reproduced image due to the look is removed by convolution integration in the distance direction between the received signal and an interpolation kernel different for each look.