阅读说明：本技术 基于自然均匀目标的合成孔径雷达在轨快速定标方法 (Synthetic aperture radar on-orbit rapid calibration method based on natural uniform target ) 是由 易明宽 李传荣 马灵玲 王新鸿 汪琪 王宁 赵永光 唐伶俐 郑青川 于 2019-09-05 设计创作，主要内容包括：一种基于自然均匀目标的合成孔径雷达在轨快速定标方法,该方法包括：对需定标合成孔径雷达的未校正图像进行单元格划分；对未校正图像剔除不均匀区域单元格；之后对未校正图像的所有方位条带数据进行平均,再对距离条带拟合,求得天线增益G并对未校正图像进行相对校正得到相对校正图像；使用已校正图像作为基准图像,基于基准图像和未校正图像特征点,对两幅图像进行配准；根据处理后的基准图像和相对校正图像计算绝对定标参数解,求得绝对定标系数与偏移系数。本发明基于已定标合成孔径雷达基准图像和相对校正图像中分布式均匀目标,对在轨合成孔径雷达进行快速定标,省略了摆放人工参考设备的流程,减少了耗费的人力物力,加快定标实验进程。(An on-orbit fast calibration method of a synthetic aperture radar based on a natural uniform target comprises the following steps: dividing cells of an uncorrected image of the synthetic aperture radar to be calibrated; removing uneven area cells from the uncorrected image; averaging all azimuth strip data of the uncorrected image, fitting distance strips, obtaining antenna gain G, and performing relative correction on the uncorrected image to obtain a relative corrected image; using the corrected image as a reference image, and registering the two images based on the reference image and the characteristic points of the uncorrected image; and calculating an absolute calibration parameter solution according to the processed reference image and the relative correction image to obtain an absolute calibration coefficient and an offset coefficient. The method is based on the calibrated synthetic aperture radar reference image and the distributed uniform targets in the relative correction image, the on-orbit synthetic aperture radar is quickly calibrated, the process of placing manual reference equipment is omitted, the consumed manpower and material resources are reduced, and the calibration experiment process is accelerated.)

1. An on-orbit fast calibration method of a synthetic aperture radar based on a natural uniform target comprises the following steps:

dividing cells of an uncorrected image of the synthetic aperture radar to be calibrated;

carrying out consistency check on an uncorrected image of the synthetic aperture radar to be calibrated to remove cells in an uneven area;

averaging all azimuth strip data of the uncorrected image after the uneven area cells are removed, then fitting the distance strip data to obtain antenna directional pattern gain G (theta), and performing relative correction on the image by using the antenna directional pattern gain G (theta) to obtain a relative correction image of the synthetic aperture radar to be calibrated;

using the corrected image as a reference image, and performing registration processing on the reference image and the uncorrected image based on the characteristic points in the reference image and the uncorrected image;

and calculating an absolute calibration parameter solution according to the registered reference image and the relative correction image of the synthetic aperture radar to be calibrated, and solving an absolute calibration coefficient gain and an offset coefficient offset.

2. The natural uniform target-based synthetic aperture radar in-orbit fast calibration method according to claim 1, wherein the registering the reference image and the uncorrected image comprises:

selecting ground control points on the reference image and the uncorrected image to register the two images;

and performing scaling processing on the reference image to enable the azimuth resolution and the distance resolution of the reference image to be consistent with those of the original image.

3. The natural uniform target-based synthetic aperture radar in-orbit fast calibration method according to claim 2, wherein the ground control points on the reference image and the uncorrected image are selected as obvious feature points on the images of the reference image and the uncorrected image.

4. The natural uniform target-based synthetic aperture radar in-orbit rapid calibration method according to claim 2, wherein the scaling processing is performed on the reference image by using a bilinear interpolation algorithm or a cubic convolution interpolation algorithm.

5. The natural uniform target based on synthetic aperture radar in-orbit fast calibration method according to claim 2, wherein the reference image is a corrected image of a calibrated satellite-borne or airborne load at the same time, at the same pitch angle and at the same azimuth angle, and the resolution of the reference image is between 0.2 and 5 times of the resolution of the image to be calibrated.

6. The method for fast on-orbit calibration of a synthetic aperture radar based on a natural uniform target according to claim 1, wherein the calculating an absolute calibration parameter solution according to the processed reference image and the relative calibration image of the synthetic aperture radar to be calibrated to obtain an absolute calibration coefficient gain and an offset coefficient offset comprises:

taking the DN value of the processed reference image as the true value of the backscattering coefficient of the target area of the synthetic aperture radar to be calibrated relative to the corrected image, the relationship between the pixel values of the two images at coordinates (i, j) can be expressed as:

DN_1i，j＝gain·DN_2i，j+offset

wherein DN_1i，jFor the DN value of the ith row and jth column unit cell of the processed reference image_2i，jThe method comprises the steps of obtaining a DN value of an image subjected to relative correction at the ith row and jth column cells in a synthetic aperture radar relative correction image to be calibrated;

order to

Where N represents the number of all pixels in the reference image scaling target area, the above equation can be written as:

DN₁＝DN₂·G

the absolute scaling coefficient gain and the offset coefficient offset are solved using the least square method:

G＝(DN₂ ^T·DN₂)^-1·DN₂ ^T·DN_1。

7. the method for fast in-orbit natural uniform target-based synthetic aperture radar calibration according to claim 1, wherein the step of performing consistency check on the uncorrected image of the synthetic aperture radar to be calibrated to remove the uneven area cells comprises:

based on the parameter N_rgAnd N_azCarrying out consistency check on an uncorrected image of the synthetic aperture radar to be calibrated by using chi-square distribution to eliminate uneven area cells;

calculating a range-wise pixel value N from an uncorrected image of a synthetic aperture radar to be calibrated_rgThe number of values of (c):

whereinCorresponding slope distance R₀The slope of the image power, P_M(R) represents the power of the image at a distance R, where R represents the slant distance of the target from the antenna phase center at that point of the image, ρ_rFor distance resolution,. epsilon_GDesired accuracy for the antenna pattern;

[]represents rounding, N_minIs the minimum number of samples in azimuth, N_razThe number of azimuth stripes and D the original drawing interval.

8. The method for fast in-orbit natural uniform target-based synthetic aperture radar calibration according to claim 1, wherein the fitting of all strip data of the uncorrected image of the synthetic aperture radar to be calibrated after the non-uniform area cells are removed comprises:

averaging each stripe of the uniform area data after the non-uniform area cells are removed, and then normalizing by taking the maximum value as a reference value to be used as the gain G (theta) of the antenna directional diagram under different incidence angles:

20log₁₀G(θ)＝a(θ-θ₀)²+b+c(θ-θ₀)⁴

where θ represents the angle of incidence of the beam at the corresponding pixel point, θ₀Parameters a, b, c are obtained by representing the beam center incident angle.

Technical Field

The invention relates to the field of radar calibration, in particular to an on-orbit rapid calibration method for a synthetic aperture radar based on a natural uniform target.

Background

Cross-scaling is a method of radiometric scaling that scales uncalibrated satellites using calibrated high-precision satellite data. There are many relevant applications of this technology in optical remote sensing. In an existing Synthetic Aperture Radar (SAR) calibration experiment, a corresponding calibration reference device placement area needs to be calculated by combining satellite orbit data, and then a reference device is placed in the corresponding area within a corresponding time period, so that the controllable time of the calibration experiment is short. Meanwhile, in a calibration experiment, if the irradiation time is missed, the experimenter needs to image the reference device placement area again at least after one satellite orbit operation period. Such a strict time limit may result in a calibration experiment that is not time-efficient. Compared with the traditional radiometric calibration method, the cross calibration method directly uses the satellite data which is accurately calibrated to calibrate the un-calibrated satellite. By the aid of the method, the step of placing the reference equipment can be omitted, manpower and material resources required by the experiment are saved, time required by the calibration experiment is saved, and cost required by the calibration experiment is greatly reduced.

The existing on-orbit relative radiometric calibration method which saves manpower and material resources is mainly a calibration method of a distributed target used in 1995 by Hakken island. Fig. 1 is a flow chart of relative radiometric calibration of a synthetic aperture radar based on a distributed reference target in the prior art, and the method mainly includes the following steps:

s11, carrying out cell division on the uncorrected image of the synthetic aperture radar to be calibrated:

calculating a range pixel value N from an image_rgThe number of values of (c):

corresponding slope distance R₀The slope of the image power, where P_M(R) represents the image power at a distance R, which represents the slant distance of the target from the antenna phase center at that point of the image. Rho_rFor distance resolution,. epsilon_GAccuracy is desired for the antenna pattern.

[]Represents rounding, N_minIs the minimum number of samples in azimuth, N_razThe number of azimuth stripes and D the original drawing interval.

S12, carrying out consistency check on the uncorrected image of the synthetic aperture radar to be calibrated to eliminate uneven area cells;

and carrying out consistency check on the image by using chi-square distribution based on the cell parameters to remove the cells in the uneven area.

S13, fitting all the strip data of the uncorrected image of the synthetic aperture radar needing to be calibrated after the uneven area cells are removed;

and averaging each strip of the uniform area data from which the abnormal data are removed, and then normalizing by taking the maximum value as a reference value to be used as the gain G (theta) of the antenna directional diagram under different incidence angles.

Using the model:

20log₁₀G(θ)＝a(θ-θ₀)²+b+c(θ-θ₀)⁴

fitting all the strip data, where θ represents the beam incident angle at the corresponding pixel point, θ₀Parameters a, b, c are obtained by representing the beam center incident angle.

S14, obtaining an absolute scaling coefficient by using the G (theta) antenna directional pattern gain obtained by fitting and the known sigma;

and finally, obtaining an absolute scaling coefficient K by using the antenna directional pattern gain of G (theta) obtained by fitting and the known sigma:

σ is the Radar Cross Section (RCS) value of the manually placed corner reflector.

Disclosure of Invention

Technical problem to be solved

The method mainly solves the problems that RCS reference equipment needs to be manually placed in the existing calibration method, the timeliness of the calibration processing flow is poor, the satellite revisit period is limited, and the synthetic aperture radar cannot be rapidly calibrated in an emergency.

(II) technical scheme

In order to achieve the above object, the present invention provides an on-orbit fast calibration method for synthetic aperture radar based on natural uniform target, comprising:

dividing cells of an uncorrected image of the synthetic aperture radar to be calibrated;

carrying out consistency check on an uncorrected image of the synthetic aperture radar to be calibrated to remove cells in an uneven area;

averaging all azimuth strip data of the uncorrected image after the uneven area cells are removed, then fitting the distance strip data to obtain antenna directional pattern gain G (theta), and performing relative correction on the image by using the antenna directional pattern gain G (theta) to obtain a relative correction image of the synthetic aperture radar to be calibrated;

using the corrected image as a reference image, and performing registration processing on the reference image and the uncorrected image based on the characteristic points in the reference image and the uncorrected image;

and calculating an absolute calibration parameter solution according to the registered reference image and the relative correction image of the synthetic aperture radar to be calibrated, and solving an absolute calibration coefficient gain and an offset coefficient offset.

The registration processing is performed on the reference image and the uncorrected image, and comprises the following steps:

selecting ground control points on the reference image and the uncorrected image to register the two images;

and performing scaling processing on the reference image to enable the azimuth resolution and the distance resolution of the reference image to be consistent with those of the original image.

Wherein, calculating an absolute calibration parameter solution according to the reference image after the registration processing and the relative correction image of the synthetic aperture radar to be calibrated, and obtaining an absolute calibration coefficient gain and an offset coefficient offset, comprising:

taking the DN value of the processed reference image as the true value of the backscattering coefficient of the target area of the synthetic aperture radar to be calibrated relative to the corrected image, the relationship between the pixel values of the two images at coordinates (i, j) can be expressed as:

DN_1i，j＝gain·DN_2i，j+offset

wherein DN_1i，jFor the DN value at the ith row and jth column unit cell of the reference image after the registration processing, DN_2i，jThe method comprises the steps of obtaining a DN value of an image after relative correction at the ith row and jth column cells in an uncorrected image of the synthetic aperture radar to be calibrated;

order to

Where N represents the number of all pixels in the reference image scaling target area, the above equation can be written as:

DN₁＝DN₂·G

the absolute scaling coefficient gain and the offset coefficient offset are solved using the least square method:

G＝(DN₂ ^T·DN₂)·DN₂ ^T·DN₁

(III) advantageous effects

1. According to the synthetic aperture radar on-orbit rapid calibration method based on the natural uniform target, the distributed uniform target (such as Amazon rainforest, uniform grassland, desert and the like) is used for carrying out relative correction processing, the process of placing manual reference equipment is omitted, manpower and material resources consumed in the experimental process can be reduced, and the external calibration experimental process is accelerated.

2. The synthetic aperture radar on-orbit rapid calibration method based on the natural uniform target uses the corrected image of the calibrated spaceborne (or airborne) load which meets certain requirements and has the same pitch angle and the same azimuth angle as a reference datum for calculating the absolute calibration parameters of the load to be calibrated, thereby omitting the process of placing manual reference equipment and accelerating the external calibration experiment process.

3. According to the synthetic aperture radar on-orbit quick calibration method based on the natural uniform target, provided by the invention, the geometric relation and the resolution of the reference image are converted to be consistent with the image to be calibrated, the two images are subjected to translational registration, the numerical values of the two images can be directly calculated and compared, and the accuracy of on-orbit quick calibration is ensured.

Drawings

FIG. 1 is a prior art relative radiometric calibration flow chart for a synthetic aperture radar based on a distributed reference target;

FIG. 2 is a flowchart of a method for fast on-orbit calibration of a synthetic aperture radar based on natural uniform targets according to an embodiment of the present invention;

fig. 3 is a flowchart of a registration process of a reference image and an uncorrected image in an in-orbit fast calibration method of a synthetic aperture radar based on a natural uniform target according to an embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to specific embodiments and the accompanying drawings.

Fig. 2 is a flowchart of a method for fast in-orbit calibration of a synthetic aperture radar based on a natural uniform target according to an embodiment of the present invention, where the method includes the following steps:

s21, dividing cells of the uncorrected image of the synthetic aperture radar to be calibrated;

s22, carrying out consistency check on the uncorrected image of the synthetic aperture radar to be calibrated to eliminate uneven area cells;

based on the parameter N_rgAnd N_azThe uncorrected image is subjected to consistency check using chi-square distribution to eliminate uneven area cells. Calculating a distance-wise pixel value N from an uncorrected image_rgThe number of values of (c):

whereinCorresponding slope distance R₀The slope of the image power, P_M(R) represents the image power at a distance R, which represents the slant distance of the target from the antenna phase center at that point of the image. Rho_rFor distance resolution,. epsilon_GAccuracy is desired for the antenna pattern.

[]Represents rounding, N_minIs the minimum number of samples in azimuth, N_razThe number of azimuth stripes and D the original drawing interval.

S23, averaging all azimuth strip data of the uncorrected image without the uneven area cells, fitting the distance strip data to obtain antenna directional pattern gain G (theta), and performing relative correction on the image by using the antenna directional pattern gain G (theta) to obtain a relative correction image of the synthetic aperture radar to be calibrated;

and averaging each stripe of the uniform area data after the non-uniform area cells are removed, and then normalizing by taking the maximum value as a reference value to be used as the gain G (theta) of the antenna directional diagram under different incidence angles.

2Olog₁₀G(θ)＝a(θ-θ₀)²+b+c(θ-θ₀)⁴

Where θ represents the angle of incidence of the beam at the corresponding pixel point, θ₀Parameters a, b, c are obtained by representing the beam center incident angle.

Then, the gain G (theta) of the antenna directional diagram obtained by fitting is used for carrying out relative correction on the uncorrected image to obtain a relative correction image of the synthetic aperture radar to be calibrated, and the correction formula is as follows:

DN_{relativecalibrated}is a relative corrected DN value, DN_uncalibratedIs an uncorrected picture DN value.

S21-S23 of the synthetic aperture radar on-orbit rapid calibration method based on the natural uniform target is consistent with S11-S13 of a calibration method flow of a distributed target mentioned in the background technology, but after relative correction, an RCS theoretical value is obtained by manually placing a corner reflector to solve absolute calibration parameters, and the absolute calibration parameters are solved by using a registration image which is subjected to radiation correction. The invention uses the registration image which is subjected to radiation correction to calculate the absolute calibration parameter, and the registration processing needs to be carried out on the two images, the specific process can be shown as fig. 3, fig. 3 is a flow chart of the registration processing to the reference image and the uncorrected image in the synthetic aperture radar on-orbit rapid calibration method based on the natural uniform target according to the embodiment of the invention, and the registration processing to the reference image and the uncorrected image specifically comprises the following steps:

and S24, using the corrected images of the calibrated satellite-borne (or airborne) loads at the same time, the same pitch angle and the same azimuth angle as reference images, and carrying out registration processing on the reference images and the uncorrected images based on the characteristic points in the local images.

Firstly, the resolution of the reference image is selected to ensure that the difference between the resolution of the reference image and the image to be calibrated is within 5 times, namely the resolution of the reference image is between 0.2 times and 5 times of the resolution of the image to be calibrated. And (4) intercepting ground features of the reference image and the uncorrected image, uniformly distributing, and correcting partial images with flat terrain (such as amazon forest, plain area and the like). The total number of contrastable cells in the truncated region should be guaranteed to be above 50. Meanwhile, the incident elevation angles of the two aperture radar images at the central line of the wave beam are required to be the same and between 20 degrees and 80 degrees, and the error between the two is not less than 1 degree. The azimuth angles are the same, the error is not less than 1 degree, and the irradiation time of the two images is ensured to be different by less than 1 day so as to prevent time decorrelation.

The registration processing of the reference image and the uncorrected image is divided into two steps:

s241, selecting ground control points on the reference image and the uncorrected image to register the two images;

and selecting n obvious characteristic points such as corners, edge points and the like on the images of the reference image and the uncorrected image as ground control points. Setting the coordinate of a cell on a reference image by n ground control points of the reference image as (x)₂₁，y₂₁)，(x₂₂，y₂₂)，……，(x_2n，y_2n) In the uncorrected image, the reference image control point corresponds to the cell coordinate (x) on the image to be calibrated₁₁，y₁₁)，(x₁₂，y₁₂)，……，(x_1n，y_1n)：

The coordinate position relationship of the two image pixels can be expressed as follows:

order toThe above equation can be written as:

XY₀＝XY·AB

and (3) solving a registration coefficient by using n ground control points selected in the front and using a least square method:

AB＝(XY^T·XY)^-1·XY^T·XY₀

registering all coordinate points of the reference image to the uncorrected image using registration coefficients:

XY_0all＝XY_all·AB

the coordinate positions of all pixel points before the reference image is not registered,and (4) registering the reference image to the uncorrected image, wherein N represents the number of all pixels in the calibration target area of the reference image.

S242, scaling the reference image to enable the azimuth resolution and the distance resolution of the reference image to be consistent with those of the original image;

and performing scaling processing on the reference image by using a bilinear interpolation algorithm or a cubic convolution interpolation algorithm to enable the azimuth resolution and the distance resolution of the reference image to be consistent with the original image.

Azimuth scaling factor D_a＝ρ_a/ρ_arefDistance scaling factor D_r＝ρ_r/ρ_rrefWhere ρ is_aFor the azimuth resolution, rho, of the synthetic aperture radar image to be corrected_arefFor azimuthal resolution of the reference image, p_rFor the azimuthal resolution, rho, of the SAR image to be corrected_rrefIs the reference image azimuthal resolution.

S25, calculating an absolute calibration parameter solution according to the reference image after registration processing and the synthetic aperture radar relative correction image to be calibrated, and solving an absolute calibration coefficient gain and an offset coefficient offset;

after the two steps of registration and scaling interpolation are performed, taking the DN value (RCS value) of the processed reference image 1 as the true value of the backscattering coefficient of the image target area of the relative correction image (image 2), and considering the relationship between the pixel values of the two images at coordinates (i, j) separately, it can be expressed as:

DN_1i，j＝gain·DN_2i，j+offset

DN_1i，jis the DN value (RCS value) at the ith row and jth column cell of image 1. DN_2i，jThe pixel value is the DN value of the image after relative correction at the ith row and jth column unit cell in the image 2.

Order toN represents the number of all pixels of the reference image scaling target area, the above equation can be written as:

DN₁＝DN₂·G

the absolute scaling coefficient gain and the offset coefficient offset are solved using the least square method:

G＝(DN₂ ^T·DN₂)^-1·DN₂ ^T·DN₁

the above-mentioned embodiments are intended to illustrate the objects, technical solutions and advantages of the present invention in further detail, and it should be understood that the above-mentioned embodiments are only exemplary embodiments of the present invention, and are not intended to limit the present invention, and any modifications, equivalents, improvements and the like made within the spirit and principle of the present invention should be included in the protection scope of the present invention.