Real-time splicing method of SAR image based on frequency domain correlation and spatial positioning combination

Abstract

The present application relates to a kind of SAR image splicing methods based on frequency domain correlation and space positioning joint, first extract the certain distance of the previous image near right edge as reference, calculate the correlation of each distance of the next image and reference data, get splicing parameter by correlation;Then through the airplane position, attitude and radar pointing parameters when two SAR images are imaged, the positioning data of image is calculated, and the image splicing parameter is calculated using image positioning data;Finally, the difference of splicing parameter calculated in the above two steps is compared, if the difference is greater than a certain threshold, the splicing parameter obtained by positioning data is used, otherwise the splicing parameter obtained by image correlation is used.The problem that airborne SAR radar cannot be spliced correctly in real time after imaging on the scene of uniform distribution is solved.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of radar image processing, and particularly relates to a SAR image real-time splicing method based on joint frequency domain correlation and spatial positioning. BACKGROUND

[0002] When the airborne SAR radar works in a strip mode, high-resolution SAR images in the strip mode are generally displayed in real time on a display terminal through screen scrolling, and a plurality of continuous images will be displayed on the screen at this time, and the instability of the aircraft flight will cause the mutual misalignment between every two continuous images, the misalignment includes the overlap in the azimuth direction and the translation in the range direction, that is, the image splicing parameters, and therefore the plurality of continuous images need to be spliced and displayed in real time. Under the requirement of real-time display, the radar needs to calculate the image splicing parameters for the display software after imaging processing, and the calculation generally adopts the method of calculating the splicing parameters based on the image content correlation similar to optical images. However, the above-mentioned method of calculating the image correlation has a certain applicable range, and can only correctly calculate the SAR image splicing parameters in a non-uniform scene or a case with a large number of landmarks, and is no longer applicable to a uniform distribution scene such as a desert, a grassland, and a gobi. The scene does not have obvious electromagnetic wave reflectors, and the correlation between the images is weak, and therefore the correlation processing based on the image content cannot be performed. SUMMARY

[0003] Technical problems to be solved

[0004] In view of the problem that the airborne SAR radar cannot correctly splice the images in real time after imaging in a uniform distribution scene, the application provides a SAR image real-time splicing method based on joint frequency domain correlation and spatial positioning, which can effectively improve the real-time splicing accuracy of the SAR image in a uniform scene under the premise of increasing a small amount of operation.

[0005] Technical scheme

[0006] A SAR image real-time splicing method based on joint frequency domain correlation and spatial positioning, characterized by the following steps:

[0007] Step 1: correlation in the frequency domain is calculated for the images to be spliced, and the splicing parameters are obtained from the correlation: a certain distance column near the right edge of a previous image is transformed into the frequency domain as a reference, each distance column of a subsequent image is selected in turn, transformed into the frequency domain, and multiplied with the reference column, and the azimuth overlap and the range offset are calculated from the multiplication result;

[0008] Step 2: calculate the positioning data of the images to be spliced, and obtain the splicing parameters from the positioning data: according to the airborne SAR strip imaging geometry model, calculate the geographic coordinates of the centers of the two images according to the states of the carrier and the radar, and calculate the relative position relationship of the two images from the geographic coordinates of the centers of the two images, the number of pixel points and the interval between the pixel points, so as to obtain the amount of azimuth overlap and the distance offset;

[0009] Step 3: calculate the difference value of the corresponding splicing parameters of step 1 and step 2, and select the splicing parameters of step 1 or step 2 according to the comparison result of the length represented by the difference value and the positioning error statistical value of the radar.

[0010] Further technical solutions of the application: step 1 is as follows:

[0011] 1a) select a longitudinal data m pixels away from the right edge of the previous image as a reference, denoted as a sequence S1, and perform fast Fourier transform to obtain frequency domain data S 1f :

[0012] S 1f =FFT(S1) (1)

[0013] 1b) sequentially select each longitudinal data in the next image, denoted as a sequence S2(n), 1≤n≤N, N is the number of horizontal pixels, and perform fast Fourier transform and conjugate to obtain frequency domain data S 2f (n):

[0014] S 2f (n)=Conj(FFT(S2(n))) (2)

[0015] 1c) calculate S 1f ·S 2f (n), and perform inverse fast Fourier transform to obtain a time domain sequence S3(n):

[0016] S3(n)=IFFT(S 1f ·S 2f (n)) (3)

[0017] 1d) for each sequence S3(n), calculate the maximum value and the index number of the maximum value in the sequence, denoted as a maximum value sequence Max(N) and an index number sequence Index(N);

[0018] 1e) calculate the maximum value a in the sequence Max(N) and the index number b of the maximum value in the sequence;

[0019] 1f) the azimuth overlap of the previous image and the next image is b+m;

[0020] 1g) If Index(b) < M / 2, the distance offset of the latter image relative to the former image is -Index(b), otherwise the distance offset is M-Index(b), the distance offset is positive value representing the latter image offset upward relative to the former image, and the negative value representing the latter image offset downward relative to the former image.

[0021] Further technical solutions of the present application: step 2 is specifically as follows:

[0022] 2a) Calculate the squint angle θ:

[0023]

[0024] In the formula, fdc is the Doppler center of the current image, λ is the radar wavelength, v A is the average velocity in the azimuth direction of the current image;

[0025] 2b) Calculate the projection of the squint angle on the ground θ g :

[0026]

[0027] Where, r c is the slant range of the scene center point, and h is the height of the aircraft;

[0028] 2c) Calculate the coordinates of the image center point Q in the OXYZ three-dimensional coordinate system, wherein OXYZ takes the position O of the aircraft located at the center of the synthetic aperture as the origin, the ideal heading of the aircraft as the Y axis direction, the X axis perpendicular to the ideal heading of the aircraft along the imaging scene direction, and the Z axis direction vertically upward, to establish a three-dimensional space rectangular coordinate system;

[0029]

[0030] 2d) On the basis of the three-dimensional coordinates (x Q , y Q , z Q ) of the image center Q, the geographical coordinates (L O , M O , h O ) of the O point, and the ideal heading γ of the aircraft, the geographical coordinates (L Q , M Q , h Q ) of the image center Q are obtained according to the coordinate conversion matrix, wherein in the geographical coordinates, L is the longitude, M is the latitude, and h is the altitude;

[0031] 2e) Set the center of the previous image as Q1, the center of the next image as Q2, A as the intersection of the ideal heading through Q1 and the distance direction through Q2, the geographic coordinates of Q1 and Q2 can be obtained according to the above steps, and then the length r and the direction of the vector Q1 Q2 are calculated , that is, the angle with the north direction;

[0032] 2f) The azimuth overlap amount of the previous image and the next image is where Δ A is the azimuth pixel interval of the image;

[0033] 2g) If it is right side view imaging, the distance offset amount of the next image relative to the previous image is If it is left side view imaging, the distance offset amount of the next image relative to the previous image is where Δ R is the distance pixel interval of the image.

[0034] Further technical solutions of the present application: step 3 is specifically as follows:

[0035] Calculate the difference value of the corresponding stitching parameters of step 1 and step 2, and take the positioning error statistical value of this radar as a comparison threshold. As long as any difference value represented by the length of the azimuth overlap amount and the distance offset amount is greater than the above threshold, it is considered that the current scene is a uniform scene, the frequency domain correlation method is invalid, and the stitching parameters obtained by the spatial positioning method need to be used. Otherwise, the stitching parameters obtained by the frequency domain correlation method are used.

[0036] A computer system, characterized in that it comprises: one or more processors, a computer readable storage medium for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors realize the above-mentioned method.

[0037] A computer readable storage medium, characterized in that it stores computer executable instructions, which when executed are used to realize the above-mentioned method.

[0038] Advantages

[0039] The present application provides a kind of SAR image real-time splicing method based on frequency domain correlation and space positioning joint, compared with pure calculation image correlation, when splicing SAR image using the method of the present application, it has the following beneficial effects:

[0040] Because for common non-uniform scene, the correlation between images is strong, the stitching parameters calculated by correlation are smaller than the stitching parameters calculated by positioning data, while for uniform scene, the correlation between images is weak, the stitching parameters calculated by correlation are larger than the stitching parameters calculated by positioning data, so this method not only ensures the accurate stitching for non-uniform scene, but also enables the stitching of uniform scene to be within acceptable error range under the premise of increasing a small amount of operation. BRIEF DESCRIPTION OF DRAWINGS

[0041] The accompanying drawings are included to provide a further understanding of the application, and are incorporated in and constitute a part of this specification. The drawings illustrate embodiments of the application and, together with the description, serve to explain the principles of the application. In the drawings:

[0042] Figure 1 For two continuous images to be stitched under uniform scene, wherein Figure 1 -a is the previous image, Figure 1 -b is the next image.

[0043] Figure 2 is the stitched image obtained by the frequency domain correlation method.

[0044] Figure 3 is the airborne SAR strip imaging geometry model.

[0045] Figure 4 is the schematic diagram of the spatial positioning method.

[0046] Figure 5 is the stitched image obtained by the joint method of frequency domain correlation and spatial positioning.

[0047] Figure 6 is the processing flowchart of the joint method of frequency domain correlation and spatial positioning. DETAILED DESCRIPTION

[0048] In order to make the objects, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not to limit the present application. In addition, the technical features involved in each embodiment of the present application described below can be combined with each other as long as there is no conflict.

[0049] The application provides a SAR image splicing method based on frequency domain correlation and spatial positioning combination. First, a certain distance data close to the right edge of a previous image is extracted as a reference, the correlation of each distance data of a next image with the reference data is calculated, and the splicing parameters are obtained from the correlation; then, the positioning data of the images are calculated through the parameters such as the airplane position, attitude and radar pointing during imaging of the two SAR images, and the image positioning data are used to calculate the image splicing parameters; finally, the difference of the splicing parameters calculated in the above two steps is compared, if the difference is greater than a certain threshold, the splicing parameters obtained from the positioning data are used, otherwise the splicing parameters obtained from the image correlation are used.

[0050] The method of the application comprises the following steps:

[0051] Step 1: the correlation of the images to be spliced is calculated in the frequency domain, and the splicing parameters are obtained from the correlation: a certain distance data close to the right edge of a previous image is selected and transformed into the frequency domain as a reference, each distance data of a next image is selected in turn, transformed into the frequency domain and multiplied with the reference data, and the azimuth overlap and the distance offset are calculated from the multiplication result;

[0052] Step 2: the positioning data of the images to be spliced are calculated, and the splicing parameters are obtained from the positioning data: according to the airborne SAR strip imaging geometric model, the geographical coordinates of the centers of the two images are calculated according to the states of the carrier and the radar, the relative position relationship of the two images can be calculated from the geographical coordinates of the centers of the two images, the number of pixel points and the pixel point interval, and thus the azimuth overlap and the distance offset are obtained;

[0053] Step 3: the difference of the corresponding splicing parameters of step 1 and step 2 is calculated, and the splicing parameters of step 1 or step 2 are selected according to the comparison result of the length represented by the difference with the positioning error statistical value of the radar.

[0054] The embodiment of the application is further described below.

[0055] Step 1: the correlation of the images to be spliced is calculated in the frequency domain, and the splicing parameters are obtained from the correlation.

[0056] According to the imaging processing steps of a general SAR image, the longitudinal direction of the image to be displayed is the distance direction, and the transverse direction is the azimuth direction, the number of longitudinal pixel points is M, and the number of transverse pixel points is N.

[0057] a) selecting Figure 1 a longitudinal data of m pixel points away from the right edge as a reference, denoted as S1, and performing fast Fourier transform to obtain the frequency domain data S 1f .

[0058] S 1f = FFT (S1) (1)

[0059] b) sequentially selecting Figure 1 -b each longitudinal data, denoted as a sequence S2(n) (1≤n≤N), is subjected to fast Fourier transform and conjugation to obtain frequency domain data S 2f (n).

[0060] S 2f (n) = Conj(FFT(S2(n))) (2)

[0061] c) calculating S 1f ·S 2f (n), and subjected to inverse fast Fourier transform to obtain a time domain sequence S3(n).

[0062] S3(n) = IFFT(S 1f ·S 2f (n)) (3)

[0063] d) for each sequence S3(n), calculating its maximum value and the index number of the maximum value in the sequence, denoted as a maximum value sequence Max(N) and an index number sequence Index(N).

[0064] e) calculating the maximum value a in the sequence Max(N) and the index number b of the maximum value in the sequence.

[0065] f) Figure 1 -a and Figure 1 -b have an orientation overlap amount of b+m.

[0066] g) if Index(b) < M / 2, then Figure 1 -b is offset from Figure 1 -a by an amount of -Index(b), otherwise, the distance offset amount is M-Index(b) (the distance offset amount is a positive value representing that Figure 1 -b is offset upward from Figure 1 -a, and is a negative value representing that Figure 1 -b is offset downward from Figure 1 -a).

[0067] As shown in Figure 2 , the stitching parameters obtained using the frequency domain correlation method are used to stitch two continuous images shown in Figure 1 , and the stitching result has a large error from the actual scene.

[0068] Step 2: calculating the positioning data of the images to be stitched, and obtaining the stitching parameters from the positioning data.

[0069] Figure 3In the airborne SAR strip imaging geometric model shown, point Q is the center point of the imaging scene. The origin is set at position O when the aircraft is at the center of the synthetic aperture. The ideal heading of the aircraft is the Y-axis, the X-axis is perpendicular to the ideal heading of the aircraft and along the direction of the imaging scene, and the Z-axis is perpendicular to the ground and upwards. A three-dimensional Cartesian coordinate system is established. c h is the slant distance from the center point of the scene, h is the height of the aircraft, and θ is the oblique angle. g This is the projection of the oblique angle onto the ground.

[0070] a) Calculate the oblique angle θ:

[0071]

[0072] In the formula, fdc is the Doppler center of the current image, λ is the radar wavelength, and v A This represents the average velocity in the azimuth direction of the current image.

[0073] b) Calculate the projection θ of the oblique angle onto the ground. g :

[0074]

[0075] c) Calculate the coordinates of the image center Q in the OXYZ three-dimensional coordinate system:

[0076]

[0077] d) The three-dimensional coordinates (x, y) at the image center Q Q ,y Q ,z Q ), geographic coordinates of point O (L O M O ,h O Based on the ideal heading γ of the aircraft and the coordinate transformation matrix, the geographic coordinates (L) of the image center Q can be obtained. Q M Q ,h Q (In geographic coordinates, L represents longitude, M represents latitude, and h represents altitude.)

[0078] e) such as Figure 4 As shown, let Figure 1 The center of -a is Q1. Figure 1 The center of -b is Q2, and A is the intersection of the ideal heading through Q1 and the distance direction through Q2. Following the steps above, the geographical coordinates of Q1 and Q2 can be obtained, and then the length r and direction of vector Q1Q2 can be calculated. (Angle with north)

[0079] f) Figure 1 -a and Figure 1 -b's azimuth overlap is where Δ A is the azimuth pixel interval of the image, and N is the number of horizontal pixels of the image.

[0080] g) If right side view imaging, then Figure 1 -b relative to Figure 1 -a distance offset is If left side view imaging, then Figure 1 -b relative to Figure 1 -a distance offset is where Δ R is the range pixel interval of the image.

[0081] Step 3: Calculate the difference of the corresponding stitching parameters of step 1 and step 2, and take the positioning error statistical value of this radar as a comparison threshold. If the length represented by any one of the difference of the azimuth overlap and the distance offset is above the threshold, it is considered that the current scene is a uniform scene, the frequency domain correlation method is invalid, and the stitching parameters obtained by the spatial positioning method need to be used. Otherwise, the stitching parameters obtained by the frequency domain correlation method are used.

[0082] As shown in Figure 5 , the stitching parameters obtained by the frequency domain correlation and spatial positioning combined method are used to stitch the two continuous images shown in Figure 1 , and the stitching result is basically consistent with the actual scene.

[0083] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of various equivalent modifications or replacements within the technical range disclosed by the present application, and these modifications or replacements should be covered within the protection scope of the present application.

Claims

1. A real-time splicing method for SAR images based on joint frequency domain correlation and spatial positioning, characterized in that The steps are as follows: Step 1: correlation of the images to be spliced is calculated in the frequency domain, and the splicing parameters are obtained from the correlation: a certain distance vector series of the edge of the previous image is selected and transformed into the frequency domain as a reference, and each distance vector series of the subsequent image is selected and transformed into the frequency domain, and the reference series is multiplied point by point, and the azimuth overlap and the distance offset are calculated from the multiplication result; Step 2: the positioning data of the images to be spliced is calculated, and the splicing parameters are obtained from the positioning data: according to the airborne SAR strip imaging geometric model, the geographical coordinates of the centers of the two images are calculated according to the states of the aircraft and the radar, and the relative position relationship of the two images is calculated from the geographical coordinates of the centers of the two images, the number of pixel points and the pixel point interval, so as to obtain the azimuth overlap and the distance offset; Step 3: the difference of the corresponding splicing parameters of step 1 and step 2 is calculated, and the length represented by the difference is compared with the positioning error statistical value of the radar to select the splicing parameters of step 1 or step 2. 2.The SAR image real-time splicing method based on frequency domain correlation and spatial positioning combination of claim 1, wherein: Step 1 is as follows: 1a) Select a vertical data at m pixels from the right edge of the previous image as reference, denoted as S1, and take its fast Fourier transform to get the frequency domain data S 1f : S 1f = FFT(S1) (1) 1b) sequentially select each longitudinal data in the latter image, denoted as a sequence S2(n), 1≤n≤N, N is the number of horizontal pixels, and perform fast Fourier transform and get the conjugate to obtain the frequency domain data S 2f (n): S 2f (n) = Conj(FFT(S2(n))) (2) 1c) Calculate S 1f • S 2f (n), and take the inverse Fourier transform to obtain the time-domain sequence S3(n): S3(n) = IFFT(S 1f • S 2f (n)) (3) 1d) For each series S3(n), the maximum value and the index number of the maximum value in the series are calculated, and the maximum value series is denoted as Max(N) and the index number series is denoted as Index(N); 1e) The maximum value a in the series Max(N) and the index number b of the maximum value in the series are calculated; 1f) The azimuth overlap of the previous image and the subsequent image is b+m; 1g) If Index(b) < M / 2, the distance offset of the subsequent image relative to the previous image is -Index(b), otherwise the distance offset is M-Index(b), and the distance offset is positive, which means that the subsequent image is offset upward relative to the previous image, and the distance offset is negative, which means that the subsequent image is offset downward relative to the previous image. 3.The SAR image real-time splicing method based on frequency domain correlation and spatial positioning combination of claim 1, characterized in that: Step 2 is as follows: 2a) Calculate the squint angle θ: where fdcis the Doppler center of the current image, λ is the radar wavelength, v A is the azimuth mean velocity of the current image; 2b) calculating the projection θ of the squint angle on the ground g : where r c is the slant range of the scene center point, and h is the height of the carrier. 2c) Calculate the coordinates of the image center point Q in the OXYZ three-dimensional coordinate system, wherein OXYZ takes the position O of the aircraft when the synthetic aperture center is located as the origin, the ideal heading of the aircraft as the Y-axis direction, the X-axis perpendicular to the ideal heading of the aircraft along the imaging scene direction, and the Z-axis direction perpendicular to the ground upward, and a three-dimensional space rectangular coordinate system is established; 2d) on the basis of the three-dimensional coordinates (x Q ,y Q ,z Q ) of the image center Q, the geographic coordinates (L O ,M O ,h O ) of the point O and the ideal heading γ of the carrier, the geographic coordinates (L Q ,M Q ,h Q ) of the image center Q are obtained according to the coordinate conversion matrix, wherein L is the longitude, M is the latitude and h is the altitude; 2e) Set the center of the previous image as Q1, the center of the next image as Q2, A as the intersection of the ideal heading through Q1 and the distance direction through Q2, obtain the geographic coordinates of Q1 and Q2 according to the above steps, and further calculate the length r and the direction of the vector Q1 Q2 i.e. the angle with the north direction; 2f) the amount of orientation overlap between the previous image and the next image is where Δ A is the orientation pixel interval of the image; 2g) if right side view imaging, the distance offset of the latter image relative to the former image is if left side view imaging, the distance offset of the latter image relative to the former image is where Δ R is the distance pixel interval of the image.

4. The SAR image real-time splicing method based on frequency domain correlation and spatial positioning combination of claim 1, characterized in that: Step 3 is as follows: The difference of the corresponding splicing parameters of step 1 and step 2 is calculated, and the positioning error statistical value of the radar is taken as a comparison threshold, and only when any difference value represented by the length of the azimuth overlap and the distance offset is less than the threshold value, it is considered that the current scene is a uniform scene, the frequency domain correlation method is invalid, and the splicing parameters obtained by the spatial positioning method need to be used, otherwise the splicing parameters obtained by the frequency domain correlation method are used.

5. A computer system, characterized by It comprises: One or more processors, a computer readable storage medium for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method of claim 1.

6. A computer-readable storage medium, characterized in that Computer executable instructions are stored, and the instructions are used to implement the method of claim 1 when executed.

Images