A method for designing massive SAR data imaging software based on fast BP algorithm

Abstract

The application relates to a kind of massive SAR data imaging software design methods based on fast BP algorithm, for the existing fast BP imaging algorithm, by accurately estimating the data amount in the operation process, propose to use the available memory of operation server, dynamically correct J operator, carry out sub-scene processing to sub-set, use disk temporary storage intermediate operation result, solve the memory limit problem of SAR full scene full resolution imaging of super large data on the operation server with lower memory, greatly reduce the hardware memory cost of operation server.

Description

Technical Field

[0001] This application relates to the field of SAR imaging, and more specifically, to a software design method for imaging massive SAR data based on the fast BP algorithm. Background Technology

[0002] To meet the imaging requirements of high-orbit (geosynchronous orbit) SAR payloads achieve better target azimuth resolution by obtaining a longer coherent accumulation time in the target azimuth direction through azimuth antenna beam-staring scanning. Range resolution is determined by system bandwidth; a larger bandwidth results in higher resolution, but also increases the computational load. To achieve good azimuth resolution, the synthetic aperture time (SAP) for a single high-orbit SAR strip mode can reach 1800s–3600s, leading to a massive amount of imaging echo data within that time. High-orbit SAR satellite strip mode can achieve imaging over a 300km × 500km area in a single startup, exhibiting an ultra-wide swath width; therefore, full-scene, full-resolution image data requires a significant amount of memory. To achieve stable full-scene imaging on limited server memory, the imaging process must address the issues of segmenting the echo and image scenes for separate processing, defining the memory requirements for each SAR imaging step, and designing a suitable software flow. Current technologies face memory limitations when performing full-scene, full-resolution SAR imaging on servers with limited memory for massive amounts of data. Summary of the Invention

[0003] To overcome at least one deficiency in the prior art, this application provides a software design method for imaging massive SAR data based on the fast BP algorithm.

[0004] Firstly, a method for designing massive SAR data imaging software based on the fast backpropagation (BP) algorithm is provided, including:

[0005] Step 1: Divide the full-scene single-view single-precision complex image into C r ×C a Individual scene images, C r C represents the number of sub-scene images at a distance. a This represents the number of sub-scene images in the orientation direction, and all sub-scene images are divided into J subsets, with J initially set to 1, including:

[0006] Option 1: Divide all sub-scene images into J subsets along the orientation or distance direction;

[0007] Option 2: Divide all sub-scene images into J subsets;

[0008] There is an overlapping area between two adjacent sub-scene images;

[0009] Step 2: Under the current J value, calculate the memory space occupied by all sub-scene images based on the width of the sub-scene images, calculate the memory space occupied by the overlapping area of ​​all sub-scene images, and calculate the memory space occupied by the 3D geographic grid points on which fast BP imaging depends.

[0010] Step 3: When performing SAR imaging based on the fast BP algorithm, the echo signal needs to be divided into sub-apertures to obtain multiple sub-aperture signals; the memory space occupied by the echo data of the sub-aperture is calculated based on the azimuth pulse number and the range sampling point number of the sub-aperture.

[0011] Step 4: Calculate the peak memory consumption during SAR imaging based on the memory space occupied by all sub-scene images, the memory space occupied by the overlapping area of ​​all sub-scene images, the memory space occupied by the 3D geographic grid points on which fast BP imaging depends, and the memory space occupied by the echo data of sub-apertures.

[0012] Step 5: Determine whether the peak memory consumption during the SAR imaging process is greater than the available memory of the computing server. If yes, increase the J value by 1 and return to step 2. If no, proceed to step 6.

[0013] Step 6: For Scheme 1, determine whether the J value is less than or equal to the number C of azimuth sub-scene images. a Or the distance to the number of sub-scene images C r For Scheme 2, determine whether the value of J is less than or equal to the number of sub-scene images C. r ×C a If so, the computing server's memory is sufficient to achieve full-scene imaging and obtain the final J value; otherwise, the computing server's memory is insufficient to achieve full-scene imaging.

[0014] In one embodiment, the full-scene single-view single-precision complex image is divided into multiple sub-scene images, and the division rules include:

[0015] Each sub-scene image has a consistent two-dimensional width that is greater than a first set value, the single-view 16-bit complex image corresponding to each sub-scene image is smaller than a second set value, and there is an overlapping area between two adjacent sub-scene images.

[0016] In one embodiment, the memory space occupied by all sub-scene images is represented by the following formula:

[0017]

[0018] Among them, M fig Let N be the memory space occupied by all sub-scene images, where j is the subset index, J is the number of subsets, and N is the number of subsets. j Let W be the number of sub-scene images in the j-th subset, and n be the index of the sub-scene image in the j-th subset.n I is the actual width of the nth sub-scene image. r I represents the distance pixel interval of the image. a The pixel spacing in the orientation direction of the image.

[0019] In one embodiment, the memory space occupied by the overlapping area of ​​all sub-scene images is represented by the following formula:

[0020]

[0021] Among them, M ovp Let N be the memory space occupied by the overlapping regions of all sub-scene images, where j is the index of the subset, J is the number of subsets, and N is the number of subsets. j Let W be the number of sub-scene images in the j-th subset, and n be the index of the sub-scene image in the j-th subset. n I is the actual width of the nth sub-scene image. r I represents the distance pixel interval of the image. a α represents the pixel spacing in the orientation direction of the image, and α is the proportion of the overlapping region to the sub-scene image.

[0022] In one embodiment, the memory space occupied by the 3D geographic grid points upon which fast BP imaging relies is expressed by the following formula:

[0023]

[0024] Among them, M tar M represents the memory space occupied by the 3D geographic grid points that fast BP imaging relies on. ovp M represents the memory space occupied by the overlapping regions of all sub-scene images. fig D represents the memory space occupied by all sub-scene images. n The sampling factor for geographic grid point coordinates along the azimuth gradient.

[0025] In one embodiment, the memory space occupied by the echo data of the sub-aperture is expressed by the following formula:

[0026] M apt =N k ×N r ×8

[0027]

[0028] Among them, M apt N represents the memory space occupied by the echo data of the sub-aperture. k N represents the azimuth pulse number of the sub-aperture. r N represents the number of echo range sampling points. a T is the number of pulses in the azimuth direction over the full aperture time. aFor the full aperture time, T k For a single sub-aperture time, To round down, This is for rounding up.

[0029] In one embodiment, calculating the peak memory consumption during SAR imaging includes:

[0030] For Scheme 1, the SAR imaging process includes: for each subset, performing sub-aperture imaging, sub-aperture upsampling fusion, and first-level stitching of the sub-scene images in sequence to obtain the stitched image corresponding to the subset, and writing it to the disk; after each subset has completed image stitching, reading the stitched image corresponding to the subset from the disk in sequence according to the subset order, and performing second-level stitching in sequence to achieve full-scene imaging;

[0031] Calculate the memory required for sub-aperture imaging of each subset:

[0032]

[0033] Among them, (M) img ) j M represents the memory required for subaperture imaging of the j-th subset. apt N represents the memory space occupied by the echo data of the sub-aperture. j Let W be the number of sub-scene images in the j-th subset, and n be the index of the sub-scene image in the j-th subset. n I is the actual width of the nth sub-scene image. r I represents the distance pixel interval of the image. a M represents the azimuth pixel spacing of the image. tar M represents the memory space occupied by the 3D geographic grid points that fast BP imaging relies on. aux Memory space occupied by auxiliary data;

[0034] Calculate the memory required for full-scene stitching:

[0035]

[0036] Among them, M spl The memory required for full-scene stitching, M fig M represents the memory space occupied by all sub-scene images. ovp The memory space occupied by the overlapping area of ​​all sub-scene images;

[0037] The peak memory consumption during SAR imaging is:

[0038] M max =max(max((M) img ) j ),Mspl )

[0039] Among them, M max This represents peak memory consumption.

[0040] In one embodiment, calculating the peak memory consumption during SAR imaging includes:

[0041] For Scheme 2, the SAR imaging process includes: dividing all sub-scene images into J subsets; for each subset, performing sub-aperture imaging and sub-aperture upsampling fusion on the sub-scene images in sequence to obtain the upsampled fused image, and writing it to disk; after all sub-scene images have completed image upsampling fusion, reading the upsampled fused images from the disk sequentially according to the order of the sub-scene images, and stitching them together one by one to achieve full-scene imaging;

[0042] Calculate the memory required for sub-aperture imaging of each subset:

[0043]

[0044] Among them, (M) img ) j M represents the memory required for subaperture imaging of the j-th subset. apt N represents the memory space occupied by the echo data of the sub-aperture. j Let W be the number of sub-scene images in the j-th subset, and n be the index of the sub-scene image in the j-th subset. n I is the actual width of the nth sub-scene image. r I represents the distance pixel interval of the image. a M represents the azimuth pixel spacing of the image. tar M represents the memory space occupied by the 3D geographic grid points that fast BP imaging relies on. aux Memory space occupied by auxiliary data;

[0045] Calculate the memory required for full-scene stitching:

[0046]

[0047] Among them, M spl The memory required for full-scene stitching, M fig M represents the memory space occupied by all sub-scene images. ovp The memory space occupied by the overlapping area of ​​all sub-scene images;

[0048] The peak memory consumption during SAR imaging is:

[0049] M max =max(max((M) img ) j),M spl )

[0050] Among them, M max This represents peak memory consumption.

[0051] Secondly, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the aforementioned software design method for massive SAR data imaging based on the fast BP algorithm.

[0052] Thirdly, a computer program product is provided, including a computer program / instruction, which, when executed by a processor, implements the above-mentioned software design method for massive SAR data imaging based on the fast BP algorithm.

[0053] Compared with the prior art, this application has the following beneficial effects: This application addresses the existing fast BP imaging algorithm by accurately estimating the amount of data during the calculation process, and proposes to use the available memory of the computing server to dynamically correct the J operator, perform sub-set processing on sub-scenes, and use disk to temporarily store intermediate calculation results. This solves the memory limitation problem of SAR full-scene full-resolution imaging of ultra-large data on computing servers with low memory, and greatly reduces the hardware memory cost of computing servers. Attached Figure Description

[0054] This application can be better understood by referring to the description given below in conjunction with the accompanying drawings, which, together with the detailed description below, are incorporated in and form part of this specification. In the drawings:

[0055] Figure 1 A flowchart illustrating the design method of massive SAR data imaging software based on the fast BP algorithm is shown.

[0056] Figure 2 A schematic diagram of one of the uneven scene division methods in Scheme 2 is shown;

[0057] Figure 3 A diagram illustrating the sub-scene division is shown. Detailed Implementation

[0058] Exemplary embodiments of the present application will be described below with reference to the accompanying drawings. For clarity and brevity, not all features of the actual embodiments are described in the specification. However, it should be understood that many embodiment-specific decisions can be made in the development of any such actual embodiment to achieve the developer’s specific objectives, and these decisions may vary as the embodiments differ.

[0059] It should also be noted that, in order to avoid obscuring this application with unnecessary details, only the device structure closely related to the solution according to this application is shown in the accompanying drawings, while other details that are not closely related to this application are omitted.

[0060] It should be understood that this application is not limited to the described embodiments by virtue of the following description with reference to the accompanying drawings. In this document, embodiments may be combined with each other, features may be substituted or borrowed between different embodiments, and one or more features may be omitted in one embodiment, where feasible.

[0061] This application provides a method for designing massive SAR data imaging software based on a fast BP (BackProjection) algorithm. Figure 1 A flowchart illustrating the design method for massive SAR data imaging software based on the fast BP algorithm is shown. (See [link]) Figure 1 The methods include:

[0062] Step 1: Divide the full-scene single-view single-precision complex image into C r ×C a Individual scene images, C r C represents the number of sub-scene images at a distance. a C represents the number of sub-scene images in the orientation direction. r and C a All values ​​can be equal to 1, and all sub-scene images are divided into J subsets, with J initially set to 1, including:

[0063] Option 1: Divide all sub-scene images into J subsets along the orientation or distance direction;

[0064] Option 2: Divide all sub-scene images into J subsets; here, the distribution can be average or non-average. For example, when the total number of sub-scene images is 9 and J equals 2, it can be divided into 4+5. Figure 2 The diagram shows one of the uneven scene division methods in Scheme 2.

[0065] There is an overlapping area between two adjacent sub-scene images.

[0066] Among them, single-view: indicates that the image has not undergone excessive view processing and retains its original resolution. Single-precision: indicates that the image data is stored as single-precision floating-point numbers (32 bits). Complex image: indicates that the image data contains real and imaginary parts, which can express the amplitude and phase information of the radar signal.

[0067] Specifically, the division rules adopted include:

[0068] (1) The two-dimensional width of each sub-scene image is consistent and greater than the first set value W. min Here, Wmin (1) The width of the first dimension of the entire scene is any value greater than 0, which is smaller than the width of the lowest dimension of the entire scene. The value is determined according to actual needs. (2) The single-view 16-bit complex image corresponding to each sub-scene image is smaller than the second set value. Here, the second set value is any value greater than 0, which is smaller than the size of the single-view 16-bit complex image data of the entire scene. The value is determined according to actual needs. (3) There is an overlapping area between two adjacent sub-scene images. Here, the overlapping area is taken as 10%. Figure 3 A diagram illustrating the sub-scene division is shown.

[0069] Use C r Represents the number of sub-scenes in the distance-oriented segmentation (C r It is an integer and C r ≥1), use C a The number of sub-scene images in the orientation segmentation (C) a It is an integer and C a If ≥1), then the total number of sub-scene images is C. r ×C a All sub-scene images are divided into J subsets, where the number of sub-scenes contained in the j-th subset is denoted by N. j This indicates that j = 1, 2, ..., J.

[0070] Step 2: Under the current J value, calculate the memory space occupied by all sub-scene images based on the width of the sub-scene images, calculate the memory space occupied by the overlapping area of ​​all sub-scene images, and calculate the memory space occupied by the 3D geographic grid points on which fast BP imaging depends.

[0071] Based on the above constraints of image width and image byte size, when the single-view 16-bit complex image of a single sub-scene is less than 8GB, its corresponding single-view single-precision complex image data must be between 14GB and 16GB. The actual image width of the sub-scene is determined according to the following formula:

[0072]

[0073] Where W represents the width of the smaller dimension of the entire scene; if the width in the azimuth direction is smaller, then W represents the width of the entire scene in the azimuth direction; if the width in the distance direction is smaller, then W represents the width of the entire scene in the distance direction. min The first set value; W n I represents the width of the nth sub-scene image, in meters. r I represents the distance pixel interval of the image, in meters. a Let W be the azimuth pixel spacing of the image, in meters. The scene width W is calculated from the above set of inequalities. n It should be as large as possible under the given conditions.

[0074] Specifically, the memory space occupied by all sub-scene images is represented by the following formula:

[0075]

[0076] Among them, M fig Let N be the memory space occupied by all sub-scene images, where j is the subset index, J is the number of subsets, and N is the number of subsets. j Let W be the number of sub-scene images in the j-th subset, and n be the index of the sub-scene image in the j-th subset. n I is the actual width of the nth sub-scene image. r I represents the distance pixel interval of the image. a The pixel spacing in the orientation direction of the image.

[0077] Specifically, the memory space occupied by the overlapping area of ​​all sub-scene images is represented by the following formula:

[0078]

[0079] Among them, M ovp Let N be the memory space occupied by the overlapping regions of all sub-scene images, where j is the index of the subset, J is the number of subsets, and N is the number of subsets. j Let W be the number of sub-scene images in the j-th subset, and n be the index of the sub-scene image in the j-th subset. n I is the actual width of the nth sub-scene image. r I represents the distance pixel interval of the image. a β represents the pixel spacing in the orientation direction of the image, and β is the proportion of the overlapping region to the sub-scene image, which can be, for example, 10%.

[0080] Specifically, the memory space occupied by the 3D geographic grid points upon which fast BP imaging relies is expressed by the following formula:

[0081]

[0082] Among them, M tar M represents the memory space occupied by the 3D geographic grid points that fast BP imaging relies on. ovp M represents the memory space occupied by the overlapping regions of all sub-scene images. fig D represents the memory space occupied by all sub-scene images. n The sampling factor for geographic grid point coordinates along the azimuth gradient.

[0083] Step 3: When performing SAR imaging based on the fast BP algorithm, the echo signal needs to be divided into sub-apertures to obtain multiple sub-aperture signals; the memory space occupied by the echo data of the sub-aperture is calculated based on the azimuth pulse number and the range sampling point number.

[0084] Specifically, the memory space occupied by the sub-aperture echo data is expressed by the following formula:

[0085] M apt =N k ×N r ×8

[0086]

[0087] Among them, M apt N represents the memory space occupied by the echo data of the sub-aperture. k N represents the azimuth pulse number of the sub-aperture. r N represents the number of echo range sampling points. a T is the number of pulses in the azimuth direction over the full aperture time. a For the full aperture time, T k For a single sub-aperture time, To round down, This is for rounding up.

[0088] Step 4: Calculate the peak memory consumption during SAR imaging based on the memory space occupied by all sub-scene images, the memory space occupied by the overlapping area of ​​all sub-scene images, the memory space occupied by the 3D geographic grid points on which fast BP imaging depends, and the memory space occupied by the echo data of sub-apertures.

[0089] Two methods can be used to calculate the peak memory consumption during SAR imaging.

[0090] In Scheme 1, the SAR imaging process includes: for each subset, performing sub-aperture imaging, sub-aperture upsampling fusion, and first-level stitching of the sub-scene images in the subset to obtain the stitched image corresponding to the subset, and writing it to the disk; after each subset has completed image stitching, the stitched image corresponding to the subset is read from the disk in sequence according to the sorting of the subsets, and second-level stitching is performed in sequence to achieve full-scene imaging;

[0091] Calculate the memory required for sub-aperture imaging of each subset:

[0092]

[0093] Among them, (M) img ) j M represents the memory required for subaperture imaging of the j-th subset. apt N represents the memory space occupied by the echo data of the sub-aperture. j Let W be the number of sub-scene images in the j-th subset, and n be the index of the sub-scene image in the j-th subset. n I is the actual width of the nth sub-scene image. r I represents the distance pixel interval of the image.a M represents the azimuth pixel spacing of the image. tar M represents the memory space occupied by the 3D geographic grid points that fast BP imaging relies on. aux Memory space occupied by auxiliary data;

[0094] Calculate the memory required for sub-aperture upsampling fusion:

[0095]

[0096] It can be seen that (M) img ) j >(M up ) j .

[0097] Calculate the memory required for full-scene stitching:

[0098]

[0099] Among them, M spl The memory required for full-scene stitching, M fig M represents the memory space occupied by all sub-scene images. ovp The memory space occupied by the overlapping area of ​​all sub-scene images;

[0100] The peak memory consumption during SAR imaging is:

[0101] M max =max(max((M) img ) j ),M spl )

[0102] Among them, M max This represents peak memory consumption.

[0103] In Scheme 2, the SAR imaging process includes: for each subset, performing sub-aperture imaging and sub-aperture upsampling fusion on the sub-scene images in the subset to obtain the upsampled fused image, and writing it to the disk; after each sub-scene image has completed image upsampling fusion, the upsampled fused images are read from the disk one by one according to the order of the sub-scene images, and stitched together one by one to achieve full-scene imaging;

[0104] Calculate the memory required for sub-aperture imaging of each subset:

[0105]

[0106] Among them, (M) img ) j M represents the memory required for subaperture imaging of the j-th subset. aptN represents the memory space occupied by the echo data of the sub-aperture. j Let W be the number of sub-scene images in the j-th subset, and n be the index of the sub-scene image in the j-th subset. n I is the actual width of the nth sub-scene image. r I represents the distance pixel interval of the image. a M represents the azimuth pixel spacing of the image. tar M represents the memory space occupied by the 3D geographic grid points that fast BP imaging relies on. aux Memory space occupied by auxiliary data;

[0107] Calculate the memory required for full-scene stitching:

[0108]

[0109] Among them, M spl The memory required for full-scene stitching, M fig M represents the memory space occupied by all sub-scene images. ovp The memory space occupied by the overlapping area of ​​all sub-scene images;

[0110] The peak memory consumption during SAR imaging is:

[0111] M max =max(max((M) img ) j ),M spl )

[0112] Among them, M max This represents peak memory consumption.

[0113] Comparing the peak memory consumption of Scheme 2 and Scheme 1, it can be seen that Scheme 2 consumes less memory in this step than Scheme 1. That is, Scheme 2 has better peak memory consumption when splicing sub-scenes. However, in actual applications, the software flow designed in Scheme 1 is more flexible.

[0114] Step 5: Determine whether the peak memory consumption during the SAR imaging process is greater than the available memory M of the computing server. If yes, increase the value of J by 1 and return to step 2. If no, proceed to step 6.

[0115] Step 6: For Scheme 1, determine whether the J value is less than or equal to the number C of azimuth sub-scene images. a Or the distance to the number of sub-scene images C r For Scheme 2, determine whether the value of J is less than or equal to the number of sub-scene images C. r ×C aIf so, the computing server's memory is sufficient to achieve full-scene imaging and obtain the final J value; otherwise, the computing server's memory is insufficient to achieve full-scene imaging.

[0116] In the imaging software design method of this embodiment, in full-resolution imaging of sub-scenes, each sub-scene imaging requires the use of all sub-aperture echo signals. Therefore, the J value should be as small as possible to reduce the frequency of repeated echo signal analysis. That is, as many sub-scene sub-aperture images as possible should be completed after analyzing the echo signal of each sub-aperture. However, the more sub-scenes there are, the more memory is required. Therefore, the J value needs to be used to adjust the number of sub-scene images completed in a single sub-aperture echo analysis.

[0117] In summary, this application has the following technical advantages over the prior art: This application addresses the existing fast BP imaging algorithm by accurately estimating the amount of data during the computation process, proposing to utilize the available memory of the computing server to dynamically correct the J operator, perform sub-set processing on sub-scenes, and temporarily store intermediate computation results on disk. This solves the memory limitation problem of performing SAR full-scene full-resolution imaging on computing servers with low memory for ultra-large amounts of data, and greatly reduces the hardware memory cost of the computing server.

[0118] This application provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the above-described software design method for massive SAR data imaging based on the fast BP algorithm.

[0119] This application provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the above-described software design method for massive SAR data imaging based on the fast BP algorithm.

[0120] The above descriptions are merely various embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A software design method for imaging massive SAR data based on the fast BP algorithm, characterized in that, include: Step 1: Divide the full-scene single-view single-precision complex image into C r ×C a Individual scene images, C r C represents the number of sub-scene images at a distance. a This represents the number of sub-scene images in the orientation direction, and all sub-scene images are divided into J subsets, with J initially set to 1, including: Option 1: Divide all sub-scene images into J subsets along the orientation or distance direction; Option 2: Divide all sub-scene images into J subsets; There is an overlapping area between two adjacent sub-scene images; Step 2: Under the current J value, calculate the memory space occupied by all sub-scene images based on the width of the sub-scene images, calculate the memory space occupied by the overlapping area of ​​all sub-scene images, and calculate the memory space occupied by the 3D geographic grid points on which fast BP imaging depends. Step 3: When performing SAR imaging based on the fast BP algorithm, the echo signal needs to be divided into sub-apertures to obtain multiple sub-aperture signals; the memory space occupied by the echo data of the sub-aperture is calculated based on the azimuth pulse number and the range sampling point number of the sub-aperture. Step 4: Calculate the peak memory consumption during SAR imaging based on the memory space occupied by all sub-scene images, the memory space occupied by the overlapping area of ​​all sub-scene images, the memory space occupied by the three-dimensional geographic grid points on which fast BP imaging depends, and the memory space occupied by the echo data of the sub-aperture. Step 5: Determine whether the peak memory consumption during the SAR imaging process is greater than the available memory of the computing server. If yes, increase the J value by 1 and return to step 2. If no, proceed to step 6. Step 6: For Scheme 1, determine whether the J value is less than or equal to the number C of azimuth sub-scene images. a Or the distance to the number of sub-scene images C r For Scheme 2, determine whether the value of J is less than or equal to the number of sub-scene images C. r ×C a If so, the computing server's memory is sufficient to achieve full-scene imaging and obtain the final J value; otherwise, the computing server's memory is insufficient to achieve full-scene imaging.

2. The method as described in claim 1, characterized in that, in, The full-scene single-view single-precision complex image is divided into multiple sub-scene images, and the division rules include: Each sub-scene image has a consistent two-dimensional width that is greater than a first set value, the single-view 16-bit complex image corresponding to each sub-scene image is smaller than a second set value, and there is an overlapping area between two adjacent sub-scene images.

3. The method as described in claim 1, characterized in that, The memory space occupied by all sub-scene images is represented by the following formula: Among them, M fig Let N be the memory space occupied by all sub-scene images, where j is the subset index, J is the number of subsets, and N is the number of subsets. j Let W be the number of sub-scene images in the j-th subset, and n be the index of the sub-scene image in the j-th subset. n I is the actual width of the nth sub-scene image. r I represents the distance pixel interval of the image. a The pixel spacing in the orientation direction of the image.

4. The method as described in claim 1, characterized in that, The memory space occupied by the overlapping area of ​​all the sub-scene images is represented by the following formula: Among them, M ovp Let N be the memory space occupied by the overlapping regions of all sub-scene images, where j is the index of the subset, J is the number of subsets, and N is the number of subsets. j Let W be the number of sub-scene images in the j-th subset, and n be the index of the sub-scene image in the j-th subset. n I is the actual width of the nth sub-scene image. r I represents the distance pixel interval of the image. a α represents the pixel spacing in the orientation direction of the image, and α is the proportion of the overlapping region to the sub-scene image.

5. The method as described in claim 1, characterized in that, The memory space occupied by the 3D geographic grid points upon which the fast BP imaging relies is expressed by the following formula: Among them, M tar M represents the memory space occupied by the 3D geographic grid points that fast BP imaging relies on. ovp M represents the memory space occupied by the overlapping regions of all sub-scene images. fig D represents the memory space occupied by all sub-scene images. n The sampling factor for geographic grid point coordinates along the azimuth gradient.

6. The method as described in claim 1, characterized in that, The memory space occupied by the echo data of the sub-aperture is expressed by the following formula: M apt =N k ×N r ×8 Among them, M apt N represents the memory space occupied by the echo data of the sub-aperture. k N represents the azimuth pulse number of the sub-aperture. r N represents the number of echo range sampling points. a T is the number of pulses in the azimuth direction over the full aperture time. a For the full aperture time, T k For a single sub-aperture time, To round down, This is for rounding up.

7. The method as described in claim 1, characterized in that, The peak memory consumption during the calculation of SAR imaging includes: For Scheme 1, the SAR imaging process includes: for each subset, performing sub-aperture imaging, sub-aperture upsampling fusion, and first-level stitching of the sub-scene images in the subset in sequence to obtain the stitched image corresponding to the subset, and writing it to the disk; after each subset has completed image stitching, reading the stitched image corresponding to the subset from the disk in sequence according to the sorting of the subsets, and performing second-level stitching in sequence to achieve full-scene imaging; Calculate the memory required for sub-aperture imaging of each subset: Among them, (M) img ) j M represents the memory required for subaperture imaging of the j-th subset. apt N represents the memory space occupied by the echo data of the sub-aperture. j Let W be the number of sub-scene images in the j-th subset, and n be the index of the sub-scene image in the j-th subset. n I is the actual width of the nth sub-scene image. r I represents the distance pixel interval of the image. a M represents the azimuth pixel spacing of the image. tar M represents the memory space occupied by the 3D geographic grid points that fast BP imaging relies on. aux Memory space occupied by auxiliary data; Calculate the memory required for full-scene stitching: Among them, M spl The memory required for full-scene stitching, M fig M represents the memory space occupied by all sub-scene images. ovp The memory space occupied by the overlapping area of ​​all sub-scene images; The peak memory consumption during SAR imaging is: M max =max(max((M img ) j ),M spl ) Among them, M max This represents peak memory consumption.

8. The method as described in claim 1, characterized in that, The peak memory consumption during the calculation of SAR imaging includes: For Scheme 2, the SAR imaging process includes: dividing all sub-scene images into J subsets; for each subset, performing sub-aperture imaging and sub-aperture upsampling fusion on the sub-scene images in sequence to obtain the upsampled fused image, and writing it to disk; after all sub-scene images have completed image upsampling fusion, reading the upsampled fused images from the disk sequentially according to the order of the sub-scene images, and stitching them together one by one to achieve full-scene imaging; Calculate the memory required for sub-aperture imaging of each subset: Among them, (M) img ) j M represents the memory required for subaperture imaging of the j-th subset. apt N represents the memory space occupied by the echo data of the sub-aperture. j Let W be the number of sub-scene images in the j-th subset, and n be the index of the sub-scene image in the j-th subset. n I is the actual width of the nth sub-scene image. r I represents the distance pixel interval of the image. a M represents the azimuth pixel spacing of the image. tar M represents the memory space occupied by the 3D geographic grid points that fast BP imaging relies on. aux Memory space occupied by auxiliary data; Calculate the memory required for full-scene stitching: Among them, M spl The memory required for full-scene stitching, M fig M represents the memory space occupied by all sub-scene images. ovp The memory space occupied by the overlapping area of ​​all sub-scene images; The peak memory consumption during SAR imaging is: M max =max(max((M img ) j ),M spl ) Among them, M max This represents peak memory consumption.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the software design method for massive SAR data imaging based on the fast BP algorithm as described in any one of claims 1-8.

10. A computer program product, characterized in that, It includes a computer program / instruction, which, when executed by a processor, implements the software design method for massive SAR data imaging based on the fast BP algorithm as described in any one of claims 1-8.

Images