Method and device for reconstructing BP imaging SAR data into RD imaging mode

Abstract

The invention discloses a method and device for reconstructing BP imaging SAR data into an RD imaging mode, and belongs to the technical field of SAR imaging and image geometric process.The method comprises the steps that fitting is conducted on a BP imaging sensor track, and a continuous function of the sensor position changing along with time is constructed; calculating a statistical characteristic value from the image distance to the slope distance; designing a set of equivalent RD virtual imaging parameters; for each pixel in the RD virtual imaging grid, calculating a ground point coordinate corresponding to the pixel through geometric positioning, reversely mapping the ground point coordinate to a BP image space, obtaining an amplitude value and a complex value, and generating an amplitude image and a single-view complex image of RD virtual imaging in combination with an orbit compensation phase; virtual imaging parameters are integrated into an auxiliary data file in a standard format, and the auxiliary data file, the amplitude image and the single-view complex image jointly form a complete reconstructed data product. According to the invention, the application value and the application range of BP imaging data are expanded, and the defects of the BP algorithm ecological tool chain are overcome.

Description

Technical Field

[0001] This invention relates to a method and apparatus for reconstructing BP imaging SAR data into RD imaging mode, specifically to a method and apparatus for reconstructing SAR image data generated using the Back Projection (BP) algorithm into data conforming to the geometric characteristics of the Range-Doppler (RD) model, belonging to the field of synthetic aperture radar (SAR) imaging and image geometry processing technology. Background Technology

[0002] Synthetic Aperture Radar (SAR), as an active microwave remote sensing sensor, has imaging processing algorithms mainly divided into two categories: frequency domain algorithms and time domain algorithms. Range-Doppler (RD) algorithms, as the most classic representative of frequency domain algorithms, have developed extremely mature geometric models and supporting processing toolchains, and are widely used in advanced applications such as geometric correction of SAR images, stereo mapping, and interferometry (InSAR).

[0003] Back projection (BP) is a precise temporal imaging method with advantages in geometric accuracy for handling complex trajectories and scenes with large squint. BP methods typically image on a given initial DEM to acquire high-quality images. However, the BP algorithm lacks an explicit and concise description of the "image-object" geometric relationship like the RD model, making it difficult to directly utilize various mature processing software and algorithms developed based on the RD model. This mismatch between "data and tools" severely limits the subsequent application value of high-quality BP imaging data.

[0004] In existing technologies, high-precision geometric processing of BP imaging data typically involves two options: one is to develop a dedicated processing tool for BP data, which is costly and time-consuming; the other is to attempt to re-image the original echo data using the RD algorithm, but this requires the original data to be available and may not be able to reproduce the imaging advantages of the BP algorithm in specific scenarios. Therefore, there is an urgent need for a method that can transform data at the data level to "adapt" BP imaging results to the RD processing ecosystem. Summary of the Invention

[0005] To address the aforementioned issues, this invention proposes a method for reconstructing BP imaging SAR data into RD imaging mode, which enables seamless integration of BP imaging data with mature RD processing toolchains, fully combining the geometric accuracy of BP imaging with the ecological convenience of RD processing.

[0006] The technical solution adopted by this invention to solve its technical problem is as follows: In a first aspect, embodiments of the present invention provide a method for reconstructing BP imaging SAR data into RD imaging mode, comprising the following steps: Step S1: Extract the imaging time and corresponding sensor position coordinates from the BP imaging data, and construct a continuous function of sensor position change over time using a polynomial curve fitting method. Step S2: Based on the geometric relationship between the ground coordinates of the BP image points and the sensor position coordinates, calculate the statistical characteristic value of the image range syncline. Step S3: Based on the continuous function and statistical characteristic values, design a set of equivalent RD virtual imaging parameters, including virtual imaging trajectory, Doppler center frequency, initial slant range, imaging time range, image width and height, wavelength and sampling resolution. Step S4: For each pixel in the RD virtual imaging grid, calculate its corresponding ground point coordinates through geometric positioning, back-map it to the BP image space, obtain the amplitude and complex values ​​from the BP imaging data through interpolation, and combine them with orbital compensation phase to generate the amplitude image and single-view complex image of the RD virtual imaging. Step S5: Integrate the virtual imaging parameters into a standard format auxiliary data file, which together with the amplitude image and single-view complex image of RD virtual imaging constitutes a complete reconstructed data product.

[0007] As one possible implementation of this embodiment, step S1 includes the following steps: Step S11: Extract multiple discrete imaging time points and their corresponding sensor position coordinates from the BP imaging data; Step S12: Perform polynomial curve fitting on the extracted sensor position coordinates along the time axis to construct a continuous function of sensor position change over time. Step S13: Based on the continuous function obtained from the fitting, the sensor velocity function is obtained by differentiation, which is used for subsequent Doppler frequency calculation.

[0008] As one possible implementation of this embodiment, the function form of the polynomial curve fitting is: , in, For direction and time, These are the parameters for orbit fitting.

[0009] As one possible implementation of this embodiment, step S2 includes the following steps: Step S21: Obtain the ground point coordinates and sensor position coordinates corresponding to all pixels in the first column of the BP image; Step S22: Calculate the slant distance corresponding to each first column cell, and calculate the mean of these slant distances as a near-range reference value; Step S23: Obtain the ground point coordinates and sensor position coordinates corresponding to all pixels in the last column of the BP image; Step S24: Calculate the slant distance corresponding to each last column of cells, and calculate the mean of these slant distances as a long-distance reference value; Step S25: Calculate the synclinal range based on the near-range reference value and the far-range reference value.

[0010] As one possible implementation of this embodiment, step S3 includes the following steps: Step S31: Use the fitted sensor trajectory continuity function as the trajectory after trajectory compensation for virtual RD imaging. Step S32: Based on the geometric relationship of each pixel in BP imaging, calculate the instantaneous Doppler frequency value, and average all instantaneous Doppler frequency values ​​to obtain the Doppler center frequency used for RD virtual imaging. Step S33: Use the calculated near-range reference value as the initial slant range for virtual imaging; Step S34: Use the start and end times of BP imaging as the start and end times of virtual imaging, respectively. Step S35: The width, height, and wavelength of the BP image are directly used as the corresponding parameters for virtual imaging to calculate the range sampling resolution and azimuth temporal resolution.

[0011] As one possible implementation of this embodiment, the formula for calculating the instantaneous Doppler frequency value is as follows: , Where λ is the wavelength, V is the sensor velocity vector, P is the ground point coordinates, and S is the sensor position coordinates; The range sampling resolution The calculation formula is: , in For near-range mean, The mean is the distance. Where W is the slant range and W is the image width; The formula for calculating the azimuth time resolution Δt is: , in , These represent the start and end times, respectively, and H represents the image height.

[0012] As one possible implementation of this embodiment, step S4 includes the following steps: Step S41: Using the RD geometric positioning model and combining it with digital elevation model data, convert the RD virtual cell coordinates into the corresponding three-dimensional coordinates of ground points. Step S42: Calculate the sub-pixel position of the ground point on the BP image based on the ground point coordinates and the BP imaging geometry. Step S43: On the BP amplitude image, with the sub-pixel position as the center, take the amplitude values ​​of the four nearest neighboring pixels around it, and use bilinear interpolation to calculate the amplitude value of the point. Step S44: On the BP complex image, with the sub-pixel position as the center, take the complex values ​​of multiple neighboring pixels around it, and use the Sinc interpolation method to calculate the basic complex value of the point. Step S45: Calculate the slant range difference between the ground point and the position of the RD virtual imaging sensor and the position of the BP actual imaging sensor, convert the slant range difference into a phase difference, and perform phase compensation on the basic complex value to obtain the final complex value. Step S46: Assign the calculated amplitude value and the final complex value to the RD virtual pixel and store them in the corresponding image matrix.

[0013] As one possible implementation of this embodiment, step S42 specifically includes: Based on the ground point coordinates, find the four BP imaging points that are closest to the virtual RD image point's location point; Based on the object point coordinates and image point coordinates of the four nearest BP imaging points, a linear fit is made to the function from the object point coordinates to the BP image point coordinates. Based on the geometric location of the ground point coordinates of the virtual RD image point, the image point coordinates on the BP image are calculated using a function obtained by linear fitting. These coordinates are usually non-integer values.

[0014] As one possible implementation of this embodiment, step S5 includes the following steps: Step S51: Integrate all RD virtual imaging parameters according to the standard RD data format requirements to generate an auxiliary data file; Step S52: The auxiliary data file is packaged with the generated RD virtual imaging amplitude image and single-view complex image to form a complete data product; Step S53: Add necessary metadata to the data product, including data generation time, processing software version, and coordinate system information; Step S54: Convert the data product to a standard format and add metadata. Then, perform an integrity check on the generated data product to ensure that all necessary data files and parameter files exist and are in the correct format.

[0015] Secondly, an embodiment of the present invention provides an apparatus for reconstructing BP imaging SAR data into RD imaging mode, comprising: The trajectory fitting module is used to extract the imaging time and corresponding sensor position coordinates from the BP imaging data, and to construct a continuous function of sensor position change over time using a polynomial curve fitting method. The parameter statistics module is used to calculate the statistical characteristic value of the image range syncline based on the geometric relationship between the ground coordinates of the BP image points and the sensor position coordinates. The parameter design module is used to design a set of equivalent RD virtual imaging parameters based on the continuous function and statistical characteristic values. The RD virtual imaging parameters include virtual imaging trajectory, Doppler center frequency, initial slant range, imaging time range, image width and height, wavelength and sampling resolution. The data reconstruction module is used to calculate the corresponding ground point coordinates for each pixel in the RD virtual imaging grid through geometric positioning, back-map it to the BP image space, obtain the amplitude and complex values ​​from the BP imaging data through interpolation, and generate the amplitude image and single-view complex image of the RD virtual imaging by combining the orbital compensation phase. The auxiliary data generation module is used to integrate virtual imaging parameters into a standard format auxiliary data file, which together with the amplitude image and single-view complex image of RD virtual imaging constitutes a complete reconstruction data product.

[0016] One of the above technical solutions has the following advantages or beneficial effects: One of the above technical solutions enables BP imaging data to be directly processed using a large number of existing commercial software and open-source tools based on RD models, such as geometric correction and InSAR, which greatly expands its application scope and improves its compatibility. One of the above technical solutions addresses the dilemma of BP imaging data being "easy to use but difficult to process," thus unlocking its quality advantages in imaging complex scenes.

[0017] One of the above technical solutions provides a technical path to transform non-standard SAR data products into standard products, which is beneficial for data sharing and joint analysis.

[0018] One of the above technical solutions relies solely on the final results and basic metadata of BP imaging, without requiring raw echo data, thus having a low implementation threshold.

[0019] One of the above technical solutions is based on an understanding of BP imaging data and its imaging process. It reconstructs the data at the data level, resulting in a data product that possesses all the characteristics of standard RD data. This allows for seamless integration with existing, mature RD data processing pipelines (such as geometric positioning, orthorectification, and interferometry), achieving seamless integration between BP imaging data and mature RD processing toolchains. It fully combines the geometric accuracy of BP imaging with the ecological convenience of RD processing, thereby greatly expanding the application value and scope of BP imaging data and compensating for the shortcomings of the BP algorithm ecosystem toolchain. For BP imaging data on initial DEMs (such as open-source DEM data), the above technical solution yields better reconstruction results. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating a method for reconstructing BP imaging SAR data into RD imaging mode according to an exemplary embodiment; Figure 2 This is a schematic block diagram of an apparatus for reconstructing BP imaging SAR data into RD imaging mode according to an exemplary embodiment; Figure 3 This is a flowchart illustrating a specific implementation of BP to RD imaging data reconstruction according to an exemplary embodiment. Detailed Implementation

[0021] To more clearly illustrate the technical features of the present invention, the present invention will be described in detail below through specific embodiments and in conjunction with the accompanying drawings.

[0022] There is currently an urgent technical need: how to combine the advantages of the BP algorithm in image quality with the maturity and convenience of the RD model in subsequent geometric processing. Simply reprocessing the original echo data with the RD algorithm is not always feasible or optimal, because the original data may not be available, or BP imaging itself is the best choice for specific scenarios (such as non-uniform motion, large strabismus).

[0023] Therefore, this invention aims to resolve this contradiction by proposing a "data-level" reconstruction method. Instead of re-imaging, it constructs a geometrically equivalent "virtual RD dataset" through a deep understanding of BP imaging results and its process. This virtual data possesses all the characteristics of standard RD data (single-view complex image (SLC), orbital parameters, Doppler parameters, etc.), thus enabling seamless integration into existing RD data processing pipelines. This is equivalent to "empowering" BP image data, allowing it to immediately benefit from all the processing tools and application results accumulated over decades of RD model development, thereby enhancing the practical value and application scope of BP imaging data.

[0024] like Figure 1 As shown in the figure, an embodiment of the present invention provides a method for reconstructing BP imaging SAR data into RD imaging mode, comprising the following steps: Step S1: Extract the imaging time and corresponding sensor position coordinates from the BP imaging data, and construct a continuous function of sensor position change over time using a polynomial curve fitting method. Step S2: Based on the geometric relationship between the ground coordinates of the BP image points and the sensor position coordinates, calculate the statistical characteristic value of the image range syncline. Step S3: Based on the continuous function and statistical characteristic values, design a set of equivalent RD virtual imaging parameters, including virtual imaging trajectory, Doppler center frequency, initial slant range, imaging time range, image width and height, wavelength and sampling resolution. Step S4: For each pixel in the RD virtual imaging grid, calculate its corresponding ground point coordinates through geometric positioning, back-map it to the BP image space, obtain the amplitude and complex values ​​from the BP imaging data through interpolation, and combine them with orbital compensation phase to generate the amplitude image and single-view complex image of the RD virtual imaging. Step S5: Integrate the virtual imaging parameters into a standard format auxiliary data file, which together with the amplitude image and single-view complex image of RD virtual imaging constitutes a complete reconstructed data product.

[0025] As one possible implementation of this embodiment, step S1 includes the following steps: Step S11: Extract multiple discrete imaging time points and their corresponding sensor position coordinates from the BP imaging data; Step S12: Perform polynomial curve fitting on the extracted sensor position coordinates along the time axis to construct a continuous function of sensor position change over time. Step S13: Based on the continuous function obtained from the fitting, the sensor velocity function is obtained by differentiation, which is used for subsequent Doppler frequency calculation.

[0026] As one possible implementation of this embodiment, the polynomial curve fitting adopts a cubic polynomial fitting method, and the functional form of the cubic polynomial fitting is: , in, For direction and time, These are the parameters for orbit fitting.

[0027] The least squares method is used to fit a polynomial curve, minimizing the sum of squared residuals between the fitted curve and the original sensor position coordinates.

[0028] The BP imaging data is the BP algorithm imaging result of airborne SAR or spaceborne SAR, and the sensor position coordinates come from navigation data recorded during the BP imaging process or subsequent precise orbit determination data.

[0029] As one possible implementation of this embodiment, step S2 includes the following steps: Step S21: Obtain the ground point coordinates and sensor position coordinates corresponding to all pixels in the first column of the BP image; Step S22: Calculate the slant distance corresponding to each first column cell, and calculate the mean of these slant distances as a near-range reference value; Step S23: Obtain the ground point coordinates and sensor position coordinates corresponding to all pixels in the last column of the BP image; Step S24: Calculate the slant distance corresponding to each last column of cells, and calculate the mean of these slant distances as a long-distance reference value; Step S25: Calculate the synclinal range based on the near-range reference value and the far-range reference value.

[0030] The ground point coordinates are obtained by back-calculation using the BP imaging geometric model, or directly read from the ground point coordinate data recorded during the BP imaging process.

[0031] The slant distance is calculated using the Euclidean distance formula: , in( , , ) represents the coordinates of a ground point. , , ) represents the sensor's position coordinates.

[0032] As one possible implementation of this embodiment, step S3 includes the following steps: Step S31: Use the fitted sensor trajectory continuity function as the trajectory after trajectory compensation for virtual RD imaging. Step S32: Based on the geometric relationship of each pixel in BP imaging, calculate the instantaneous Doppler frequency value, and average all instantaneous Doppler frequency values ​​to obtain the Doppler center frequency used for RD virtual imaging. Step S33: Use the calculated near-range reference value as the initial slant range for virtual imaging; Step S34: Use the start and end times of BP imaging as the start and end times of virtual imaging, respectively. Step S35: The width, height, and wavelength of the BP image are directly used as the corresponding parameters for virtual imaging to calculate the range sampling resolution and azimuth temporal resolution.

[0033] As one possible implementation of this embodiment, the formula for calculating the instantaneous Doppler frequency value is as follows: , Where λ is the wavelength, V is the sensor velocity vector, P is the ground point coordinates, and S is the sensor position coordinates; The range sampling resolution The calculation formula is: , in For near-range mean, The mean is the distance. Where W is the slant range and W is the image width; The formula for calculating the azimuth time resolution Δt is: , in , These represent the start and end times, respectively, and H represents the image height.

[0034] Step S3 further includes step S36: determining the side-view direction parameters of the virtual imaging based on the side-view direction of the BP imaging.

[0035] Step S3 further includes calculating the pulse repetition frequency of the virtual imaging, using the following formula: , where Δt is the azimuth time resolution.

[0036] As one possible implementation of this embodiment, step S4 includes the following steps: Step S41: Using the RD geometric positioning model and combining it with digital elevation model data, convert the RD virtual cell coordinates into the corresponding three-dimensional coordinates of ground points. Step S42: Calculate the sub-pixel position of the ground point on the BP image based on the ground point coordinates and the BP imaging geometry. Step S43: On the BP amplitude image, with the sub-pixel position as the center, take the amplitude values ​​of the four nearest neighboring pixels around it, and use bilinear interpolation to calculate the amplitude value of the point. Step S44: On the BP complex image, with the sub-pixel position as the center, take the complex values ​​of multiple neighboring pixels around it, and use the Sinc interpolation method to calculate the basic complex value of the point. Step S45: Calculate the slant range difference between the ground point and the position of the RD virtual imaging sensor and the position of the BP actual imaging sensor, convert the slant range difference into a phase difference, and perform phase compensation on the basic complex value to obtain the final complex value. Step S46: Assign the calculated amplitude value and the final complex value to the RD virtual pixel and store them in the corresponding image matrix.

[0037] As one possible implementation of this embodiment, the RD geometric positioning model is based on the range-Doppler equation and solves the ground point coordinates through an iterative algorithm. The convergence condition of the iteration is that both the slant range error and the Doppler frequency error are less than a preset threshold.

[0038] As one possible implementation of this embodiment, step S42 specifically involves: querying the four closest BP imaging points to the virtual RD image point's location point based on the ground point coordinates; linearly fitting a function from the object point coordinates to the BP image point coordinates based on the object point coordinates and image point coordinates of the four closest BP imaging points; and calculating the image point coordinates of the virtual RD image point on the BP image using the fitted linear function based on the geometrically located ground point coordinates of the virtual RD image point, which is usually a non-integer value.

[0039] The interpolation kernel function of the Sinc interpolation method is: , in Let be the complex value of the neighboring pixels in the BP complex image, (x, y) be the coordinates of the subpixel position, and Δx and Δy be the sampling intervals.

[0040] The specific calculation for the phase compensation is as follows: , , , Where λ is the wavelength.

[0041] When the subpixel position exceeds the boundary of the BP image, the nearest neighbor interpolation method or the boundary filling method is used for processing.

[0042] The weighting coefficients of the bilinear interpolation method are determined based on the distance between the sub-pixel position and its four neighboring pixels.

[0043] As one possible implementation of this embodiment, step S5 includes the following steps: Step S51: Integrate all RD virtual imaging parameters according to the standard RD data format requirements to generate an auxiliary data file; Step S52: The auxiliary data file is packaged with the generated RD virtual imaging amplitude image and single-view complex image to form a complete data product; Step S53: Add necessary metadata to the data product, including data generation time, processing software version, and coordinate system information, etc. Step S54: Convert the data product to a standard format and add metadata. Then, perform an integrity check on the generated data product to ensure that all necessary data files and parameter files exist and are in the correct format.

[0044] The auxiliary data file includes the following parameters: RD virtual imaging trajectory data, image width and height, Doppler center frequency, imaging start and end time, range sampling resolution, azimuth time resolution, radar side-looking direction, operating wavelength, initial slant range, pulse repetition frequency, and radar operating mode.

[0045] The encapsulation method uses compression packaging, which packages multiple files into a single compressed file. The compression format includes ZIP, TAR, or a dedicated binary format.

[0046] When the standard format is CEOS format, the auxiliary data files include Leader File, Image File, and Trailer File, and the file header information and parameter records are set according to the CEOS standard.

[0047] The metadata also includes data quality indicators, processing accuracy information, geographic reference information, and time zone information.

[0048] The integrity verification includes file size checking, file header verification, data verification and calculation, and parameter consistency checking.

[0049] like Figure 2 As shown in the figure, an embodiment of the present invention provides an apparatus for reconstructing BP imaging SAR data into RD imaging mode, comprising: The trajectory fitting module is used to extract the imaging time and corresponding sensor position coordinates from the BP imaging data, and to construct a continuous function of sensor position change over time using a polynomial curve fitting method. The parameter statistics module is used to calculate the statistical characteristic value of the image range syncline based on the geometric relationship between the ground coordinates of the BP image points and the sensor position coordinates. The parameter design module is used to design a set of equivalent RD virtual imaging parameters based on the continuous function and statistical characteristic values. The RD virtual imaging parameters include virtual imaging trajectory, Doppler center frequency, initial slant range, imaging time range, image width and height, wavelength and sampling resolution. The data reconstruction module is used to calculate the corresponding ground point coordinates for each pixel in the RD virtual imaging grid through geometric positioning, back-map it to the BP image space, obtain the amplitude and complex values ​​from the BP imaging data through interpolation, and generate the amplitude image and single-view complex image of the RD virtual imaging by combining the orbital compensation phase. The auxiliary data generation module is used to integrate virtual imaging parameters into a standard format auxiliary data file, which together with the amplitude image and single-view complex image of RD virtual imaging constitutes a complete reconstruction data product.

[0050] The technical implementation path of this invention is as follows: First, based on the original parameters of BP imaging, an equivalent set of RD virtual imaging parameters is designed; second, for each virtual pixel in the RD imaging grid, the corresponding ground point coordinates are calculated; based on these coordinates, the amplitude value is obtained from the amplitude data of BP imaging through bilinear interpolation, and the complex value is obtained from the complex data through Sinc interpolation; finally, through orbit compensation technology, a single-view complex image (SLC) that geometrically strictly corresponds to the RD model is generated.

[0051] This invention effectively utilizes existing mature RD model processing tools to perform high-precision geometric processing on imaging data, expanding the application processing level of BP imaging data.

[0052] This invention aims to address the difficulty of utilizing mature RD geometric processing tools for BP imaging SAR data. Its core implementation involves constructing a virtual RD dataset that is geometrically equivalent to the original BP imaging data through a series of rigorous steps. For example... Figure 3 As shown, the specific process of reconstructing BP imaging SAR data into RD imaging mode according to the present invention is as follows.

[0053] Step 1, BP imaging trajectory fitting: The imaging time and sensor position of the BP image points are extracted, and the sensor trajectory is fitted using a cubic curve to construct a functional relationship between the sensor position and time. The purpose of step 1 is to obtain a continuous and smooth sensor trajectory function, which is used for subsequent RD imaging trajectory compensation.

[0054] From the metadata accompanying the BP imaging data, a series of discrete imaging time points and their corresponding sensor 3D position coordinates are extracted. Subsequently, a cubic polynomial curve was used to perform least-squares fitting on the coordinate components in the X, Y, and Z directions. The fitted function is as follows: , in, For direction and time, These are the parameters for orbit fitting.

[0055] This function provides a mathematical description of the "compensated trajectory" in subsequent virtual RD imaging.

[0056] Step 2, BP imaging parameter statistics: Based on the relationship between the object coordinates of the BP image points and the sensor position coordinates, the average maximum and average minimum slant range values ​​corresponding to the first and last columns of pixels are calculated respectively. Step 2 aims to obtain some global geometric parameters of BP imaging.

[0057] Using the slant distances of all pixels in the first column of the BP imaging image to their corresponding sensor positions, the mean of these slant distances is calculated (as a near-range reference); similarly, the slant distances of all pixels in the last column are calculated, and the mean is calculated (as a far-range reference). These statistical values ​​(such as the near-range mean) Long-distance mean This will be used to determine the range of the virtual RD image.

[0058] Step 3, RD virtual imaging parameter design: Step 3 is a key step in this invention, the purpose of which is to construct a set of self-consistent and standard RD imaging data.

[0059] First, the RD virtual imaging parameters are obtained. The core of this process is to invert and derive equivalent RD virtual imaging parameters from the metadata and results of BP imaging. The design and calculation of the RD virtual imaging parameters use the cubic curve fitting trajectory of the BP sensor trajectory as the trajectory after virtual imaging trajectory compensation, the Doppler mean value corresponding to the BP imaging as the virtual imaging Doppler value, the mean slant range of the first column of the BP image as the initial slant range of the virtual imaging, the BP start imaging time as the initial virtual imaging time, the BP end imaging time as the virtual imaging end time, the BP imaging wavelength as the virtual imaging wavelength, the BP image width as the virtual imaging image width, and the BP image height as the virtual imaging image height. Then, the RD virtual imaging slant range resolution and azimuth time resolution are calculated based on the azimuth imaging time, the difference between slant range lengths, the imaging width, and the height.

[0060] Virtual track: directly use the cubic polynomial obtained from step 1 for fitting. As the ideal trajectory after trajectory compensation in virtual RD imaging.

[0061] Doppler center frequency For each BP cell, based on its ground point Imaging time Sensor position and speed (This can be obtained by differentiating the orbital function), and the wavelength. Calculate its instantaneous Doppler The calculation formula is: , in, For the sensor velocity vector, The vector from the sensor to the ground point. Slope distance λ is the wavelength.

[0062] For all pixels The arithmetic mean is used to obtain the Doppler center frequency used in virtual RD imaging. .

[0063] Initial slope distance Use the average slant distance of the first column of pixels obtained in step 2. .

[0064] Imaging time, size, and wavelength: Inherited from the start time of BP azimuth imaging End time Image width (Distance in pixels) and height (Number of pixels in azimuth direction), and wavelength .

[0065] Resolution calculation: Range sampling resolution : for , Azimuth time resolution : , These parameters collectively define the geometric framework of virtual RD imaging.

[0066] Step 4, BP to RD imaging data reconstruction: Based on the width and height of the virtual imaging, blank amplitude and complex image matrices are created. For each pixel in the RD virtual imaging grid, the ground point coordinates obtained by its location are back-mapped to the BP image space. The amplitude value of the pixel is obtained from the BP amplitude data using bilinear interpolation, and the fundamental complex value is obtained from the BP complex data using Sinc interpolation. Combined with orbital compensation phase, the complete RD virtual imaging amplitude image and single-view complex image (SLC) are finally generated.

[0067] Step 4 performs the actual pixel value reconstruction. First, create a file of size [size missing]. The blank amplitude matrix and complex matrix. Then, for each pixel in the virtual RD image. (c represents the azimuth direction, r represents the range direction), execute the following sub-steps: Step 4.1: Target localization. Using the RD geometric localization model, the pixel coordinates are... By combining the data with known DEM data and using iterative calculations (such as the distance-Doppler equation), the three-dimensional coordinates of the ground point corresponding to the pixel can be calculated. .

[0068] Step 4.2: Calculation of RD and BP imaging sensor positions. Based on ground points Calculate the sensor positions corresponding to RD imaging and BP imaging, as well as the BP imaging points. (Row, column). This position It is usually subpixel.

[0069] Step 4.3: Amplitude Value Calculation. Locate the position on the BP amplitude image. The four nearest neighbors are calculated using bilinear interpolation. The amplitude value of the point is assigned to the RD virtual cell. .

[0070] Step 4.4: Complex numerical calculation and phase compensation. This is crucial for ensuring the phase consistency of complex data.

[0071] a. On BP complex images, for location Sinc interpolation is performed on the surrounding pixels to obtain an initial complex value. ; b. Based on ground points The coordinates of the image point p in the BP imaging are interpolated to obtain: BP actual imaging time Corresponding sensor location , Virtual RD Imaging Time (Defined by virtual RD imaging timing, typically) The corresponding sensor location ; c. Calculate the distance difference: ; d. Calculate the compensation phase: ; e. Generate the final complex value: ,Will Assign virtual pixels to RD This operation "injects" the geometric relationships of the virtual RD imaging into the complex phase.

[0072] Step 5, Extraction of RD Virtual Imaging Auxiliary Data: All RD virtual imaging parameters designed in step 3 are organized and stored according to the specifications of standard RD data products. This auxiliary data typically includes a descriptive metadata file (XML or text format) covering the RD virtual imaging's orbit / track data, image width and height, Doppler center frequency, imaging start and end times, range sampling resolution, azimuth time resolution, radar side-looking direction, operating wavelength, and initial slant range. These parameters are integrated into standard auxiliary data. This auxiliary data file, together with the virtual RD amplitude image and single-look complex image (SLC) generated in step 4, constitutes the final data reconstruction product. Users can directly import this product into SAR processing software that supports RD models for subsequent geometric correction, mapping, or interferometric processing.

[0073] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. For data imaged by BP on an initial DEM (such as open-source DEM data), the above technical solutions will yield better reconstruction results. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. These modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method of reconstructing BP imaging SAR data into an RD imaging mode, characterized by, The method comprises the following steps: Step S1, extracting the imaging time and corresponding sensor position coordinates in the BP imaging data, and constructing a continuous function of the sensor position changing with time by using a polynomial curve fitting method; Step S2, calculating the statistical characteristic value of the image slant range based on the geometric relationship between the ground coordinates of the BP image points and the sensor position coordinates; Step S3, designing a set of equivalent RD virtual imaging parameters based on the continuous function and the statistical characteristic value, the RD virtual imaging parameters comprising a virtual imaging track, a Doppler center frequency, an initial slant range, an imaging time range, an image width and height, a wavelength and a sampling resolution; Step S4, for each image element in the RD virtual imaging grid, calculating the corresponding ground point coordinates by geometric positioning, inversely mapping to the BP image space, obtaining the amplitude value and complex value from the BP imaging data by an interpolation method, and combining the track compensation phase to generate the amplitude image and single-view complex image of the RD virtual imaging; Step S5, integrating the virtual imaging parameters into an auxiliary data file in a standard format, and jointly constituting a complete reconstruction data product with the amplitude image and single-view complex image of the RD virtual imaging.

2. The method of claim 1, wherein, The step S1 comprises the following steps: Step S11, extracting a plurality of discrete imaging time points and corresponding sensor position coordinates from the BP imaging data; Step S12, performing polynomial curve fitting on the extracted sensor position coordinates along the time axis respectively, and constructing a continuous function of the sensor position changing with time; Step S13, obtaining the sensor velocity function by derivation according to the obtained continuous function, for subsequent Doppler frequency calculation.

3. The method of claim 2, wherein, The function form of the polynomial curve fitting is: ， wherein, is the azimuth time, is the orbit fitting parameter.

4. The method of claim 1, wherein, The step S2 comprises the following steps: Step S21, obtaining the ground point coordinates and sensor position coordinates corresponding to all image elements of the first column of the BP image; Step S22, calculating the slant range corresponding to each first column image element, and calculating the mean value of the slant ranges as a near distance reference value; Step S23, obtaining the ground point coordinates and sensor position coordinates corresponding to all image elements of the last column of the BP image; Step S24, calculating the slant range corresponding to each last column image element, and calculating the mean value of the slant ranges as a far distance reference value; Step S25, calculating the distance slant range range based on the near distance reference value and the far distance reference value.

5. The method of claim 1, wherein, The step S3 comprises the following steps: Step S31, taking the obtained continuous function of the sensor track as the track after track compensation of the virtual RD imaging; Step S32, calculating the instantaneous Doppler frequency value based on the geometric relationship of each image element in the BP imaging, and averaging all the instantaneous Doppler frequency values to obtain the Doppler center frequency used in the RD virtual imaging; Step S33, taking the calculated near distance reference value as the initial slant range of the virtual imaging; Step S34, taking the start time and end time of the BP imaging as the start time and end time of the virtual RD imaging respectively; Step S35, taking the width, height and wavelength of the BP image as the corresponding parameters of the virtual imaging, and calculating the distance sampling resolution and the azimuth time resolution.

6. The method of claim 5, wherein, The calculation formula of the instantaneous Doppler frequency value is: ， Where λ is wavelength, V is sensor velocity vector, P is ground point coordinate, S is sensor position coordinate; The distance is sampled at a resolution The formula for calculating the distance is: ， wherein is the near range mean, is the far range mean, is the slant range, and W is the image width; The formula for calculating the azimuth time resolution Δt is: ， wherein , are the start and end times, respectively, and H is the height of the video.

7. The method of claim 1, wherein, The step S4 comprises the following steps: Step S41, using the RD geometric positioning model, combined with digital elevation model data, the RD virtual pixel coordinates are converted into corresponding ground point three-dimensional coordinates; Step S42, according to the ground point coordinates and BP imaging geometric relationship, the sub-pixel position on the BP image is calculated; Step S43, on the BP amplitude image, taking the amplitude values of the four nearest pixels around the sub-pixel position as the center, the amplitude value of the point is calculated by using the bilinear interpolation method; Step S44, on the BP complex image, taking the complex values of multiple adjacent pixels around the sub-pixel position as the center, the basic complex value of the point is calculated by using the Sinc interpolation method; Step S45, the slant range difference between the ground point and the RD virtual imaging sensor position and the BP actual imaging sensor position is calculated, the slant range difference is converted into phase difference, and the basic complex value is phase compensated to obtain the final complex value; Step S46, the calculated amplitude value and the final complex value are respectively assigned to the RD virtual pixel, and stored in the corresponding image matrix.

8. The method of claim 7, wherein, The step S42 specifically comprises: According to the ground point coordinates, the four nearest BP imaging points of the positioning point of the virtual RD image point are queried; According to the object point coordinates and image point coordinates of the four nearest BP imaging points, a linear function of the object point coordinates to the BP image point coordinates is fitted; According to the geometric positioning ground point coordinates of the virtual RD image point, the image point coordinates of the point on the BP image are calculated by using the fitted linear function, which is usually a non-integer value.

9. The method of reconstructing BP imaging SAR data into an RD imaging mode according to any one of claims 1-8, characterized in that, The step S5 comprises the following steps: Step S51, integrating all RD virtual imaging parameters according to the requirements of the standard RD data format, to generate an auxiliary data file; Step S52, encapsulating the auxiliary data file with the generated RD virtual imaging amplitude image and single-view complex image to form a complete data product; Step S53, adding necessary metadata to the data product, the metadata including data generation time, processing software version and coordinate system information; Step S54, converting the data product into a standard format, adding metadata, and then performing integrity check on the generated data product to ensure that all necessary data files and parameter files exist and are in correct format.

10. An apparatus for reconstructing BP imaging SAR data into an RD imaging mode, the apparatus comprising: a processor configured to: receive a BP imaging SAR data; and reconstruct the BP imaging SAR data into an RD imaging mode. It comprises: An orbit fitting module, configured to extract the imaging time and corresponding sensor position coordinates in the BP imaging data, and construct a continuous function of the sensor position changing with time by using a polynomial curve fitting method; A parameter statistical module, configured to calculate statistical characteristic values of the image distance slant range based on the geometric relationship between the ground coordinates of the BP image points and the sensor position coordinates; A parameter design module, configured to design a set of equivalent RD virtual imaging parameters based on the continuous function and the statistical characteristic values, the RD virtual imaging parameters including virtual imaging orbit, Doppler center frequency, initial slant range, imaging time range, image width and height, wavelength and sampling resolution; An orbit fitting module, configured to extract the imaging time and corresponding sensor position coordinates in the BP imaging data, and construct a continuous function of the sensor position changing with time by using a polynomial curve fitting method; A parameter statistical module, configured to calculate statistical characteristic values of the image distance slant range based on the geometric relationship between the ground coordinates of the BP image points and the sensor position coordinates; A parameter design module, configured to design a set of equivalent RD virtual imaging parameters based on the continuous function and the statistical characteristic values, the RD virtual imaging parameters including virtual imaging orbit, Doppler center frequency, initial slant range, imaging time range, image width and height, wavelength and sampling resolution; The data reconstruction module is configured to, for each pixel in the RD virtual imaging grid, calculate the corresponding ground point coordinate by geometric positioning, reversely map to the BP image space, obtain the amplitude value and complex value from the BP imaging data by an interpolation method, and generate the amplitude image and single-view complex image of the RD virtual imaging in combination with the orbit compensation phase. The auxiliary data generation module is configured to integrate the virtual imaging parameters into an auxiliary data file in a standard format, and jointly form a complete reconstruction data product with the amplitude image and single-view complex image of the RD virtual imaging.

Images