An on-board interferometric phase self-adapting compression method and device for an interferometric imaging radar altimeter

Abstract

This application discloses an on-board adaptive compression method and apparatus for interferometric imaging radar altimeter. The method includes: Step S1: Imaging the altimeter echo data to obtain a single-view complex image; Step S2: Image registration of the two single-view complex images, generating an interferogram through conjugate multiplication; Step S3: Extracting the interferometric phase from the interferogram to construct a normalized unit-modulus complex signal; Step S4: Performing scene-type-based adaptive sparse spectrum compression; Step S5: Quantizing and encoding the compressed sparse spectrum coefficients and their corresponding frequency domain position indices. This invention overcomes the limitations of resolution and downlink bandwidth, preserves resolution as much as possible while compressing data volume, improves reconstruction quality in complex scenes, possesses adaptive denoising capabilities, and has an algorithm architecture suitable for real-time on-board implementation.

Description

Technical Field

[0001] This invention belongs to the field of radar signal processing and satellite remote sensing technology, specifically relating to a data processing technology for a spaceborne wide-swath interferometric imaging radar altimeter, particularly an adaptive compression method and device for interferometric phase data that can adapt to different observation scenarios such as open water areas and the sea-land interface, with water bodies as the main observation object. Background Technology

[0002] Wide-swath interferometric imaging radar altimeter (InIRA) is a core instrument of the new generation of spaceborne microwave remote sensing. Leveraging near-nadir observation geometry and short-baseline interferometry, it achieves wide-swath, high-precision observations of sea surface height, sea surface height anomalies, and sub-mesoscale ocean dynamics. It has become a core tool for studying key processes such as global ocean circulation, sub-mesoscale eddies, and ocean energy cascades. Engineering practices, including the InIRA carried by my country's Tiangong-2 space station and the KaRIn payload of the international SWOT (Surface Water and Ocean Topography) satellite, demonstrate that this type of remote sensing equipment can simultaneously form observation points in the azimuth and range directions, achieving three-dimensional imaging of the sea surface and preserving richer texture and spectral information. This significantly overcomes the one-dimensional along-orbit sampling limitations of traditional nadir altimeters, providing two-dimensional high-resolution observational data support for ocean dynamics research and triggering a revolution in ocean observation technology.

[0003] However, while wide-swath interferometric imaging radar altimeters possess high-resolution observation capabilities, they also generate massive amounts of raw echo data. Taking the SWOT satellite as an example, its Ka-band radar interferometer (KaRIn) generates raw data rates exceeding 4Gb / s, while the bandwidth of the satellite-to-ground data transmission link is typically extremely limited. This contradiction between the enormous data rate and limited downlink bandwidth has become the main bottleneck restricting the all-weather, global coverage observation capabilities of this type of payload. To address this limitation, SWOT employs a "dual-stream" downlink strategy in its engineering: the high-rate (HR) link retains as much high-fidelity raw echo data as possible to support high-resolution ground processing, while the low-rate (LR) link covers a wider area, performing imaging and interferometric processing on-board before downlinking image data. Specifically, limited by downlink bandwidth, SWOT's LR link data uses a compression path of "non-focused imaging + multi-look averaging." The non-focused algorithm of Doppler beam sharpening (DBS) reduces the imaging resolution, and multi-look averaging further reduces the data volume, ultimately reducing the resolution of the main downlink data product to approximately 500m. While the on-board processing mechanism for LR link data can meet downlink bandwidth requirements, it causes irreversible loss of effective resolution: wave information at wavelengths below 500m is almost entirely lost, and coarse product resolution also leads to blurred coastlines and loss of coastline and nearshore wave details. To achieve higher resolution downlink interferometric phase maps within the existing downlink bandwidth, a phase compression method with a higher compression ratio is required.

[0004] Current interferometric data compression techniques have several shortcomings: First, compression primarily focuses on raw echoes or complex samples, resulting in low compression rates; direct compression techniques for spaceborne interferometric phase maps are still immature. Second, existing methods and traditional image processing techniques struggle to balance compression efficiency with reconstructed image resolution. Third, scene adaptability is insufficient; the primary observation targets of interferometric imaging radar altimeters are oceans and water bodies, while low-coherence land information will affect the compression efficiency of traditional image compression methods, leading to a reduction in compression rates. Summary of the Invention

[0005] The technical problem this invention aims to solve is to address the contradiction between limited downlink bandwidth and high-resolution observation requirements in on-board processing of existing wide-swath interferometric imaging radar altimeters. This invention provides an adaptive phase compression method for interferometric imaging altimeters. This method can effectively preserve high-resolution texture information of water bodies and geometric features at the land-sea interface at extremely low data transmission rates.

[0006] To address the aforementioned technical problems, this invention provides an on-board interferometric phase adaptive compression method for an interferometric imaging radar altimeter, characterized by comprising: Step S1: After imaging the altimeter echo data, obtain a single-view complex image; Step S2: Perform image registration on the two single-view complex images and generate an interferogram by conjugate multiplication; Step S3: Extract the interference phase from the interferogram and construct a normalized unit-modulus complex signal; Step S4: Perform scene-type-based adaptive sparse spectral compression; Step S5: Quantize and encode the compressed sparse spectral coefficients and their corresponding frequency domain position indices.

[0007] As one aspect of the above method, step S4 specifically includes: Step S4-1: Identify the observation scene type based on satellite ephemeris and pre-stored map information on the satellite; if the identification result indicates that the scene is open water, proceed to step S4-2; if the identification result indicates that the scene includes land, proceed to steps S4-3 to S4-5. Step S4-2: Directly perform frequency domain transformation on the unit modulus complex signal; then proceed to step S4-6; Step S4-3: Calculate the coherence coefficients based on the two single-view complex images: Step S4-4: Use the coherence coefficient to perform spatial domain mask preprocessing on the unit modulus complex signal; Step S4-5: Perform frequency domain transformation on the unit modulus complex signal preprocessed by the spatial domain mask; then proceed to step S4-6. Steps S4-6: Based on the preset compression ratio or threshold value of the spectral coefficients, retain the spectral coefficients that meet the conditions, and set the rest to zero.

[0008] As another embodiment of the above method, the characteristic is that, in step S4-3, the calculation is based on two-way single-view complex images. and The coherence coefficient is: ; In step S4-4, the obtained coherence coefficient is used to generate a mask.

[0009] As another embodiment of the above method, the feature is that step S4-4 utilizes the coherence coefficient to perform spatial domain mask preprocessing on the signal, specifically including the following sub-steps: Step S4-4-1: Generate a coherent binary mask; Set a coherence threshold According to the coherence coefficient Generate spatial mask The formula is as follows: ; in, The values ​​are pixel coordinates. A value of 1 represents a highly coherent effective water body observation target area, while a value of 0 represents a low-coherence non-target land area or noise area. Step S4-4-2: Perform spatial mask weighting; Using the space mask For unit modulus complex signals Perform dot product weighting: ; in This is the preprocessed unit modulus complex signal.

[0010] As another embodiment of the above method, the feature is that, in the above steps S4-6: sparse spectral coefficients are screened according to amplitude sorting or energy threshold.

[0011] In another embodiment of the above method, the amplitude spectrum of the spectrum is calculated and sorted in steps S4-6, and the number of reserved points is determined based on the downlink bandwidth. Only retain the largest amplitude. One coefficient is set to zero, and the rest are set to zero to obtain the final sparse spectrum.

[0012] As a further embodiment of the above method, step S3 further includes performing phase deflating processing on the interferogram, extracting the interference phase from the deflated interferogram, and constructing a normalized unit-modulus complex signal. : ; in, It is the imaginary unit.

[0013] As a further embodiment of the above method, the single-view complex image in step S1 is a single-view complex image that retains the original spatial resolution and phase information.

[0014] To achieve the above objectives, the present invention also provides an on-board interferometric phase adaptive compression device for an interferometric imaging radar altimeter, used to implement the above-mentioned interferometric phase adaptive compression method, characterized in that it includes: The data acquisition module is used to process the altimeter echo data into images to obtain a single-view complex image. The interferogram generation module is used to perform image registration on two single-view complex images and generate an interferogram through conjugate multiplication. The unit modulus complex signal module is used to extract the interference phase from the interferogram and construct a normalized unit modulus complex signal; The spectrum compression module is used to perform scene-type-based adaptive sparse spectrum compression; The encoding module is used to quantize and encode the compressed sparse spectral coefficients and their corresponding frequency domain position indices.

[0015] The present invention also provides an on-board interferometric phase adaptive compression device for an interferometric imaging radar altimeter, used to implement the interferometric phase adaptive compression method, characterized in that it includes: Memory, used to store data and computer programs; A processor for executing the computer program to implement the steps of the interferometric phase adaptive compression method.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. Breaking through the constraints of resolution and downlink bandwidth: This invention abandons the existing technology of reducing data volume by reducing resolution through "multi-view averaging", and retains resolution as much as possible while compressing data volume.

[0017] 2. Improved Reconstruction Quality in Complex Scenes: For non-stationary scenes such as the land-sea interface, the coherence masking strategy proposed in this invention effectively solves the compression difficulties caused by low coherence regions in traditional transform domain methods. Experiments show that this method can clearly reconstruct complex coastline geometries and avoid edge blurring.

[0018] 3. Adaptive denoising function: This invention utilizes the sparsity of the wave interferometric phase map signal in the frequency domain and the broadband characteristics of noise, and automatically filters out broadband background noise through an amplitude spectrum coefficient selection mechanism. This enables signal denoising while performing data compression, improving the signal-to-noise ratio of the final retrieved sea surface height.

[0019] 4. The algorithm architecture is suitable for real-time implementation on satellite: The FFT (Fast Fourier Transform) and threshold mask operations involved in this invention are all standard linear or logical operations, requiring no complex iterative process and not including interpolation operations, making them suitable for real-time operation in satellite embedded processors such as FPGAs or DSPs. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall process of the on-board interferometric phase map compression method for interferometric imaging altimeter provided in this embodiment of the invention; Figure 2 This is a schematic diagram of the interferometric measurement geometry of the wide-swath interferometric imaging radar altimeter according to an embodiment of the present invention; Figure 3 This is a schematic diagram comparing the two-dimensional spectral characteristics of complex domain interferometric phase under two typical observation scenarios (open water scenario and land-sea boundary scenario) obtained from the analysis of embodiments of the present invention; Figure 4This is a schematic diagram comparing the phase reconstruction effect with the original phase map in an open water scenario with an extremely low data retention ratio of 0.05% according to an embodiment of the present invention. Figure 5 This is a comparative schematic diagram of the original phase map (top), multi-view average (left), and phase compression reconstruction effect (right) using the coherence mask compression method of this invention in a land-sea interface scenario according to an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions provided in this application are further illustrated below with reference to the embodiments.

[0022] This invention provides a method for compressing on-board interferometric phase image data of an interferometric imaging radar altimeter. This method operates on interferometric phase images in the complex domain, which can be obtained by interferometric processing of registered single-view complex (SLC) data and removing terrain features. By constructing a sparse signal model in the complex domain and combining it with a scene adaptation strategy, this method can achieve high-fidelity reconstruction of high-resolution sea surface textures and complex coastline features with extremely low data transmission bandwidth.

[0023] like Figure 1 As shown in the figure, a satellite-based interferometric phase adaptive compression method for an interferometric imaging radar altimeter is provided by a specific embodiment of the present invention. The method includes the following steps: Step S1: After imaging the altimeter echo data, obtain a Single-Look Complex (SLC) image.

[0024] The system receives raw echo data from the interferometric imaging radar altimeter, performs high-resolution imaging processing in the onboard processing unit, and generates two single-look complex images that retain the original spatial resolution and phase information. and The high-resolution imaging processing described herein is an imaging process that maintains the original single-view spatial resolution.

[0025] Step S2: Generate an interferogram.

[0026] For the two single-view complex images and Image registration is performed, and an interferogram is generated through conjugate multiplication. Optionally, phase deflating processing can then be performed on the interferogram to reduce spectral spread caused by flat phase and improve the sparsity of subsequent frequency domain signals.

[0027] The interferogram The calculation formula is: ; where * denotes the complex conjugate operation.

[0028] Step S3: Construct a unit modulus complex domain phase signal; Extracting the interference phase from the interferogram after leveling. To eliminate the discontinuities caused by real phase entanglement and adapt to frequency domain transformations, the interference phase is mapped onto the unit circle in the complex domain to construct a normalized unit modulus complex signal. The unit modulus complex signal The construction can be modeled as follows: in, The imaginary unit is used. This mapping transforms the periodically wrapped phase of the real phase into a continuous complex signal, providing a foundation for subsequent frequency domain sparse analysis.

[0029] Step S4: Perform scene-type-based adaptive sparse spectral compression; Identify scene types; for scenes containing land (i.e., land-sea boundary scenes), calculate based on two single-view complex images (…). and The coherence coefficient of ) ; The obtained coherence coefficient is used to generate a mask for subsequent extraction of the effective water phase signal, followed by the effective unit modulus complex signal. Perform two-dimensional frequency domain transformation and extract frequency domain sparse coefficients.

[0030] If the scene identification result indicates that the scene is an open water area, then the unit modulus complex signal is directly processed. Perform two-dimensional frequency domain transformation and extract frequency domain sparse coefficients.

[0031] Specifically, the steps include the following: Step S4-1: Identify the observation scene type based on satellite ephemeris and on-board pre-stored map information; if the identification result indicates that the scene is open water, proceed to step S4-2; if the identification result indicates that the scene includes land (i.e., the scene where land and sea meet), proceed to steps S4-3 to S4-5. It should be noted that this scene recognition step can be performed in advance, independently of the interference processing.

[0032] Step S4-2, directly process the unit modulus complex signal Perform a two-dimensional frequency domain transformation (two-dimensional discrete Fourier transform or fast Fourier transform) for frequency domain sparse coefficient extraction; then proceed to steps S4-6. Step S4-3, trigger based on two single-view complex images ( and Calculation of coherence coefficient: ; Step S4-4: Perform spatial domain masking preprocessing on the signal using coherence characteristics; The signal is preprocessed using spatial domain masking based on coherence characteristics to suppress spectral diffusion in low coherence noise regions. Step S4-5: Perform a two-dimensional discrete Fourier transform or fast Fourier transform on the signal preprocessed by the spatial domain mask to convert the signal to the frequency domain for extracting frequency domain sparse coefficients; then proceed to step S4-6. Steps S4-6: Based on the preset compression ratio, bandwidth limit, or threshold for the size of the spectral coefficients, retain the main spectral coefficients that meet the conditions, and set the rest to zero, thereby truncating the low-amplitude background noise coefficients.

[0033] Furthermore, as a preferred embodiment of the present invention, the spatial domain mask preprocessing of the signal using coherence characteristics described in step S4-4 specifically includes the following sub-steps: Sub-step S4-4-1: Generate a coherence binary mask (for land-sea interface scenarios). Set a coherence threshold. According to the complex coherence coefficient Generate spatial mask The formula is as follows: ; in, The values ​​are pixel coordinates. A value of 1 represents a highly coherent, effective water body observation target area, while a value of 0 represents a low-coherence, non-target land area or noise area.

[0034] Sub-step S4-4-2: Perform spatial mask weighting (for land-sea boundary scenarios). Utilize the spatial mask... For unit modulus complex signals Perform dot product weighting to obtain the preprocessed signal. : ; By using masking operations in this spatial domain, the signal in low-coherence regions is forcibly set to zero to prevent energy leakage in low-coherence regions such as land from submerging the weak sea surface gravity wave signal during subsequent frequency domain transformation.

[0035] Two-dimensional frequency domain transformation: Perform a two-dimensional Fourier transform on the signal to convert it to the frequency domain. The input signal for the transformation is adaptively selected according to the scene type: if it is a land-sea boundary scene, then the weighted signal output in step S4-4-2 is used. Perform the transformation to obtain the mask spectrum. For open water scenes, directly apply the original unit analog complex signal. Perform the transformation to obtain the original spectrum. Taking a scenario at the boundary between land and sea as an example, the transformation formula is expressed as: ; in, Represents a two-dimensional Fourier transform. These are frequency domain coordinates.

[0036] Furthermore, as a preferred embodiment, in steps S4-6 above: sparse spectral coefficients are screened based on amplitude sorting or energy threshold.

[0037] Calculate and sort the amplitude spectrum of the spectrum in steps S4-6, and determine the number of reserved points based on the downlink bandwidth. Only retain the largest amplitude. Set one coefficient to zero and the rest to zero to obtain the final sparse spectrum. .

[0038] Taking a land-sea boundary scenario as an example, the truncation operation is represented as: ; If the scene is an open body of water, then in the above formula... Replace with the original spectrum .

[0039] Step S5: Quantization encoding and data download.

[0040] The retained sparse spectral coefficients and their corresponding frequency domain position indices are quantized and encoded to form compressed data packets, which are then transmitted via the satellite downlink.

[0041] To verify the compression performance of the method of the present invention in high-resolution wide-swath observation, the high code rate single-view complex product (L1B_HR_SLC) of the SWOT satellite Ka-band radar interferometer (KaRIn) is used as the input data source in the following embodiments.

[0042] Please see Figure 1 This is a schematic diagram illustrating the overall process of the on-board interferometric phase map compression method for interferometric imaging altimeters provided in this embodiment of the invention. The method mainly includes the following steps: Step S1: Acquire a single-view complex (SLC) image; In this embodiment, the input data is the standard scientific product of the SWOT mission, Level 1B High Rate Single-Look Complex (L1B_HR_SLC). This data is generated by the ground system using a high-phase-preserving back-projection (BP) algorithm, possessing high spatial resolution (approximately 0.75 meters in the slant range direction and approximately 5 meters in the azimuth direction), and can preserve the high-frequency phase texture and speckle statistics of interferometry to the greatest extent. To simulate the actual data flow processed on-board and eliminate invalid observations, this embodiment performs range cropping on the original SLC image, removing echoes near the zero angle of incidence (Nadir) region, and retaining only the small angle of incidence observation swaths (incident angle approximately...). Valid data. The cropped data size is 2000. 22826 pixels (for Example 1) and 2000 22948 pixels (for Example 2).

[0043] Step S2: Interference processing and signal feature extraction See Figure 2 This demonstrates the measurement geometry of a wide-swath interferometric imaging radar altimeter. The satellite platform utilizes two spatially separated antennas. and (Baseline length is) The baseline inclination angle is Simultaneously receive data from sea surface points. The echo. Using the registered two single-view complex images. and Interference maps are generated through conjugate multiplication. ; ; according to Figure 2 The geometric relationship shown, target point height Interference phase before going to flat ground The solution is obtained. Among them, the interference phase error... Transmitted to high inversion error Sensitivity relationship: ; in, For radar wavelength, For antenna Slope distance to the target Baseline tilt angle, For local incident angle, This represents the effective vertical baseline length. This relationship indicates the inversion error of sea level height. With interferometric phase error Proportional.

[0044] Due to the input data in this embodiment ( and The image is generated based on the back projection algorithm, which accurately projects the echo data onto the reference plane during the imaging process. The flat phase in the interferogram obtained after the conjugate multiplication of the two single-view complex images has been basically eliminated. Therefore, the interferogram can be regarded as having completed the flat phase removal process.

[0045] Step S3: Complex Domain Phase Mapping Due to the observed real interference phase exist The periodic winding of the signal introduces nonlinear edge high-frequency noise when directly frequency-domain transformed. This embodiment constructs a normalized unit-modulus complex signal by mapping the phase onto the unit circle in the complex domain. : ; This step transforms the nonlinear phase transition into a continuous signal in the complex plane, ensuring that the sparsity of the signal in the frequency domain can be correctly represented.

[0046] Step S4: Scene-Adaptive Sparse Compression See Figure 3 Spectral analysis based on measured data shows that the interferometric phase exhibits drastically different spectral sparsity characteristics under different observation scenarios: the energy in the open ocean scenario is highly concentrated in the low frequency (left figure), while the energy in the land-sea boundary scenario is severely diffused due to land noise and the approximate step signal caused by the coastline (right figure). Therefore, this embodiment of the invention implements an adaptive processing strategy based on prior knowledge: firstly, based on the satellite's real-time ephemeris and pre-stored coarse-resolution map information, the scenario type of the current observation area is quickly identified. This identification process does not rely on radar echo data, so it can be processed in advance or in parallel. Based on the identification results, the system automatically selects to enter the "open water processing path" or the "land-sea boundary processing path".

[0047] Example 1: Open water scene This embodiment uses SWOT data A from the eastern waters of the Labrador Peninsula for verification. This area is mainly dominated by small-scale ocean waves (wavelength approximately 220m), exhibiting highly coherent water textures.

[0048] Scene recognition: The ephemeris and map matching results show that there is no land within the observation area, and it is determined to be an "open water scene".

[0049] Processing flow: The system skips coherence coefficient calculation and masking operations, directly calling sub-step S4-2. Specific operation: For unit-modulus complex signals... Perform a two-dimensional fast Fourier transform (2D-FFT) to obtain the spectrum. .

[0050] Coefficient truncation: Set the data retention ratio based on extremely low bandwidth requirements. This embodiment tested from... arrive The retention ratios of multiple groups. Under the extreme compression conditions, that is, only the largest spectral amplitude is retained. One coefficient, the rest Setting the coefficients to zero yields the sparse spectrum. .

[0051] Effect analysis: See Figure 4 . Figure 4 The left image is the original phase map, where the sea surface texture is obscured by background speckle noise. Figure 4 The right figure is a compressed reconstruction diagram of the present invention. Given the lack of an absolute true value for sea surface height, this embodiment uses the well-validated results of traditional multi-view processing as a large-scale benchmark for cross-validation. Calculations show that, under the extreme compression condition of a data retention ratio of only 0.05%, the average deviation between the reconstructed elevation data and the multi-view results at a spatial scale of 5 km is on the order of millimeters. This error level meets the accuracy requirements for sea surface height in sub-mesoscale marine dynamic observations, proving that this method does not introduce destructive large-scale distortions at high compression ratios.

[0052] Experimental conclusion: Despite the extremely low retention ratio, the reconstructed phase map still clearly retains the dominant wave stripe features. This is because the wave signal is highly sparse in the frequency domain, while background random noise is distributed across the entire frequency band. By selecting coefficients in a large frequency domain, background noise is effectively filtered out automatically, achieving 2000-fold data compression while improving the signal-to-noise ratio of the image.

[0053] Example 2: Land-sea boundary scenario This embodiment uses SWOT data B from the northern Pacific coast of Peru for verification. This region includes a low-coherence land surface, complex coastline geometry, and nearshore waves.

[0054] Scene recognition: The ephemeris and map matching results show that the observation area contains land, which is determined to be a "land-sea boundary scene", triggering the coherence processing mechanism.

[0055] Processing flow: Execute sub-steps S4-3 to S4-6 sequentially.

[0056] Mask generation (S4-3): Call step S1 to acquire two single-view complex images that participate in the interferometric processing in step S2. Calculate the complex coherence coefficient The average sliding window size for the coherence coefficient is 6 × 30 (range × azimuth). Based on the difference in coherence between land and sea, the coherence threshold is... Set to 0.9. Generate a spatial binary mask. ,Will A mark of 0 represents a land area. The mark 1 represents a water area.

[0057] Mask weighting (S4-4): Performs dot product operation in the spatial domain This step strongly suppresses strong noise generated in low-coherence regions such as land, preventing it from leaking energy and drowning out the weak nearshore wave signal during subsequent frequency domain transformation.

[0058] Frequency domain transformation (S4-5): For the preprocessed signal Perform a two-dimensional Fourier transform to obtain the mask spectrum. Because the masking operation effectively removes broadband noise, the signal still exhibits sparsity in the frequency domain.

[0059] Coefficient cutoff (S4-6): Set the retention ratio Only keep The top 1% of coefficients with the largest amplitudes yield the sparse spectrum. .

[0060] Effect analysis: Please refer to Figure 5 The diagram shows a comparison of the local reconstruction performance of the multi-view averaging method (left) and the method of this invention (right) at comparable data compression ratios. Calculations show that, compared to the multi-view processing results, the additional error introduced by compression is on the order of millimeters at a spatial scale of 5 km and does not dominate the overall error budget.

[0061] Existing technical limitations: Multi-view averaging leads to reduced resolution, erases wave details, produces a jagged coastline, and fails to distinguish narrow river paths flowing into the sea. The publicly available SWOT_LR product manual (.pdf) demonstrates the SWOT LR processing method: defocused multi-view, resolution of 500m, and sampling interval of 250m.

[0062] Advantages of this invention: At the desired retention ratio, the image reconstructed by the method of this invention not only clearly preserves the wave texture near the shore but also accurately restores the complex coastline contours and the main path of inland rivers. This demonstrates that mask preprocessing effectively solves the signal spectrum diffusion problem in non-stationary scenes, achieving a high compression ratio.

[0063] Step S5: Quantization Encoding and Transmission The sparse spectral coefficients (real and imaginary parts) selected in step S4 and their corresponding two-dimensional frequency indices are quantized. Based on the statistical distribution characteristics of the coefficients, the data stream is further compressed, packaged into the final downlink data frame, and transmitted.

[0064] The specific embodiments of the present invention may also include some extended solutions: 1. The frequency domain transformations in steps S4-2 and S4-5 include not only Fourier transforms but also DCT transforms combining the real and imaginary parts. The focus is on implementing the frequency domain transformation.

[0065] 2. The masking scheme in step S4-4-1 can also consider using a signal-to-noise ratio scheme. The key is in mask calculation to eliminate land areas.

[0066] 3. The key to coefficient truncation in steps S4-6 is selecting the main spectral coefficients with concentrated energy. The top K coefficients with the largest amplitude and energy can be selected, or a set coefficient threshold can be used to truncate low-amplitude background noise coefficients. Alternatively, the main spectral coefficients with concentrated energy can be selected based on the distribution characteristics of the spectral coefficient magnitudes. All these methods for selecting the main spectral coefficients should be covered within the spirit and scope of the technical solution of this invention.

[0067] The present invention also provides an on-board interferometric phase adaptive compression device for an interferometric imaging radar altimeter, used to implement the above-mentioned interferometric phase adaptive compression method, characterized in that it includes: The data acquisition module is used to process the altimeter echo data into images to obtain a single-view complex image. The interferogram generation module is used to perform image registration on two single-view complex images and generate an interferogram through conjugate multiplication. The unit modulus complex signal module is used to extract the interference phase from the interferogram and construct a normalized unit modulus complex signal; The spectrum compression module is used to perform scene-type-based adaptive sparse spectrum compression; The encoding module is used to quantize and encode the compressed sparse spectral coefficients and their corresponding frequency domain position indices.

[0068] This invention also provides an on-board interferometric phase adaptive compression device for an interferometric imaging radar altimeter, used to implement the interferometric phase adaptive compression method, comprising: Memory is used to store data and computer programs; memory may include volatile memory and / or permanent memory. A processor is used to execute the computer program to implement the steps of the described interferometric phase adaptive compression method. The processor includes a general-purpose central processing unit and / or a special-purpose processor or controller capable of performing the above functions.

[0069] This invention addresses the shortcomings of current interferometric data compression techniques: First, compression primarily focuses on raw echoes or complex samples, resulting in low compression rates; direct compression techniques for spaceborne interferometric phase maps are still immature. Second, existing methods and traditional image processing techniques struggle to balance compression efficiency with reconstructed image resolution; this invention utilizes the frequency domain sparsity of phase signals to solve this problem. Third, scene adaptability is insufficient; the primary observation targets of interferometric imaging radar altimeters are oceans and water bodies, while low-coherence land information will affect the compression efficiency of traditional image compression methods, leading to a reduction in compression rates. The technical solution provided by the above-described embodiments overcomes the aforementioned technical problems of the prior art and achieves significant technical effects.

[0070] In summary, the embodiments of this invention demonstrate that by performing adaptive sparsity processing on high-resolution data in the complex domain, it can still meet the stringent requirements for texture detail and geometry in sub-mesoscale ocean dynamics observations, even with retention ratios as low as 0.05% (open water scenarios) and 1% (land-sea boundary scenarios). The method proposed in this invention is applicable to different imaging systems and is suitable for onboard real-time data processing of next-generation wide-swath interferometric imaging altimeters.

[0071] Compared with the prior art, the present invention has the following beneficial effects: 1. Breaking through the constraints of resolution and downlink bandwidth: This invention abandons the existing technology of reducing data volume by reducing resolution through "multi-view averaging", and retains resolution as much as possible while compressing data volume.

[0072] 2. Improved Reconstruction Quality in Complex Scenes: For non-stationary scenes such as the land-sea interface, the coherence masking strategy proposed in this invention effectively solves the compression difficulties caused by low coherence regions in traditional transform domain methods. Experiments show that this method can clearly reconstruct complex coastline geometries and avoid edge blurring.

[0073] 3. Adaptive denoising function: This invention utilizes the sparsity of the wave interferometric phase map signal in the frequency domain and the broadband characteristics of noise, and automatically filters out broadband background noise through an amplitude spectrum coefficient selection mechanism. This enables signal denoising while performing data compression, improving the signal-to-noise ratio of the final retrieved sea surface height.

[0074] 4. The algorithm architecture is suitable for real-time implementation on satellite: The FFT (Fast Fourier Transform) and threshold mask operations involved in this invention are all standard linear or logical operations, requiring no complex iterative process and not including interpolation operations, making them suitable for real-time operation in satellite embedded processors such as FPGAs or DSPs. Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the embodiments, those skilled in the art should understand that modifications or equivalent substitutions to the technical solutions of the present invention do not depart from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An on-board interferometric phase adaptive compression method for an interferometric imaging radar altimeter, characterized in that, include: Step S1: After imaging the altimeter echo data, obtain a single-view complex image; Step S2: Perform image registration on the two single-view complex images and generate an interferogram by conjugate multiplication; Step S3: Extract the interference phase from the interferogram and construct a normalized unit-modulus complex signal; Step S4: Perform scene-type-based adaptive sparse spectral compression; Step S5: Quantize and encode the compressed sparse spectral coefficients and their corresponding frequency domain position indices.

2. The interference phase adaptive compression method according to claim 1, characterized in that, Step S4 specifically includes: Step S4-1: Identify the observation scene type based on satellite ephemeris and pre-stored map information on the satellite; if the identification result indicates that the scene is open water, proceed to step S4-2; if the identification result indicates that the scene includes land, proceed to steps S4-3 to S4-5. Step S4-2: Directly perform frequency domain transformation on the unit modulus complex signal; then proceed to step S4-6; Step S4-3: Calculate the coherence coefficients based on the two single-view complex images: Step S4-4: Use the coherence coefficient to perform spatial domain mask preprocessing on the unit modulus complex signal; Step S4-5: Perform frequency domain transformation on the unit modulus complex signal preprocessed by the spatial domain mask; then proceed to step S4-6. Steps S4-6: Based on the preset compression ratio or threshold value of the spectral coefficients, retain the spectral coefficients that meet the conditions, and set the rest to zero.

3. The interference phase adaptive compression method according to claim 2, characterized in that, In step S4-3, the calculation is based on two-way single-view complex images. and The coherence coefficient is: ； In step S4-4, a mask is generated using the obtained coherence coefficient.

4. The interferometric phase adaptive compression method according to claim 3, characterized in that, Step S4-4 uses the coherence coefficient to perform spatial domain mask preprocessing on the signal, which specifically includes the following sub-steps: Step S4-4-1: Generate a coherent binary mask; Set a coherence threshold According to the coherence coefficient Generate spatial mask The formula is as follows: ； in, The values ​​are pixel coordinates. A value of 1 represents a highly coherent effective water body observation target area, while a value of 0 represents a low-coherence non-target land area or noise area. Step S4-4-2: Perform spatial mask weighting; Using the space mask For unit modulus complex signals Perform dot product weighting: ； in This is the preprocessed unit modulus complex signal.

5. The interference phase adaptive compression method according to claim 2, characterized in that, In steps S4-6 above: sparse spectral coefficients are screened based on amplitude sorting or energy threshold.

6. The interferometric phase adaptive compression method according to claim 2, characterized in that, In steps S4-6, the amplitude spectrum of the spectrum is calculated and sorted, and the number of reserved points is determined based on the downlink bandwidth. Only retain the largest amplitude. One coefficient is set to zero, and the rest are set to zero to obtain the final sparse spectrum.

7. The interferometric phase adaptive compression method according to claim 1, characterized in that, Step S3 also includes performing phase deflating on the interferogram, extracting the interference phase from the deflated interferogram, and constructing a normalized unit-modulus complex signal. : ； in, It is the imaginary unit.

8. The interferometric phase adaptive compression method according to claim 1, characterized in that, In step S1, the single-view complex image is a single-view complex image that retains the original spatial resolution and phase information.

9. An on-board interferometric phase adaptive compression device for an interferometric imaging radar altimeter, used to implement the interferometric phase adaptive compression method according to any one of claims 1-8, characterized in that, include: The data acquisition module is used to process the altimeter echo data into images to obtain a single-view complex image. The interferogram generation module is used to perform image registration on two single-view complex images and generate an interferogram through conjugate multiplication. The unit modulus complex signal module is used to extract the interference phase from the interferogram and construct a normalized unit modulus complex signal; The spectrum compression module is used to perform scene-type-based adaptive sparse spectrum compression; The encoding module is used to quantize and encode the compressed sparse spectral coefficients and their corresponding frequency domain position indices.

10. An on-board interferometric phase adaptive compression device for an interferometric imaging radar altimeter, used to implement the interferometric phase adaptive compression method according to any one of claims 1-8, characterized in that, include: Memory, used to store data and computer programs; A processor for executing the computer program to implement the steps of the interferometric phase adaptive compression method.

Images