A multi-source remote sensing satellite navigation data processing method and system

Abstract

The present application relates to the technical field of remote sensing satellite navigation data processing, and specifically discloses a multi-source remote sensing satellite navigation data processing method and system, wherein a plurality of miniature inertial measurement units are arranged around the imaging area of a remote sensing satellite optical imager to obtain local vibration information at different positions around the imaging area. After filtering, the local vibration information separates high-frequency disturbance components that cannot be captured by a conventional attitude measurement system. Combined with the spatial distribution information of the miniature inertial measurement units and a spatial difference algorithm, a fine attitude disturbance field map is constructed, which contains the attitude disturbance experienced by each pixel point on the imaging area. Finally, the attitude disturbance field map is input into a remote sensing image to obtain the corresponding attitude disturbance and perform compensation and correction, thereby solving the problem of weak and non-periodic residual vibration of the attitude control mechanism of the remote sensing satellite, which leads to the introduction of difficult-to-detect local deformation in the geometric processing of the remote sensing image.

Description

Technical Field

[0001] This application relates to the field of remote sensing satellite navigation data processing technology, and more specifically, to a multi-source remote sensing satellite navigation data processing method and system. Background Technology

[0002] When a remote sensing satellite is in orbit, its attitude control mechanisms, such as momentum wheels or reaction wheels, generate very small residual vibrations after adjusting its attitude. These momentum wheels change the satellite's angular momentum through high-speed rotation, thereby adjusting the satellite's attitude. During acceleration, deceleration, or reversal, even the most sophisticated momentum wheels cannot completely avoid generating weak mechanical vibrations. These high-frequency vibrations are transmitted to the onboard optical imager through the satellite's structure. Although the satellite's attitude measurement system (such as star sensors and gyroscopes) can accurately measure and compensate for overall attitude changes, these extremely weak and high-frequency residual vibrations may exceed the frequency range that the attitude measurement system can effectively handle, or be below its detection threshold, and therefore cannot be completely filtered out. This means that the recorded instantaneous satellite attitude information will contain a weak, non-periodic perturbation component. While this perturbation has little impact on the overall quality of a single remote sensing image, it can introduce imperceptible local deformations in the image's geometric processing. For example, if the sensor platform experiences a slight vibration at the moment of exposure, it can cause an extremely small deflection in the viewing direction of different pixels in the image, resulting in subtle geometric distortion within the image.

[0003] There is currently no effective technical solution to the above problems. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for processing multi-source remote sensing satellite navigation data, which aims to solve the problem that weak perturbations in the attitude information of remote sensing images lead to imperceptible local deformations in image geometric processing.

[0005] To achieve the above objectives, the technical solution of this application is as follows:

[0006] As one aspect of this application, a multi-source remote sensing satellite navigation data processing method is provided, which compensates and corrects remote sensing images by employing multiple miniature inertial measurement units positioned around the imaging area of ​​a remote sensing satellite optical imager. The method includes the following steps:

[0007] Multiple miniature inertial measurement units are used to acquire local vibration information at different locations around the imaging area;

[0008] The acquired local vibration information is filtered to separate the high-frequency disturbance component from the local vibration information;

[0009] The spatial distribution information of multiple micro inertial measurement units around the imaging area is obtained, and an attitude perturbation field map is constructed by using a spatial difference algorithm and combining it with high-frequency perturbation components. The attitude perturbation field map contains the attitude perturbation amount that each pixel in the imaging area needs to experience.

[0010] The attitude perturbation field map is input into the remote sensing image that needs geometric correction. The corresponding attitude perturbation amount is obtained according to the position area of ​​the pixels in the imaging area of ​​the remote sensing image, and the remote sensing image is compensated and corrected based on the attitude perturbation amount.

[0011] Furthermore, the step of acquiring local vibration information at different locations around the imaging area using multiple miniature inertial measurement units also includes:

[0012] A pulse signal is provided by the receiver of a remote sensing satellite. This pulse signal is used to synchronize the time reference of all micro inertial measurement units, so that the time period for each micro inertial measurement unit to acquire local vibration information is consistent.

[0013] Furthermore, the step of filtering the acquired local vibration information to separate the high-frequency disturbance component from the local vibration information includes:

[0014] The acquired local vibration information is preprocessed to remove zero bias and noise.

[0015] The preprocessed local vibration information is filtered to remove low-frequency disturbance components caused by changes in the overall attitude of the remote sensing satellite, and high-frequency disturbance components are separated from the local vibration information.

[0016] Furthermore, the step of filtering the acquired local vibration information to separate the high-frequency disturbance component from the local vibration information specifically includes:

[0017] A corresponding micro motion sensor and a micro interface health sensor are deployed near the location of each micro inertial measurement unit in the imaging area. The motion data of the corresponding micro inertial measurement unit is obtained by using the micro motion sensor, and the physical connection status data between the corresponding micro inertial measurement unit and the imaging area is obtained by using the micro interface health sensor.

[0018] The motion data of the micro motion sensor and the physical connection status data of the micro interface health sensor are correlated and analyzed to identify and quantify the pseudo motion signals caused by the tiny sliding displacement of the corresponding micro inertial measurement unit.

[0019] The quantized false motion signals are removed from the local vibration information at different locations around the imaging area obtained by multiple miniature inertial measurement units to obtain the true local vibration information.

[0020] The actual local vibration information in the imaging area is filtered and the corresponding high-frequency disturbance components are separated.

[0021] Furthermore, the step of acquiring the spatial distribution information of multiple miniature inertial measurement units around the imaging area and constructing an attitude perturbation field map by combining high-frequency perturbation components includes:

[0022] Acquire the spatial distribution information of each micro inertial measurement unit around the imaging area and the high-frequency disturbance components corresponding to each micro inertial measurement unit;

[0023] The attitude perturbation of any pixel in the imaging area is calculated using the thin plate spline interpolation algorithm.

[0024] An attitude perturbation field map is constructed based on the attitude perturbation of any pixel in the imaging region.

[0025] Furthermore, the step of inputting the attitude perturbation field map into the remote sensing image that needs geometric correction, obtaining the corresponding attitude perturbation amount based on the position region of the pixels in the imaging area, and compensating and correcting the remote sensing image based on the attitude perturbation amount specifically includes:

[0026] Acquire remote sensing images that require geometric correction;

[0027] Based on the position of each pixel in the remote sensing image to be processed in the imaging area, the attitude perturbation amount corresponding to the position of each pixel in the remote sensing image to be processed in the imaging area is obtained from the attitude perturbation field map.

[0028] The corresponding attitude disturbances are superimposed and compensated into the overall attitude parameters of the remote sensing satellite.

[0029] Furthermore, after the steps of inputting the attitude perturbation field map into the remote sensing image requiring geometric correction, obtaining the corresponding attitude perturbation amount based on the position region of the pixels in the imaging area, and compensating and correcting the remote sensing image based on the attitude perturbation amount, the method further includes:

[0030] Obtain the exposure time window for each pixel of the compensated and corrected remote sensing image;

[0031] High-frequency perturbation components corresponding to each pixel of the remote sensing image within the exposure time window are extracted from the local vibration information at different locations around the imaging area obtained from each micro inertial measurement unit.

[0032] Based on the high-frequency perturbation components of each pixel in the remote sensing image at its exposure time point and their spatial distribution in the imaging area, a sequence of local attitude perturbation field maps is constructed in which the exposure time windows of each pixel in the remote sensing image are continuously ordered.

[0033] Based on the position of each pixel in the remote sensing image on the imaging area and its corresponding exposure time point, the corresponding attitude perturbation amount is queried from the local attitude perturbation field map sequence.

[0034] The queried attitude perturbation is used as the input for geometric correction of the pixels in the remote sensing image, and a second geometric correction is performed on the remote sensing image based on the attitude perturbation.

[0035] Furthermore, the step of extracting the high-frequency perturbation component corresponding to each pixel of the remote sensing image during exposure from the local vibration information at different locations around the imaging area obtained from each micro inertial measurement unit includes:

[0036] Dynamically adjust the effective exposure time window length for each pixel of the remote sensing image;

[0037] Extract high-frequency disturbance components corresponding to the adjusted effective exposure time window from each local vibration information;

[0038] Calculate the weighted average of the high-frequency disturbance components within the adjusted effective exposure time window, where the weights are allocated based on the rate of change and amplitude of the high-frequency disturbance components within the adjusted effective exposure time window.

[0039] Furthermore, the step of dynamically adjusting the effective exposure time window length for each pixel of the remote sensing image includes:

[0040] Real-time monitoring of instantaneous frequency and amplitude change rate in local vibration information acquired by miniature inertial measurement units;

[0041] Based on the instantaneous frequency and amplitude change rate in the local vibration information obtained by the miniature inertial measurement unit, the effective exposure time window length of each pixel in the remote sensing image is calculated and adjusted.

[0042] As a second aspect of this application, a multi-source remote sensing satellite navigation data processing system is provided, further comprising:

[0043] A vibration information acquisition module is used to acquire local vibration information at different locations around the imaging area using multiple miniature inertial measurement units.

[0044] A disturbance component separation and extraction module is used to filter the acquired local vibration information and separate high-frequency disturbance components from the local vibration information.

[0045] The perturbation field map construction module is used to acquire the spatial distribution information of multiple micro inertial measurement units around the imaging area, and constructs the attitude perturbation field map by using a spatial difference algorithm and combining it with high-frequency perturbation components. The attitude perturbation field map contains the attitude perturbation amount that each pixel in the imaging area needs to experience.

[0046] The compensation and correction module is used to input the attitude disturbance field map into the remote sensing image that needs to be geometrically corrected, obtain the corresponding attitude disturbance amount according to the position area of ​​the pixels of the remote sensing image on the imaging area, and perform compensation and correction on the remote sensing image based on the attitude disturbance amount.

[0047] As described above, the multi-source remote sensing satellite navigation data processing method and system provided in this application can acquire local vibration information at different locations around the imaging area by setting up multiple miniature inertial measurement units around the imaging area of ​​the remote sensing satellite optical imager. After filtering, these local vibration information can separate high-frequency disturbance components that are difficult for traditional attitude measurement systems to capture. Combining the spatial distribution information of the miniature inertial measurement units and the spatial difference algorithm, this application can construct a fine attitude disturbance field map, which contains the attitude disturbance amount that each pixel in the imaging area needs to experience. Finally, the attitude disturbance field map is input into the remote sensing image, and the corresponding attitude disturbance amount is obtained according to the pixel position and compensated and corrected. This solves the problem in the prior art that the weak, high-frequency, non-periodic residual vibrations generated by the remote sensing satellite attitude control mechanism lead to imperceptible local deformation in the geometric processing of remote sensing images. This application can achieve pixel-level fine geometric correction of remote sensing images, significantly improve the geometric accuracy of remote sensing images, and overcome the shortcomings of the prior art in image geometric distortion caused by weak disturbances. Attached Figure Description

[0048] Figure 1 A flowchart illustrating a multi-source remote sensing satellite navigation data processing method provided in this application embodiment;

[0049] Figure 2 This is a system structure block diagram of a multi-source remote sensing satellite navigation data processing system provided in an embodiment of this application.

[0050] Figure reference numerals: 100, Multi-source remote sensing satellite navigation data processing system; 101, Vibration information acquisition module; 102, Disturbance component separation and extraction module; 103, Disturbance field map construction module; 104, Compensation and correction module. Detailed Implementation

[0051] To better illustrate the present invention, the invention will now be described in further detail with reference to the accompanying drawings.

[0052] It should be understood that, in order to make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. The components of the embodiments of this disclosure described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this disclosure provided in the accompanying drawings is not intended to limit the scope of the claimed disclosure, but merely represents selected embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without inventive effort are within the scope of protection of this disclosure.

[0053] The following description uses at least one specific embodiment as an example. In this embodiment:

[0054] Firstly, such as Figure 1 As shown, a multi-source remote sensing satellite navigation data processing method is provided, which compensates and corrects remote sensing images by employing multiple miniature inertial measurement units positioned around the imaging area of ​​a remote sensing satellite optical imager. The method includes the following steps:

[0055] Step S1: Use multiple miniature inertial measurement units to acquire local vibration information at different locations around the imaging area;

[0056] Step S2: Filter the acquired local vibration information to separate the high-frequency disturbance component from the local vibration information;

[0057] Step S3: Obtain the spatial distribution information of multiple micro inertial measurement units around the imaging area, and construct an attitude perturbation field map by using a spatial difference algorithm and combining it with high-frequency perturbation components. The attitude perturbation field map includes the attitude perturbation amount that each pixel in the imaging area needs to experience.

[0058] Step S4: Input the attitude perturbation field map into the remote sensing image that needs to be geometrically corrected, obtain the corresponding attitude perturbation amount according to the position area of ​​the pixels in the imaging area of ​​the remote sensing image, and compensate and correct the remote sensing image based on the attitude perturbation amount.

[0059] The solution in this embodiment aims to significantly improve the geometric correction accuracy of remote sensing images by measuring and compensating for local high-frequency vibrations in the imaging area of ​​the remote sensing satellite optical imager, thereby solving the problem that traditional attitude measurement systems cannot effectively handle local image deformation caused by weak high-frequency disturbances.

[0060] Specifically, the miniature inertial measurement unit mentioned in this application refers to a sensor that integrates a miniature gyroscope and a miniature accelerometer. Its main function is to measure the angular velocity and linear acceleration at its location, thereby acquiring local vibration information. These units are strategically deployed around the imaging area of ​​a remote sensing satellite optical imager to capture minute vibrations at different locations within that area.

[0061] The imaging area refers to the sensor array or focal plane area where the optical imager actually acquires images at a certain moment.

[0062] In the method proposed in this embodiment, multiple miniature inertial measurement units (MMUs) are first used to acquire local vibration information at different locations around the imaging area. These MMUs can be deployed and data acquired in various ways. For example, a distributed sensor network can be used, with multiple MMUs evenly distributed along the edge of the imaging area, and the acquired vibration data transmitted to a central processing unit via wired or wireless means. During data acquisition, each MMU continuously records its triaxial angular velocity and triaxial acceleration data at a preset sampling frequency.

[0063] Subsequently, the acquired local vibration information is filtered to separate high-frequency disturbance components. The filtering process aims to remove noise and low-frequency components from the data, retaining only the high-frequency vibrations that have the greatest impact on the geometric accuracy of the image. For example, digital filters, such as high-pass or band-pass filters, can be used to process the raw vibration data.

[0064] Next, the spatial distribution information of multiple miniature inertial measurement units (MMUs) around the imaging area is acquired, and an attitude perturbation field map is constructed by combining this information with high-frequency perturbation components. The spatial distribution information can be obtained through pre-measurement, for example, by determining the precise position of each MMU relative to the imaging area using laser ranging technology. The attitude perturbation field map is used to transform discrete local vibration information into a continuous attitude perturbation distribution. For example, spatial interpolation algorithms, such as Kriging interpolation, inverse distance weighted interpolation, or thin-plate spline interpolation, can be used to calculate the attitude perturbation experienced by each pixel in the imaging area based on the position of each MMU and its corresponding high-frequency perturbation components. This attitude perturbation can be expressed as angular displacement or angular velocity, reflecting the minute attitude deflection of the pixels in the imaging area at the moment of exposure.

[0065] Finally, the attitude perturbation field map is input into the remote sensing image requiring geometric correction. The corresponding attitude perturbation amount is obtained based on the position of each pixel in the imaging area, and the remote sensing image is then compensated and corrected based on this perturbation amount. Upon receiving the remote sensing image to be processed, the precise position of each pixel in the image within the imaging area must first be determined. After obtaining the attitude perturbation amount, it is superimposed onto the overall attitude parameters of the remote sensing satellite, thereby correcting attitude errors caused by high-frequency vibrations.

[0066] This application achieves precise sensing of local high-frequency vibrations in the imaging region by introducing distributed miniature inertial measurement units. By filtering these local vibration information, high-frequency disturbance components can be accurately separated, avoiding interference from low-frequency overall attitude changes on local disturbance analysis. More importantly, this application constructs an attitude disturbance field map, transforming discrete local vibration measurements into a continuous attitude disturbance distribution over the imaging region, ensuring that each pixel in the imaging region receives its precise attitude disturbance amount. This pixel-level attitude disturbance compensation and correction effectively eliminates irregular geometric deformations within the image, significantly improving the geometric correction accuracy of remote sensing images.

[0067] In some of the embodiments described above in this application, multiple micro inertial measurement units are used to acquire local vibration information at different locations around the imaging area. However, if the time references between these micro inertial measurement units are inconsistent, the acquired local vibration information may have temporal deviations, which in turn affect the accuracy of subsequent attitude disturbance field map construction and the precision of remote sensing image compensation and correction.

[0068] In response, this embodiment proposes an optimization scheme to ensure the consistency of the time period for each micro inertial measurement unit to acquire local vibration information.

[0069] The step of acquiring local vibration information at different locations around the imaging area using multiple miniature inertial measurement units also includes:

[0070] A pulse signal is provided by the receiver of a remote sensing satellite. This pulse signal is used to synchronize the time reference of all micro inertial measurement units, so that the time period for each micro inertial measurement unit to acquire local vibration information is consistent.

[0071] The receiver of a remote sensing satellite can be understood as a positioning receiving device in a remote sensing satellite navigation system. It can receive signals from navigation satellites and generate precise time information. This receiver is configured to output a periodic pulse signal, such as a pulse-per-second (PPS) signal or other high-precision time synchronization signal. This pulse signal is designed to serve as a unified time reference for all miniature inertial measurement units (MMUs). For example, when the pulse signal is received, all MMUs simultaneously start or reset their data acquisition cycles, or calibrate their internal timestamps with the pulse signal's timestamp. The purpose is to eliminate the minute time drift or start-up time differences that may occur due to independent timing by each MMU.

[0072] Therefore, by using a unified time reference, it is ensured that each micro inertial measurement unit collects local vibration information within the same time window, thus making the acquired local vibration information highly consistent and comparable in the time dimension.

[0073] In some preferred embodiments, the receiver of the remote sensing satellite can be a high-precision GPS receiver capable of outputting a pulse-per-second (PPS) signal. This GPS receiver continuously outputs the PPS signal when the remote sensing satellite's optical imager begins its imaging mission. Each miniature inertial measurement unit is designed to immediately activate its internal data acquisition clock and begin recording local vibration data upon receiving the rising edge of the PPS signal.

[0074] For example, if a miniature inertial measurement unit (INS) calibrates its internal clock to 00:00:00.000 after receiving a PPS signal, then all other miniature INS will also calibrate their internal clocks to 00:00:00.000 upon receiving the same PPS signal. This ensures that all miniature INS are strictly synchronized in subsequent data acquisition cycles, such as acquiring data every millisecond. Therefore, when compensation and correction are needed for remote sensing images at a specific moment, the local vibration information corresponding to that moment can be extracted from all miniature INS. This information is perfectly aligned in time, providing a solid data foundation for constructing accurate attitude disturbance field maps.

[0075] In some of the above embodiments, filtering of the acquired local vibration information is proposed to separate high-frequency disturbance components. However, in practical applications, the local vibration information acquired by the miniature inertial measurement unit may contain inherent sensor bias, environmental noise, and low-frequency disturbance components caused by changes in the overall attitude of the remote sensing satellite. Without targeted preprocessing and frequency-band filtering, these non-high-frequency disturbance components may obscure the accurate extraction of high-frequency disturbance components, thereby affecting the accuracy of subsequent remote sensing image compensation and correction.

[0076] Therefore, the step of filtering the acquired local vibration information to separate the high-frequency disturbance component from the local vibration information includes:

[0077] Step S21a: Preprocess the acquired local vibration information to remove zero bias and noise from the local vibration information;

[0078] Step S22b: Filter the preprocessed local vibration information to remove low-frequency disturbance components caused by changes in the overall attitude of the remote sensing satellite, and separate high-frequency disturbance components from the local vibration information.

[0079] Specifically, preprocessing of the acquired local vibration information aims to eliminate or significantly reduce systematic errors and random interference in the data. Zero bias refers to the non-zero value output by the sensor under no input or static conditions, which can be removed through calibration, differential processing, or model compensation. Noise refers to random, irregular signal fluctuations, which can be suppressed using various digital filtering techniques such as mean filtering, median filtering, and Kalman filtering. The purpose of preprocessing is to improve the signal-to-noise ratio and data quality of the local vibration information, laying the foundation for subsequent precise filtering.

[0080] The preprocessed local vibration information is filtered to remove low-frequency disturbance components caused by changes in the overall attitude of the remote sensing satellite. During operation in orbit, the overall attitude of a remote sensing satellite undergoes slow but significant changes, which typically manifest as low-frequency signals.

[0081] To accurately separate the high-frequency disturbance components caused by local micro-vibrations of the optical imager, high-pass or band-pass filters are needed to filter out these low-frequency components. For example, a high-pass filter can be used to decompose the preprocessed local vibration information into different frequency components and accurately separate the high-frequency disturbance components.

[0082] Specifically, by removing zero bias and noise, the quality of the raw data is improved, and the risk of error propagation is reduced. By filtering out low-frequency disturbance components, the interference of overall satellite attitude changes on local micro-vibrations is avoided, resulting in purer extracted high-frequency disturbance components that more accurately characterize the actual micro-vibration state of the optical imager's imaging area. This provides a more accurate input for the subsequent construction of attitude disturbance field maps, thereby improving the overall accuracy and effectiveness of geometric correction and compensation of remote sensing images.

[0083] In some preferred embodiments, the preprocessing of the acquired local vibration information can be performed as follows: First, the micro inertial measurement unit is statically calibrated to acquire and store its zero-bias parameter, which is then subtracted from the raw data after acquisition. Second, to address noise, a moving average filter can be used to preliminarily smooth the local vibration information, reducing the impact of random noise. For example, a moving average window of length N can be set, and the average of the N data points within the window can be used as the output for the current point.

[0084] Furthermore, when filtering the preprocessed local vibration information, a high-pass filter can be used to filter out low-frequency disturbance components.

[0085] Specifically, a second-order Butterworth high-pass filter with a cutoff frequency of 0.5 Hz can be designed. This filter can effectively suppress low-frequency components caused by overall attitude changes of remote sensing satellites (typically below 0.5 Hz), while allowing high-frequency disturbance components above 0.5 Hz to pass through. In this way, it can be ensured that the high-frequency disturbance components separated from local vibration information are accurate, thus providing a reliable data foundation for subsequent attitude disturbance field map construction and remote sensing image compensation and correction.

[0086] Traditional multi-source remote sensing satellite navigation data processing methods may fail to adequately consider the minute slippage or displacement at the interface between the micro inertial measurement unit and the imaging area when filtering the acquired local vibration information to separate high-frequency disturbance components. As a result, false motion signals may be mixed into the local vibration information, leading to inaccurate separation of high-frequency disturbance components and affecting the accuracy of remote sensing image compensation and correction.

[0087] In this embodiment, additional sensors are introduced and data correlation analysis is performed to identify and remove false motion signals, thereby obtaining more realistic local vibration information and improving the accuracy of high-frequency disturbance component separation.

[0088] Specifically, the step of filtering the acquired local vibration information to separate the high-frequency disturbance component from the local vibration information includes:

[0089] Step S21b: Deploy corresponding micro motion sensors and micro interface health sensors near the location of each micro inertial measurement unit in the imaging area. Use the micro motion sensors to obtain the motion data of the corresponding micro inertial measurement unit and use the micro interface health sensors to obtain the physical connection status data between the corresponding micro inertial measurement unit and the imaging area.

[0090] Step S22b: Perform correlation analysis on the motion data of the micro motion sensor and the physical connection status data of the micro interface health sensor to identify and quantify the pseudo motion signal caused by the tiny sliding displacement of the corresponding micro inertial measurement unit.

[0091] Step S23b: Remove the quantized false motion signals from the local vibration information at different locations around the imaging area obtained from multiple micro inertial measurement units to obtain the true local vibration information;

[0092] Step S24b: Filter the real local vibration information on the imaging area and separate the corresponding high-frequency disturbance components.

[0093] Specifically, a miniature motion sensor can be understood as a highly sensitive displacement or acceleration sensor, the purpose of which is to accurately measure the minute motion of a miniature inertial measurement unit in a local area. For example, a microelectromechanical system (MEMS) accelerometer or gyroscope can be used to capture the instantaneous displacement or attitude change of the miniature inertial measurement unit.

[0094] Miniature interface health sensors are used to monitor the physical connection status between the miniature inertial measurement unit and the imaging area. For example, piezoelectric sensors, strain gauges or contact sensors can be used to detect the pressure, strain or contact integrity of the connection interface. The purpose is to assess the stability of the connection and thus determine whether there are any abnormalities that may cause micro-slippage.

[0095] Motion data acquired by a miniature motion sensor and physical connectivity data acquired by a miniature interface health sensor are used for correlation analysis. This correlation analysis aims to identify pseudo-motion signals caused by minute slippage or displacement between the miniature inertial measurement unit and its mounting interface, rather than by the overall vibration of the remote sensing satellite, by comparing data from two different sources. For example, if the miniature motion sensor detects minute movement while the miniature interface health sensor simultaneously reports abnormal connectivity (such as pressure fluctuations or minor separation), it can be inferred that the movement may be pseudo-motion.

[0096] Therefore, pseudo-motion signals refer to non-real vibration signals generated in the output of the micro-inertial measurement unit (MIM) due to the incomplete stability of the physical connection between the micro-inertial measurement unit and the imaging area. When subjected to minor external or internal stresses, the micro-inertial measurement unit undergoes slight relative sliding or displacement relative to the imaging area. These signals are not true attitude disturbances of the remote sensing satellite as a whole or the imaging area, but rather measurement errors caused by local connection instability.

[0097] Through the above correlation analysis, these pseudo motion signals can be identified and quantified. For example, the amplitude and frequency characteristics of the pseudo motion signals can be accurately calculated by establishing a mathematical model between motion data and connection state data or by using machine learning algorithms for pattern recognition.

[0098] After obtaining the quantized pseudo-motion signal, it is removed from the local vibration information acquired by multiple miniature inertial measurement units, thus eliminating errors caused by sensor installation defects. The real local vibration information is then input to the filtering module for processing to filter out low-frequency disturbance components caused by changes in the overall attitude of the remote sensing satellite, and finally separate the corresponding high-frequency disturbance components.

[0099] As a specific implementation, it is assumed that multiple miniature inertial measurement units (MMUs) are deployed around the imaging area of ​​a remote sensing satellite optical imager. Near the mounting base of one of the MMUs, a miniature motion sensor and a miniature interface health sensor are deployed. The miniature motion sensor continuously monitors the minute displacements and vibrations of the MMU, while the miniature interface health sensor detects the contact pressure and strain at the interface between the MMU and the imaging area in real time.

[0100] When a remote sensing satellite is in orbit, a miniature inertial measurement unit (MIG / IPU) acquires local vibration information. Simultaneously, a miniature motion sensor records its motion data, and a miniature interface health sensor records physical connection status data. If the miniature motion sensor detects a high-frequency, minute displacement signal, while the miniature interface health sensor simultaneously reports a momentary pressure drop or slight loosening at the connection interface, the system will use a pre-defined correlation analysis model (e.g., a machine learning-based classifier) ​​to determine that the displacement signal is likely a false motion signal caused by a minute slippage between the miniature MIG / IPU and the imaging area.

[0101] Once the spurious motion signal is identified, its amplitude and frequency characteristics are quantized. For example, the root mean square or peak value of the spurious signal over a specific time period can be calculated. This quantized spurious motion signal is then subtracted from the original local vibration information acquired by the miniature inertial measurement unit (MMU). This yields more accurate local vibration information, free from the influence of minor slippage. Finally, filtering this accurate local vibration information allows for more precise separation of high-frequency disturbance components, which are then used for subsequent attitude disturbance field mapping and remote sensing image compensation correction. In this way, even with minor installation defects in the miniature MMU, the measurement accuracy of attitude disturbances can be guaranteed.

[0102] The above technical solution effectively solves the problem of spurious motion signals introduced by the slight slippage at the interface between the miniature inertial measurement unit and the imaging area in traditional methods. By introducing miniature motion sensors and miniature interface health sensors, and performing correlation analysis on multi-source data, the system can accurately identify and quantify these spurious motion signals and remove them from the original local vibration information. Therefore, the obtained local vibration information is more realistic and reliable, ensuring higher accuracy for the high-frequency disturbance components separated by subsequent filtering.

[0103] In this embodiment, an optimization is proposed for the steps of acquiring the spatial distribution information of multiple micro inertial measurement units around the imaging area and constructing an attitude disturbance field map by combining high-frequency disturbance components, so as to improve the construction accuracy of the attitude disturbance field map.

[0104] Specifically, the step of acquiring the spatial distribution information of multiple miniature inertial measurement units around the imaging area and constructing an attitude perturbation field map by combining high-frequency perturbation components includes:

[0105] Step S31: Obtain the spatial distribution information of each micro inertial measurement unit around the imaging area and the high-frequency disturbance component corresponding to each micro inertial measurement unit.

[0106] Step S32: The attitude perturbation of any pixel in the imaging area is calculated using the thin plate spline interpolation algorithm;

[0107] Step S33: Construct an attitude perturbation field map based on the attitude perturbation of any pixel in the imaging area.

[0108] Specifically, acquiring the spatial distribution information of each micro inertial measurement unit (IMU) around the imaging area refers to determining the precise position of each IMU within the imaging area. This positional information can be obtained through pre-calibration. Meanwhile, the high-frequency disturbance components corresponding to each IMU refer to the disturbance data reflecting local attitude changes, separated from the local vibration information after filtering.

[0109] The core of this scheme is the calculation of the attitude perturbation at any pixel in the imaging region using the thin-plate spline interpolation algorithm. The thin-plate spline interpolation algorithm is an energy-minimization-based interpolation method that can construct a smooth and continuous surface based on discrete observation point data, thereby accurately estimating the attitude perturbation at any location within the imaging region.

[0110] In practical applications, constructing an attitude perturbation field map based on the attitude perturbation of any pixel in the imaging region refers to aggregating the attitude perturbation of each pixel in the imaging region calculated using the thin-plate spline interpolation algorithm to form a two-dimensional or three-dimensional data field. This attitude perturbation field map intuitively represents the attitude perturbation experienced by each pixel in the imaging region at a specific moment, providing a refined basis for subsequent remote sensing image compensation and correction.

[0111] By employing the aforementioned technical solution and utilizing the thin-plate spline interpolation algorithm to construct the attitude perturbation field map, the construction accuracy and spatial resolution of the attitude perturbation field map can be significantly improved. This provides a more refined and accurate input of attitude perturbation quantities for the geometric correction of remote sensing images, ultimately effectively improving the compensation and correction effect and imaging quality of remote sensing images, and reducing image distortion caused by attitude perturbations.

[0112] As a preferred example in this embodiment, assume that N miniature inertial measurement units (MMUs) are deployed around the imaging area of ​​the remote sensing satellite optical imager. The spatial coordinates of these MMUs are (x1, y1, z1), (x2, y2, z2), ..., (xN, yN, zN). At a certain moment, each MMU acquires the corresponding high-frequency disturbance component, represented as (δθ1, δφ1, δψ1), (δθ2, δφ2, δψ2), ..., (δθN, δφN, δψN), where δθ, δφ, and δψ represent the attitude disturbances in the pitch, roll, and yaw directions, respectively.

[0113] Specifically, first, the system acquires the spatial distribution information of these miniature inertial measurement units and their corresponding high-frequency perturbation components. Then, using these discrete data points as input, a thin-plate spline interpolation algorithm is applied. This algorithm generates a continuous attitude perturbation function across the entire imaging area based on these discrete attitude perturbation data. For example, for any pixel (xp, yp, zp) in the imaging area, the thin-plate spline interpolation algorithm can calculate the attitude perturbation (δθp, δφp, δψp) experienced by that pixel at the current moment. By performing this calculation on all pixels in the imaging area, an attitude perturbation field map containing the attitude perturbation experienced by each pixel in the imaging area can be constructed. This field map accurately reflects the spatial distribution of attitude perturbations within the imaging area, providing a precise basis for subsequent remote sensing image compensation and correction.

[0114] Furthermore, the step of inputting the attitude perturbation field map into the remote sensing image that needs geometric correction, obtaining the corresponding attitude perturbation amount based on the position region of the pixels in the imaging area, and compensating and correcting the remote sensing image based on the attitude perturbation amount specifically includes:

[0115] Step S41: Acquire the remote sensing image that needs to be geometrically corrected;

[0116] Step S42: Based on the position of each pixel in the remote sensing image to be processed in the imaging area, query the attitude perturbation field map to obtain the attitude perturbation amount corresponding to the position of each pixel in the remote sensing image to be processed in the imaging area.

[0117] Step S43: Superimpose and compensate the corresponding attitude disturbances into the overall attitude parameters of the remote sensing satellite.

[0118] In this process, based on the position of each pixel in the remote sensing image to be processed within the imaging region, the attitude perturbation amount corresponding to that position in the imaging region is retrieved from the attitude perturbation field map. This can be understood as follows: since the attitude perturbation field map already contains the attitude perturbation amount experienced by each pixel in the imaging region, for each pixel in the remote sensing image, its precise spatial position in the imaging region can be used to search within the pre-constructed attitude perturbation field map, thereby accurately obtaining the local attitude perturbation amount experienced by that pixel during imaging. The purpose is to provide refined perturbation data for subsequent compensation and correction.

[0119] Specifically, superimposing and compensating the corresponding attitude perturbation amounts into the overall attitude parameters of the remote sensing satellite means combining the local attitude perturbation amounts obtained above for each pixel in the imaging area with the overall attitude parameters of the remote sensing satellite at the time of imaging. This superposition compensation can be understood as a local correction of the overall attitude parameters of the remote sensing satellite, so that the geometric position correction of each pixel can take into account the small attitude perturbations it experiences, thereby improving the accuracy of geometric correction.

[0120] In this embodiment, more precise and detailed compensation for the geometric correction of remote sensing images can be achieved. Compared to traditional methods that rely solely on the overall attitude parameters of the remote sensing satellite for correction, this application obtains the local attitude perturbation amount corresponding to each pixel in the imaging area of ​​the remote sensing image and superimposes it into the overall attitude parameters, effectively eliminating the impact of local high-frequency perturbations on the geometric accuracy of the remote sensing image. Therefore, the geometric positioning accuracy and image quality of the remote sensing image can be significantly improved, providing more reliable basic data for subsequent remote sensing data applications.

[0121] In some of the above implementations, a scheme was proposed to compensate and correct remote sensing images by constructing an attitude perturbation field map. However, in actual remote sensing imaging, individual pixels in the remote sensing image may be exposed at different times, and high-frequency perturbation components may undergo rapid dynamic changes throughout the imaging cycle. If compensation is based solely on a static attitude perturbation field map, it may not be possible to completely eliminate residual geometric errors caused by differences in pixel exposure time and the dynamic nature of perturbations. This limitation is particularly pronounced when processing high-precision remote sensing images.

[0122] In response, this embodiment further proposes a method for secondary geometric correction of remote sensing images to further improve the accuracy of compensation and correction.

[0123] After the steps of inputting the attitude perturbation field map into the remote sensing image requiring geometric correction, obtaining the corresponding attitude perturbation amount based on the position region of the pixels in the imaging area, and compensating and correcting the remote sensing image based on the attitude perturbation amount, the method further includes:

[0124] Obtain the exposure time window for each pixel of the compensated and corrected remote sensing image;

[0125] High-frequency perturbation components corresponding to each pixel of the remote sensing image within the exposure time window are extracted from the local vibration information at different locations around the imaging area obtained from each micro inertial measurement unit.

[0126] Based on the high-frequency perturbation components of each pixel in the remote sensing image at its exposure time point and their spatial distribution in the imaging area, a sequence of local attitude perturbation field maps is constructed in which the exposure time windows of each pixel in the remote sensing image are continuously ordered.

[0127] Based on the position of each pixel in the remote sensing image on the imaging area and its corresponding exposure time point, the corresponding attitude perturbation amount is queried from the local attitude perturbation field map sequence.

[0128] The queried attitude perturbation is used as the input for geometric correction of the pixels in the remote sensing image, and a second geometric correction is performed on the remote sensing image based on the attitude perturbation.

[0129] Specifically, after completing the initial compensation and correction, it is necessary to obtain the precise exposure time window for each pixel in the remote sensing image. Extracting the high-frequency perturbation component corresponding to each pixel in the remote sensing image within the exposure time window from the local vibration information at different locations around the imaging area obtained from each miniature inertial measurement unit refers to accurately extracting the high-frequency perturbation data within that specific time period from the continuously acquired local vibration information for each pixel's specific exposure time window.

[0130] Therefore, based on the high-frequency perturbation components of each pixel in the remote sensing image at its exposure time point and their spatial distribution in the imaging area, a sequence of local attitude perturbation field maps is constructed, continuously ordered according to the exposure time window of each pixel in the remote sensing image. This sequence is no longer a static perturbation field map, but consists of a series of temporally continuous and spatially refined perturbation field maps, with each field map or field map segment precisely corresponding to the exposure time of a specific pixel or group of pixels in the remote sensing image.

[0131] Furthermore, based on the position of each pixel in the remote sensing image on the imaging area and its corresponding exposure time, the corresponding attitude perturbation amount is queried from the local attitude perturbation field map sequence. The purpose is to find the attitude perturbation information that best matches the exposure time and spatial position of each pixel.

[0132] Finally, the queried attitude perturbation is used as the input for geometric correction of the pixels in the remote sensing image, and a second geometric correction is performed on the remote sensing image based on the attitude perturbation, thereby further improving the geometric accuracy of the remote sensing image.

[0133] The solution presented in this embodiment can significantly improve the geometric correction accuracy of remote sensing images, especially in high-resolution remote sensing images and dynamic vibration environments. It can effectively eliminate residual geometric errors caused by rapid changes in high-frequency disturbance components and differences in pixel exposure time. As a result, the geometric accuracy and data quality of remote sensing images are greatly improved, providing a more reliable foundation for subsequent remote sensing data applications.

[0134] As an exemplary example in this embodiment, a specific example is described below. Assume a remote sensing satellite equipped with a high-resolution optical imager is conducting Earth observation. The imager uses a rolling shutter mechanism, causing different rows of pixels in the remote sensing image to be exposed sequentially within microsecond time intervals. After completing preliminary geometric compensation correction based on the overall attitude perturbation field map, the system accurately records the exposure start time and exposure duration of each row of pixels in the remote sensing image, i.e., obtains the exposure time window for each pixel in the remote sensing image. For example, the first row of pixels is exposed at time T1, the second row of pixels is exposed at time T1+Δt, where Δt indicates the exposure interval time, and so on. Simultaneously, multiple miniature inertial measurement units (MMUs) positioned around the imaging area continuously collect local vibration information at high frequency. For each row of pixels in the remote sensing image, the system extracts the high-frequency perturbation components experienced during the exposure of that row of pixels from the continuous vibration data stream of the MMUs, based on its specific exposure time window. Subsequently, combining these pixel-level high-frequency perturbation components and their spatial distribution in the imaging area, the system constructs a temporally continuous sequence of local attitude perturbation field maps. When secondary geometric correction of remote sensing imagery is required, the system queries and retrieves the most accurate, time-synchronized attitude perturbation amount from the pre-constructed local attitude perturbation field map sequence, based on the position of each pixel (or row of pixels) in the imaging area and its corresponding exposure time. Finally, these retrieved pixel-level attitude perturbation amounts are superimposed onto the overall attitude parameters of the remote sensing satellite, and secondary geometric correction is performed on the remote sensing imagery based on this.

[0135] Furthermore, the step of extracting the high-frequency perturbation component corresponding to each pixel of the remote sensing image during exposure from the local vibration information at different locations around the imaging area obtained from each micro inertial measurement unit includes:

[0136] Dynamically adjust the effective exposure time window length for each pixel of the remote sensing image;

[0137] Extract high-frequency disturbance components corresponding to the adjusted effective exposure time window from each local vibration information;

[0138] Calculate the weighted average of the high-frequency disturbance components within the adjusted effective exposure time window, where the weights are allocated based on the rate of change and amplitude of the high-frequency disturbance components within the adjusted effective exposure time window.

[0139] Specifically, dynamically adjusting the effective exposure time window length for each pixel in a remotely sensed image refers to adaptively adjusting the exposure time window used to extract high-frequency disturbance components based on real-time monitored vibration characteristics, such as instantaneous frequency and amplitude change rate. This adjustment ensures that the selected window length better matches the vibration period and intensity at the current moment, avoiding the averaging out of high-frequency information due to an excessively long window or the introduction of excessive noise due to an excessively short window.

[0140] The extraction of high-frequency disturbance components corresponding to the adjusted effective exposure time window from various local vibration information sources refers to, after determining the dynamically adjusted effective exposure time window for each pixel, extracting data segments within that time window from the local vibration information acquired by the micro inertial measurement unit and separating the high-frequency disturbance components from them. This process can utilize the previously described filtering method to ensure that the extracted components contain only high-frequency disturbances.

[0141] In practical applications, the weighted average of high-frequency disturbance components within the adjusted effective exposure time window is calculated. The weights are allocated based on the rate of change and amplitude of the high-frequency disturbance components within the adjusted effective exposure time window. This means that after acquiring the high-frequency disturbance components within the adjusted window, different weights are assigned to these components according to their characteristics within the window. For example, disturbance components with larger rates of change or higher amplitudes can be assigned higher weights to highlight their dominant role in pixel pose perturbation; conversely, components with gradual changes or smaller amplitudes can be assigned lower weights. The aim is to more accurately characterize the average high-frequency disturbance experienced by the pixel during exposure, reducing errors caused by vibration nonlinearity or instantaneous impacts.

[0142] The proposed solution dynamically adjusts the effective exposure time window length for each pixel of the remote sensing image, enabling the extraction window for high-frequency disturbance components to adaptively match the instantaneous vibration characteristics of the remote sensing satellite platform or optical imager. When the vibration frequency is high or the amplitude changes drastically, the window length can be shortened to capture finer instantaneous disturbances; when the vibration is relatively stable, the window length can be appropriately extended to improve the statistical stability of the extraction results.

[0143] In some preferred embodiments, a specific example is given below. Suppose that during the operation of a remote sensing satellite in orbit, due to attitude adjustment or the instantaneous activation of a micro-vibration source, the imaging area of ​​the optical imager experiences a high-frequency, high-amplitude instantaneous impact vibration within a certain time period. If a fixed-length exposure time window is used to extract the high-frequency disturbance component, this fixed window may not be able to completely cover or accurately capture the complete characteristics of this instantaneous impact, or the amplitude of the impact may be averaged because the window includes relatively stable data before and after the impact, thus underestimating the actual disturbance intensity.

[0144] For example, near the exposure time of a pixel, if the instantaneous frequency of the vibration signal rapidly increases from 50Hz to 200Hz, and the amplitude triples in a very short time, the system will calculate and dynamically adjust the effective exposure time window length for that pixel based on this real-time data. For instance, the original 10-millisecond window may be shortened to 2 milliseconds to better focus on the core time period of the impact. Simultaneously, when calculating the weighted average of the high-frequency disturbance components within this 2-millisecond window, higher weights are assigned to disturbance data points near the impact peak with the largest amplitude and fastest rate of change, while relatively lower weights are assigned to the impact edges or parts with smaller amplitudes. In this way, the extracted high-frequency disturbance components can more accurately reflect the true attitude disturbance of the pixel during the instantaneous impact, avoiding errors caused by averaging, thus providing more accurate compensation data for the secondary geometric correction of remote sensing images.

[0145] Furthermore, the step of dynamically adjusting the effective exposure time window length for each pixel of the remote sensing image includes:

[0146] Real-time monitoring of instantaneous frequency and amplitude change rate in local vibration information acquired by miniature inertial measurement units;

[0147] Based on the instantaneous frequency and amplitude change rate in the local vibration information obtained by the miniature inertial measurement unit, the effective exposure time window length of each pixel in the remote sensing image is calculated and adjusted.

[0148] Specifically, real-time monitoring of the instantaneous frequency and amplitude change rate in local vibration information acquired by a miniature inertial measurement unit (MIGU) refers to continuously analyzing the local vibration information acquired by the MIGU to obtain the trends in its instantaneous frequency and amplitude changes. The instantaneous frequency can be understood as the frequency characteristics of the vibration signal at a specific moment, which can be extracted using short-time Fourier transform signal processing techniques. The amplitude change rate reflects how quickly the vibration intensity changes over time and can be calculated using a differential detection algorithm on the envelope of the vibration signal. The purpose is to obtain the real-time dynamic characteristics of the vibration environment, providing precise input for subsequent window length adjustments.

[0149] The calculation and adjustment of the effective exposure time window length for each pixel in a remotely sensed image, based on the instantaneous frequency and amplitude change rate of local vibration information acquired by a miniature inertial measurement unit, refers to determining the optimal effective exposure time window length for each pixel using a preset adjustment strategy or adaptive algorithm, based on the real-time monitored instantaneous frequency and amplitude change rate. For example, when the instantaneous frequency or amplitude change rate of the high-frequency vibration component is high, it may be necessary to shorten the effective exposure time window length to more accurately capture instantaneous disturbances; conversely, when the vibration is relatively stable, the window length can be appropriately extended to improve the signal-to-noise ratio. This calculation and adjustment process can be implemented based on empirical models, lookup tables, or machine learning algorithms. Its purpose is to enable the exposure time window length to dynamically adapt to the actual vibration environment, ensuring that the high-frequency disturbance components most representative of the pixel's exposure period can be extracted under different vibration conditions.

[0150] Through the above technical solution, the effective exposure time window length of each pixel in the remote sensing image can be dynamically and intelligently adjusted according to the real-time instantaneous frequency and amplitude change rate of the local vibration information obtained by the miniature inertial measurement unit. Compared with the solution that only performs simple dynamic adjustment, this application can more accurately adapt to the complex and variable vibration environment of the remote sensing satellite platform, and effectively avoid the problem of inaccurate extraction of high-frequency disturbance components caused by the mismatch between the window length and the actual vibration characteristics.

[0151] As an exemplary example in this embodiment, it is assumed that during the operation of a remote sensing satellite in orbit, due to attitude adjustments or minor movements of internal mechanical components, the local vibration information monitored by the miniature inertial measurement unit exhibits rapid changes in instantaneous frequency and amplitude. For example, at a certain moment, the instantaneous frequency of the vibration signal rapidly increases from 50Hz to 150Hz, while the amplitude also increases significantly. At this time, the system will monitor these changes in real time. Specifically, a short-time Fourier transform can be used to perform spectral analysis on the local vibration information to obtain the instantaneous frequency; simultaneously, the amplitude change rate can be obtained by performing a Hilbert transform on the envelope of the vibration signal and taking its derivative. According to a preset adjustment strategy, for example, when the instantaneous frequency exceeds a certain threshold (such as 100Hz) and the amplitude change rate exceeds another threshold, the system will automatically calculate and shorten the effective exposure time window length of the corresponding pixel in the remote sensing image, for example, from the original 10 milliseconds to 5 milliseconds. In this way, during periods of intense vibration and high frequency, the high-frequency disturbance components within the window can be captured more precisely, avoiding the situation where the instantaneous intense disturbance is averaged out due to an excessively long window. Conversely, when the vibration is detected to be stable and the instantaneous frequency and amplitude change rate are both low, the system may appropriately extend the exposure time window, for example, by adjusting it from 5 milliseconds back to 8 milliseconds, in order to improve the signal-to-noise ratio and ensure that disturbance information can be accurately extracted even during the stable period.

[0152] Secondly, such as Figure 2As shown, a multi-source remote sensing satellite navigation data processing system 100 is provided, which also includes:

[0153] Vibration information acquisition module 101 is used to acquire local vibration information at different locations around the imaging area using multiple miniature inertial measurement units;

[0154] The disturbance component separation and extraction module 102 is used to filter the acquired local vibration information and separate the high-frequency disturbance component from the local vibration information.

[0155] The perturbation field map construction module 103 is used to acquire the spatial distribution information of multiple micro inertial measurement units around the imaging area, and construct the attitude perturbation field map by using a spatial difference algorithm and combining it with high-frequency perturbation components. The attitude perturbation field map contains the attitude perturbation amount that each pixel in the imaging area needs to experience.

[0156] The compensation and correction module 104 is used to input the attitude disturbance field map into the remote sensing image that needs to be geometrically corrected, obtain the corresponding attitude disturbance amount according to the position area of ​​the pixels of the remote sensing image on the imaging area, and perform compensation and correction on the remote sensing image based on the attitude disturbance amount.

[0157] As described above, the multi-source remote sensing satellite navigation data processing method and system provided in this application can acquire local vibration information at different locations around the imaging area by setting up multiple miniature inertial measurement units around the imaging area of ​​the remote sensing satellite optical imager. After filtering, these local vibration information can separate high-frequency disturbance components that are difficult for traditional attitude measurement systems to capture. Combining the spatial distribution information of the miniature inertial measurement units and the spatial difference algorithm, this application can construct a fine attitude disturbance field map, which contains the attitude disturbance amount that each pixel in the imaging area needs to experience. Finally, the attitude disturbance field map is input into the remote sensing image, and the corresponding attitude disturbance amount is obtained according to the pixel position and compensated and corrected. This solves the problem in the prior art that the weak, high-frequency, non-periodic residual vibrations generated by the remote sensing satellite attitude control mechanism lead to imperceptible local deformation in the geometric processing of remote sensing images. This application can achieve pixel-level fine geometric correction of remote sensing images, significantly improve the geometric accuracy of remote sensing images, and overcome the shortcomings of the prior art in image geometric distortion caused by weak disturbances.

[0158] Finally, it should be noted that the above-described embodiments are merely specific implementations of this disclosure, used to illustrate the technical solutions of this disclosure, and not to limit them. The protection scope of this disclosure is not limited thereto. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features within the scope of the technology disclosed in this disclosure. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure, and should all be covered within the protection scope of this disclosure.

Claims

1. A method for processing multi-source remote sensing satellite navigation data, characterized in that it uses multiple miniature inertial measurement units positioned around the imaging area of ​​a remote sensing satellite optical imager to compensate and correct remote sensing images, The method includes the following steps: Multiple miniature inertial measurement units are used to acquire local vibration information at different locations around the imaging area; The acquired local vibration information is filtered to separate the high-frequency disturbance component from the local vibration information; The spatial distribution information of multiple micro inertial measurement units around the imaging area is obtained, and an attitude perturbation field map is constructed by using a spatial difference algorithm and combining it with high-frequency perturbation components. The attitude perturbation field map contains the attitude perturbation amount that each pixel in the imaging area needs to experience. The attitude perturbation field map is input into the remote sensing image that needs to be geometrically corrected. The corresponding attitude perturbation amount is obtained according to the position area of ​​the pixels in the imaging area of ​​the remote sensing image, and the remote sensing image is compensated and corrected based on the attitude perturbation amount. The step of filtering the acquired local vibration information to separate the high-frequency disturbance component specifically includes: A corresponding micro motion sensor and a micro interface health sensor are deployed near the location of each micro inertial measurement unit in the imaging area. The motion data of the corresponding micro inertial measurement unit is obtained by using the micro motion sensor, and the physical connection status data between the corresponding micro inertial measurement unit and the imaging area is obtained by using the micro interface health sensor. The motion data of the micro motion sensor and the physical connection status data of the micro interface health sensor are correlated and analyzed to identify and quantify the pseudo motion signals caused by the tiny sliding displacement of the corresponding micro inertial measurement unit. The quantized false motion signals are removed from the local vibration information at different locations around the imaging area obtained by multiple miniature inertial measurement units to obtain the true local vibration information. The actual local vibration information in the imaging area is filtered and the corresponding high-frequency disturbance components are separated.

2. The multi-source remote sensing satellite navigation data processing method according to claim 1, characterized in that, The step of acquiring local vibration information at different locations around the imaging area using multiple miniature inertial measurement units also includes: A pulse signal is provided by the receiver of a remote sensing satellite. This pulse signal is used to synchronize the time reference of all micro inertial measurement units, so that the time period for each micro inertial measurement unit to acquire local vibration information is consistent.

3. The multi-source remote sensing satellite navigation data processing method according to claim 1, characterized in that, The step of filtering the acquired local vibration information to separate the high-frequency disturbance component from the local vibration information includes: The acquired local vibration information is preprocessed to remove zero bias and noise. The preprocessed local vibration information is filtered to remove low-frequency disturbance components caused by changes in the overall attitude of the remote sensing satellite, and high-frequency disturbance components are separated from the local vibration information.

4. The multi-source remote sensing satellite navigation data processing method according to claim 1, characterized in that, The steps of acquiring the spatial distribution information of multiple miniature inertial measurement units around the imaging area, and constructing an attitude perturbation field map using a spatial difference algorithm combined with high-frequency perturbation components include: Acquire the spatial distribution information of each micro inertial measurement unit around the imaging area and the high-frequency disturbance components corresponding to each micro inertial measurement unit; The attitude perturbation of any pixel in the imaging area is calculated using the thin plate spline interpolation algorithm. An attitude perturbation field map is constructed based on the attitude perturbation of any pixel in the imaging region.

5. The multi-source remote sensing satellite navigation data processing method according to claim 1, characterized in that, The steps of inputting the attitude perturbation field map into the remote sensing image requiring geometric correction, obtaining the corresponding attitude perturbation amount based on the position region of the pixels in the imaging area, and compensating and correcting the remote sensing image based on the attitude perturbation amount specifically include: Acquire remote sensing images that require geometric correction; Based on the position of each pixel in the remote sensing image to be processed in the imaging area, the attitude perturbation amount corresponding to the position of each pixel in the remote sensing image to be processed in the imaging area is obtained from the attitude perturbation field map. The corresponding attitude disturbances are superimposed and compensated into the overall attitude parameters of the remote sensing satellite.

6. The multi-source remote sensing satellite navigation data processing method according to claim 1, characterized in that, After the steps of inputting the attitude perturbation field map into the remote sensing image requiring geometric correction, obtaining the corresponding attitude perturbation amount based on the position region of the pixels in the imaging area, and compensating and correcting the remote sensing image based on the attitude perturbation amount, the method further includes: Obtain the exposure time window for each pixel of the compensated and corrected remote sensing image; High-frequency perturbation components corresponding to each pixel of the remote sensing image within the exposure time window are extracted from the local vibration information at different locations around the imaging area obtained from each micro inertial measurement unit. Based on the high-frequency perturbation component of each pixel in the remote sensing image at its exposure time point and its position in the imaging area Spatial distribution: Construct a sequence of local attitude perturbation field maps that are continuously sorted according to the exposure time window of each pixel in the remote sensing image; Based on the position of each pixel in the remote sensing image on the imaging area and its corresponding exposure time point, the corresponding attitude perturbation amount is queried from the local attitude perturbation field map sequence. The queried attitude perturbation is used as the input for geometric correction of the pixels in the remote sensing image, and a second geometric correction is performed on the remote sensing image based on the attitude perturbation.

7. The multi-source remote sensing satellite navigation data processing method according to claim 6, characterized in that, The step of extracting the high-frequency perturbation component corresponding to each pixel of the remote sensing image within the exposure time window from the local vibration information at different locations around the imaging area obtained from each micro inertial measurement unit includes: Dynamically adjust the effective exposure time window length for each pixel of the remote sensing image; Extract high-frequency disturbance components corresponding to the adjusted effective exposure time window from each local vibration information; Calculate the weighted average of the high-frequency disturbance components within the adjusted effective exposure time window, where the weights are allocated based on the rate of change and amplitude of the high-frequency disturbance components within the adjusted effective exposure time window.

8. The multi-source remote sensing satellite navigation data processing method according to claim 7, characterized in that, The step of dynamically adjusting the effective exposure time window length for each pixel of the remote sensing image includes: Real-time monitoring of instantaneous frequency and amplitude change rate in local vibration information acquired by miniature inertial measurement units; Based on the instantaneous frequency and amplitude change rate in the local vibration information obtained by the miniature inertial measurement unit, the effective exposure time window length of each pixel in the remote sensing image is calculated and adjusted.

9. A multi-source remote sensing satellite navigation data processing system, characterized in that, Also includes: A vibration information acquisition module is used to acquire local vibration information at different locations around the imaging area using multiple miniature inertial measurement units. A disturbance component separation and extraction module is used to filter the acquired local vibration information and separate high-frequency disturbance components from the local vibration information. The perturbation field map construction module is used to acquire the spatial distribution information of multiple micro inertial measurement units around the imaging area, and constructs the attitude perturbation field map by using a spatial difference algorithm and combining it with high-frequency perturbation components. The attitude perturbation field map contains the attitude perturbation amount that each pixel in the imaging area needs to experience. The compensation and correction module is used to input the attitude disturbance field map into the remote sensing image that needs to be geometrically corrected, obtain the corresponding attitude disturbance amount according to the position area of ​​the pixels of the remote sensing image on the imaging area, and perform compensation and correction on the remote sensing image based on the attitude disturbance amount. The disturbance component separation and extraction module is also used for: A corresponding micro motion sensor and a micro interface health sensor are deployed near the location of each micro inertial measurement unit in the imaging area. The motion data of the corresponding micro inertial measurement unit is obtained by using the micro motion sensor, and the physical connection status data between the corresponding micro inertial measurement unit and the imaging area is obtained by using the micro interface health sensor. The motion data of the micro motion sensor and the physical connection status data of the micro interface health sensor are correlated and analyzed to identify and quantify the pseudo motion signals caused by the tiny sliding displacement of the corresponding micro inertial measurement unit. The quantized false motion signals are removed from the local vibration information at different locations around the imaging area obtained by multiple miniature inertial measurement units to obtain the true local vibration information. The actual local vibration information in the imaging area is filtered and the corresponding high-frequency disturbance components are separated.

Images