Image reconstruction method and device for very long baseline interferometry data

Abstract

The invention provides an image reconstruction method and device for very long baseline interference measurement data, and relates to the technical field of radio astronomy observation, and the method comprises the steps: processing the very long baseline interference measurement data of a target observation object in batches, reading the visibility data of each observation baseline, and processing the abnormal data, obtaining target visibility data; performing image deconvolution on a dirty image corresponding to the target visibility data to obtain a clean image and a clean model of the target observation object with dirty lobe interference removed; based on a target antenna gain factor corresponding to the observation baseline solved by the cleaning model, correcting observation data of the target observation object, and iteratively optimizing the cleaning model to determine a target cleaning model and correction parameters; and on the basis of the preset parameters and the target cleaning model of the at least one target observation object determined by the nonlinear least square method, the target cleaning model and the correction parameters, automatic image reconstruction is carried out, and the insufficient capacities of VLBI data batch processing, visual output and the like are relieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of radio astronomy observation, and in particular to a method and device for image reconstruction of very long baseline interferometry data. BACKGROUND

[0002] Very Long Baseline Interferometry (VLBI) is an interferometry technique that combines the observations of multiple ground-based radio telescopes to synthesize high-resolution images of radio sources through time synchronization and related data processing methods. This technique has important application value in frontier scientific fields such as astrophysics, precise geodesy, and black hole imaging research. With the continuous improvement of observation sensitivity and the continuous expansion of observation bandwidth, VLBI data presents new characteristics such as large data volume, significant expansion of dynamic range, and complex structural features, which makes the limitations of traditional data processing procedures based on proprietary software and manual intervention in processing efficiency, subjective bias control, and large-scale processing capacity increasingly prominent, and it is difficult to meet the strict requirements of modern scientific research on data processing efficiency, standardization, and result reproducibility.

[0003] Currently, the main image reconstruction methods in the VLBI field include the CLEAN deconvolution algorithm, self-calibration technology, and model fitting methods. However, these processing methods generally have many manual intervention links, long processing flow chains, and lack of uniformity in module interfaces. In view of the above technical bottlenecks, existing software tools still have obvious deficiencies in terms of automation processing level, batch processing capability, and visualized result output. SUMMARY

[0004] The present application aims to provide a method and device for image reconstruction of very long baseline interferometry data to alleviate the technical problems of insufficient batch processing and visualized output capabilities of VLBI data in the prior art.

[0005] In a first aspect, the present application provides a method for image reconstruction of very long baseline interferometry data, comprising: batch processing very long baseline interferometry data for at least one target observation object, arranging to generate a structured visibility table, and reading visibility data of each observation baseline from the structured visibility table; eliminating observation points with amplitude anomalies in the visibility data of each observation baseline, deleting visibility data with phase anomalies exceeding a preset time threshold, shielding visibility data corresponding to abnormal antennas, filtering visibility data corresponding to signal strength abnormal fault time periods, and obtaining target visibility data; performing image deconvolution on the dirty map corresponding to the target visibility data to obtain a clean image and a clean model for at least one target observation object without dirty lobe interference; correcting observation data for at least one target observation object based on the target antenna gain factor corresponding to the observation baseline solved based on the clean model, and iteratively optimizing the clean model to determine a target clean model and a correction parameter of the target clean model; determining a preset parameter of at least one target observation object based on the target clean model and the correction parameter by using a nonlinear least square method; performing automatic image reconstruction according to the preset parameter and the target clean model.

[0006] In an optional embodiment, the steps of batch processing very long baseline interferometry data for at least one target observation object, arranging a structured visibility table, and reading visibility data of each observation baseline from the structured visibility table, comprise: batch processing very long baseline interferometry data for at least one target observation object to obtain observation data for characterizing signal strength and phase of each observation baseline in length and direction respectively; wherein the observation data comprises spatial frequency coordinates and visibility values in complex number form; calculating visibility data of each observation baseline based on the observation data; the visibility data comprises baseline projection point coordinates, amplitude and phase; arranging the visibility data according to the observation baseline and time sequence to generate a structured visibility table for the at least one target observation object; reading corresponding visibility data from the structured visibility table based on an integral time parameter corresponding to a time scale of the target observation object and a sampling density of the baseline projection point coordinates.

[0007] In an optional embodiment, the steps of obtaining target visibility data by rejecting observation points with abnormal amplitude in the visibility data of each observation baseline, deleting visibility data with abnormal phase exceeding a preset time threshold, shielding visibility data corresponding to abnormal antennas, and filtering visibility data corresponding to a fault time period with abnormal signal strength, comprise: generating a time sequence curve of each observation baseline based on the visibility data of each observation baseline read from the structured visibility table; determining a global mean and amplitude standard deviation of each observation baseline according to the number of observation points in the time sequence curve of each observation baseline and the amplitude of each observation point; determine an amplitude threshold range of each observation baseline based on the global mean and the amplitude standard deviation, and remove abnormal observation points in the each observation baseline whose amplitudes exceed the amplitude threshold range to obtain first visibility data; process the first visibility data by using a central difference method to determine whether a phase gradient of adjacent time points of the each observation baseline exceeds a preset gradient threshold value; if a duration period during which the phase gradient exceeds the preset gradient threshold value exceeds a time threshold value, delete visibility data corresponding to the duration period to obtain second visibility data; mask data corresponding to an abnormal antenna whose amplitude standard deviation exceeds a preset standard deviation threshold value from the second visibility data to obtain third visibility data; filter data corresponding to a failure time period whose signal strength is lower than a preset strength threshold value in the third visibility data to obtain target visibility data.

[0008] In an optional implementation, the step of performing image deconvolution on a dirty map corresponding to the target visibility data to obtain a clean image and a clean model for at least one target observation object, comprises: perform inverse transformation on the target visibility data by using Fourier transformation to obtain a dirty map for at least one target observation object, and dirty lobe information and a synthesized beam information of the dirty map; determine a position coordinate and a peak intensity of a current highest brightness observation point from the dirty map based on an amplitude absolute value, and record the position coordinate of the current highest brightness observation point in the dirty map, and a first Gaussian convolution of the peak intensity of the current highest brightness observation point in the dirty map and the synthesized beam information as a point source model component; subtract a second Gaussian convolution of the dirty lobe information and the synthesized beam information from the dirty map to obtain a first residual map; repeat the following steps until a residual peak value of the first residual map is less than a preset noise threshold value, or terminate when a preset iteration number is reached: determine a position coordinate and a peak intensity of a current highest brightness observation point from the first residual map based on an amplitude absolute value, and record the position coordinate of the current highest brightness observation point in the first residual map, and a first Gaussian convolution of the peak intensity of the current highest brightness observation point in the first residual map and the synthesized beam information as a point source model component; subtract a second Gaussian convolution of the dirty lobe and the synthesized beam information from the first residual map to obtain a second residual map, and take the second residual map as a new first residual map; process the point source model component to obtain a clean model; Add the point source model component and the first residual map output by the iteration termination to determine a clean image removing the dirty lobe interference.

[0009] In an optional embodiment, based on the target antenna gain factor corresponding to the observation baseline solved by the clean model, the observation data of at least one target observation object is corrected, and the step of iteratively optimizing the clean model to determine a target clean model and a correction parameter of the target clean model comprises: Based on the visibility data calculated and output by the clean model, the observation data and the antenna gain factor of at least one target observation object in a preset time period are used to construct a gain model of each observation baseline; Based on the gain model, a least square fitting target function of the observation baseline is constructed, and a target antenna gain factor is solved by nonlinear minimization; The target antenna gain factor is used to correct the observation data to obtain corrected visibility data; Based on the dirty map corresponding to the corrected visibility data, an image deconvolution is performed to determine a current clean model, and the current clean model is used as a new clean model, and the above steps are repeated until the residual error of the clean model converges to terminate, and a target clean model and a correction parameter of the target clean model are determined.

[0010] In an optional embodiment, based on the nonlinear least square method, the target clean model and the correction parameter, the step of determining a preset parameter of at least one target observation object comprises: Using the nonlinear least square method, the actual observed visibility value acted by the correction parameter is fitted to the source structure corresponding to at least one target observation object in the target clean image corresponding to the target clean model in the visibility domain to determine a preset parameter of at least one target observation object.

[0011] In an optional embodiment, according to the preset parameter and the target clean model, the step of automatically reconstructing an image comprises: Based on the visibility data calculated and output by the target clean model and the preset parameter, a clean image of at least one target observation object is reconstructed.

[0012] In a second aspect, the present application provides an image reconstruction device for very long baseline interferometry data, comprising: A batch processing module is used to batch process very long baseline interferometry data of at least one target observation object, arrange a structured visibility table, and read visibility data of each observation baseline from the structured visibility table; A preprocessing module is configured to remove observation points with abnormal amplitudes from the visibility data of each observation baseline, delete visibility data with abnormal phases exceeding a preset time threshold, shield visibility data corresponding to abnormal antennas, and filter visibility data corresponding to fault time periods with abnormal signal strengths, to obtain target visibility data. A deconvolution module is configured to perform image deconvolution on dirty maps corresponding to the target visibility data, to obtain clean images and a clean model for at least one target observation object, which are free of dirty lobe interference. A self-calibration module is configured to correct observation data for at least one target observation object based on target antenna gain factors of the observation baselines solved by the clean model, and iteratively optimize the clean model to determine a target clean model and a correction parameter of the target clean model. A determination module is configured to determine preset parameters of at least one target observation object based on a non-linear least square method, the target clean model, and the correction parameter. A reconstruction module is configured to perform automatic image reconstruction according to the preset parameters and the target clean model.

[0013] In a third aspect, the present application provides an electronic device, which comprises a memory, a processor, and a program stored in the memory and capable of running on the processor, and the processor implements the method according to any one of the preceding embodiments when executing the program.

[0014] In a fourth aspect, the present application provides a computer-readable storage medium, which stores a computer program, and the computer program implements the method according to any one of the preceding embodiments when executed.

[0015] The method and device for image reconstruction of very long baseline interferometry data provided by the embodiments of the present application can batch process and arrange structured visibility tables for very long baseline interferometry data of at least one target observation object, so that visibility data of each observation baseline can be read from the structured visibility tables; filter observation points with various abnormalities and corresponding visibility data in the visibility data of each observation baseline, to obtain target visibility data for subsequent steps, and then perform image deconvolution to obtain clean images and a clean model for at least one target observation object, which are free of dirty lobe interference; solve target antenna gain factors according to the clean model, implement self-calibration of the clean model, extract preset parameters of at least one target observation object based on the self-calibrated target clean model and a correction parameter according to a non-linear least square method; and finally reconstruct clean images of at least one target observation object by using the target clean model and the preset parameters, to ensure the image reconstruction effect of the very long baseline interferometry data.

[0016] Other features and advantages of the present application will be set forth in the descriptions that follow, and in part will be apparent from the description or can be learned by practice of the application. The purposes and other advantages of the application will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.

[0017] In order to make the above objectives, features and advantages of the present application more apparent, the following will specifically describe a preferred embodiment in combination with the accompanying drawings, and the detailed description is as follows. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the specific embodiments of the present application or the technical solutions in the prior art, the following will briefly introduce the drawings needed to be used in the specific embodiments or the prior art description. Obviously, the drawings described below are some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0019] Figure 1 A flow chart of a very long baseline interferometric data image reconstruction method provided by an embodiment of the present application; Figure 2 A UV coverage map formed by observation data provided by an embodiment of the present application; Figure 3 An amplitude schematic diagram under UV coverage provided by an embodiment of the present application; Figure 4 A calibrated data schematic diagram of a time series of a single baseline provided by an embodiment of the present application; Figure 5 A dirty map schematic diagram provided by an embodiment of the present application; Figure 6 A residual map schematic diagram provided by an embodiment of the present application; Figure 7 Another residual map schematic diagram provided by an embodiment of the present application; Figure 8 A clean image schematic diagram provided by an embodiment of the present application; Figure 9 A Gaussian model fitting schematic diagram provided by an embodiment of the present application; Figure 10 A target clean image schematic diagram provided by an embodiment of the present application; Figure 11 A reconstruction map schematic diagram for a target object provided by an embodiment of the present application; Figure 12 A functional module schematic diagram of a very long baseline interferometric data image reconstruction device provided by an embodiment of the present application; Figure 13A hardware architecture schematic diagram of an electronic device provided by an embodiment of the present application is shown. DETAILED DESCRIPTION

[0020] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions of the present application will be described clearly and completely below with reference to the drawings. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0021] Therefore, the image reconstruction method and device for VLBI data provided by the embodiments of the present application can realize more accurate full-process automatic processing from original data calibration, imaging to physical parameter extraction, so as to improve processing efficiency and scientific output ability.

[0022] In order to facilitate the understanding of the embodiments, first, a VLBI data image reconstruction method provided by the embodiments of the present application is described in detail. The method can be applied to automatic control devices such as servers, upper computers and controllers.

[0023] Figure 1 A flowchart of the VLBI data image reconstruction method provided by the embodiments of the present application is shown.

[0024] Reference Figure 1 The method comprises the following steps: In step S102, the VLBI data for at least one target observation object is batch-processed, a structured visibility table is generated, and visibility data of each observation baseline is read from the structured visibility table.

[0025] In the embodiments of the present application, a target observation object is taken as an example, and the target observation object is a celestial object observed in a radio wave band.

[0026] Exemplarily, step S102 can be implemented by the following steps, comprising: In step 1.1, the VLBI data for at least one target observation object is batch-processed to obtain observation data for characterizing signal intensity and phase of each observation baseline in length and direction respectively.

[0027] In radio interferometry, the visibility value is a Fourier transform sample point of the brightness distribution of an astronomical object, each visibility is a complex number, representing the signal strength and phase at a specific baseline length and direction. The observation data consists of a set of (u, v, w) spatial frequency coordinates and complex number form visibility values (V), u, v are spatial frequencies in units of observation wavelength, w is the baseline component perpendicular to the observation plane. UVFITS is a standard file format used to store VLBI / VLA interferometric data, recording baseline number, time stamp, frequency channel, polarization mode, etc. Using the preset data processing software can batch extract the key fields (observation data) from the very long baseline interferometry data; such as reading the (u, v) coordinates and the corresponding complex visibility values

real part real, imaginary part imag

[0028] Step 1.2, based on the observation data, calculate the visibility data of each observation baseline according to the following formula; the visibility data includes the baseline projection point coordinates, amplitude and phase.

[0029] After analyzing the observation data, convert u, v to appropriate units (usually in units of λ), and also to kilo-lambda (kλ), visualize the coordinates: (u, v), the data point position is the baseline projection point, as shown in Figure 2 ; use the following formula to calculate other display information such as: amplitude , phase , amplitude information in the UV coverage case as shown in Figure 3 , the subsequent step can detect the observation point with amplitude anomaly according to the amplitude shown in Figure 3

[0030] Step 1.3, arrange the visibility data according to the observation baseline and time sequence, and generate a structured visibility table for at least one target observation object.

[0031] Based on the visibility data, and antenna baseline, frequency channel, time stamp, polarization mode, etc. Metadata jointly constitute the visibility table; and arrange the complex visibility information of each baseline in time sequence to generate a structured "visibility table" for subsequent processing.

[0032] ​Time ordering: all records are sorted in ascending order by time (UTC or MJD); Baseline identification: each row of records uniquely identifies an antenna pair, such as (S1-S2); Complex visibility unfolding: Table 1 below is an example of specific data of a complex visibility table, which shows the visibility data structure of different antenna pairs according to the increase of time. The main sorting is MJD, which is the simplified Julian day, and UTC is used as auxiliary information. Each row is the unique antenna pair data at a specific time point. The same antenna pair will appear later, but the time is unique. In this way, the visibility data arrangement is realized.

[0033] Table 1 is a visibility table

[0034] Step 1.4, based on the time scale of the target observation object and the integral time parameter corresponding to the sampling density of the baseline projection point coordinates, read the corresponding visibility data from the structured visibility table.

[0035] In the process of reading the visibility data, a time resampling mechanism is introduced, and appropriate integral time parameters are selected according to the time scale of the target source structure corresponding to the target observation object and the uv sampling density.

[0036] In VLBI observation, the time granularity of the original observation data is particularly small, which can reach millisecond level. For data processing, too short integration time may result in a lot of redundant data and low signal-to-noise ratio (SNR). Too long time integration may cause distortion of the signal, especially for high-speed changing or complex structure radio sources. The method of time resampling is to select a reasonable time, and the phase is generally controlled within 1 radian, and the signal amplitude attenuation is less than 2%. For low frequency L band, 1.4 GHz, the integration time is 10-30 s; for high frequency 22 GHz, 1-5 s.

[0037] It should be noted that the target observation object corresponding to the target observation object, i.e. the target source structure, can be divided into two types: if it is a complex structure target (AGN jet, black hole shadow edge), it needs to be integrated to maintain the structural details. If it is a simple structure target (bright point source, such as pulsar), the integration time can be appropriately lengthened.

[0038] Step S104, the visibility data of each observation baseline is filtered, the phase abnormality of the visibility data exceeding the preset time threshold is deleted, the visibility data corresponding to the abnormal antenna pair is shielded, and the visibility data corresponding to the fault time period of the signal strength anomaly is filtered, to obtain the target visibility data.

[0039] In some embodiments, the read visibility data is screened to ensure the accuracy of subsequent reconstruction map, and the step S104 includes: Step 2.1, based on the visibility data of each observation baseline read from the structured visibility table, generate a time series curve of each observation baseline.

[0040] wherein the observation baseline can be understood as two antennas, and the visibility data of each observation baseline can be understood as the visibility data read between two different antennas, and the visibility data of each observation baseline at different time series are different.

[0041] Step 2.2, according to the number of observation points in the time series curve of each observation baseline and the amplitude of each observation point, determine the global mean and amplitude standard deviation of each observation baseline.

[0042] Based on the visibility data, a time series curve A(t) is generated for each baseline, and its global mean μ and amplitude standard deviation σ are calculated according to the following formula:

[0043]

[0044] wherein N is the number of observation points in the time series curve corresponding to each baseline, and Amp is the amplitude of the time series curve corresponding to each baseline.

[0045] Step 2.3, based on the amplitude threshold range of each observation baseline determined by the global mean and amplitude standard deviation, remove the abnormal observation points in each observation baseline whose amplitude exceeds the amplitude threshold range to obtain first visibility data; Based on the global mean μ and amplitude standard deviation σ, the amplitude threshold range [μ-kσ, μ+kσ] (preferably k=3) can be determined; observation points exceeding this threshold range are marked as abnormal to achieve the screening of amplitude abnormal observation points. Wherein, if the antenna system fails, the data processing is abnormal, the atmosphere ionosphere is affected, the external radio frequency interference is affected, etc. will cause amplitude anomaly, in order to ensure the accuracy of the subsequent reconstructed image, the observation points of such amplitude anomaly are removed.

[0046] In addition, if there is a strong source, a sliding window local statistics can be introduced to improve robustness.

[0047] It should be noted that the strong source can be understood as a point source or compact structure with particularly high brightness (brightness exceeding a general preset threshold) in the observation field. In the visibility curve, the visibility value of the strong source will be significantly higher than the average. When there is a strong source, a sliding window is introduced to delete the visibility data of the observation points with amplitude anomaly in the preset time period, so as to avoid affecting the subsequent imaging operation.

[0048] Step 2.4, the first visibility data is processed by using the central difference method to judge whether the phase gradient of adjacent time points of each observation baseline exceeds a preset gradient threshold; Here, the phase gradient of adjacent time points is calculated in a central difference manner; it is checked whether the jump of the phase gradient of adjacent time points exceeds a reasonable physical threshold (such as 90° or π radians). Where the jump is caused by the influence of atmospheric turbulence, the influence of the ionosphere, the loss of lock of the local oscillator of the device, and the like, the visibility data of the phase anomaly is processed.

[0049] Step 2.5, if the phase gradient exceeds the preset gradient threshold for a duration period exceeding the time threshold, the visibility data corresponding to the duration period is deleted to obtain second visibility data.

[0050] If the continuous jump exceeds the set duration period (time threshold), it is marked as systematic phase drift, and the entire data needs to be removed.

[0051] Step 2.6, the data corresponding to the abnormal antenna whose amplitude standard deviation exceeds the preset standard deviation threshold is shielded from the second visibility data to obtain third visibility data. All visibility data participating in a single antenna (baseline containing the antenna) is viewed, such as Figure 4 If the amplitude standard deviation is much larger than the preset standard deviation threshold / network average value (> 2xσ), it is judged as an "abnormal antenna"; all visibility data related to the antenna is shielded.

[0052] Step 2.7, the data corresponding to the fault time period whose signal strength is lower than the preset strength threshold is filtered from the third visibility data to obtain target visibility data.

[0053] Here, the SNR signal-to-noise ratio of each time period is analyzed in units of observation time; if the signal strength is much lower than the set threshold (such as below the system noise baseline), it is determined as a fault time period; the data corresponding to the fault time period is filtered in a time window manner.

[0054] In addition, cross verification can also be combined with the antenna lifting events in the system log; The antenna lifting event is often used to represent the period of time when the antenna can observe signals; the visibility data is verified according to the antenna lifting event in the system log, which can be understood as a system observation log used to save various information during observation; for example, when checking the quality of the visibility data, if the visibility data of a certain period of time is abnormal, the antenna lifting event in the system observation log is checked at this time to learn the signal observable period and the antenna available state, so as to exclude the abnormal visibility data caused by the antenna lifting event (signal fault); for example, if the visibility data of the current period of time is abnormal, it is learned from the system log that the antenna is unavailable, and the specific information in the log is as follows: ANT14 status: DOWN @ 02:47:00 ~ 02:55:15; therefore, the abnormal visibility data is caused by the antenna being unavailable, rather than a problem with the signal itself.

[0055] In step S106, the dirty map corresponding to the target visibility data is image deconvolved to obtain a clean image and a clean model for at least one target observation object, which removes the interference of the dirty lobe.

[0056] The target visibility data (binary FITS format) after the foregoing preprocessing is deconvolved, the main lobe influence is gradually stripped, the signal is decomposed into a series of point source models, and thus a clean image and a clean model that restores the brightness distribution of celestial bodies are obtained.

[0057] In some embodiments, in order to remove the main lobe and side lobe effects caused by the synthesized beam, obtain a clean image and a clean model that restores the true brightness distribution of celestial bodies, the following steps can be implemented: Step 3.1, the target visibility data is deconvolved by using Fourier transform to obtain a dirty map for at least one target observation object, and dirty lobe information and synthesized beam information of the dirty map.

[0058] Here, the UV visibility is deconvolved by using Fourier transform FFT to generate a dirty map DirtyImage, and to obtain dirty lobe information and synthesized beam (PSF) information corresponding to the dirty map; the synthesized beam information is usually saved together with the image, and the synthesized beam information includes the FITS header standard keywords (BMAJ, BMIN, BPA): BMAJ: BeamMajorAxis (major axis length), unit degree.

[0059] BMIN: BeamMinorAxis (minor axis length), unit degree.

[0060] BPA: BeamPositionAngle, main axis orientation, unit degree (measured from north to east).

[0061] Step 3.2: Determine the position coordinates and peak intensity of the current highest brightness observation point from the dirty image based on the absolute value of the amplitude, and record the position coordinates of the current highest brightness observation point in the dirty image, as well as the peak intensity of the current highest brightness observation point in the dirty image and the first Gaussian convolution of the synthetic beam information as point source model components.

[0062] like Figure 5 As shown, the highest brightness observation point, i.e., the center of the bright source, can be found from the dirty map using a box, which serves as the clean area.

[0063] Step 3.3: Subtract the second Gaussian convolution of the dirty lobe and the synthesized beam information from the dirty map to obtain the first residual map.

[0064] in, Figure 6 To remove Figure 5 The first residual map of the current observation point with the highest brightness is used to extract the flux of the selected area (the flux is the radiative flux of the peak height point in the dirty map, in Jansky, which represents the electromagnetic radiation intensity of the point on the sky plane) and establish a visibility model (a visibility mathematical model composed of a series of model components, such as the Gaussian components of the point source). Figure 6 The plus sign indicates the delta function flow model extracted by the clean algorithm.

[0065] It should be noted that the clean algorithm extracts the radiative flux at the peak height point to construct a flux model (point source model); the flux model is subjected to Fourier transform to determine the visibility of each point source, and then a visibility model is synthesized. The residual between the visibility model and the observed data (observed visibility data) is used to fit and self-calibrate the visibility model; the calibrated visibility model is convolved with the synthesized beam to generate the final clean image.

[0066] Step 3.4: Repeat steps ab until the peak value of the residual in the first residual plot is less than the preset noise threshold, or until the preset number of iterations is reached. a. Based on the absolute value of the amplitude, determine the position coordinates and peak intensity of the current highest brightness observation point from the first residual map, and record the position coordinates of the current highest brightness observation point in the first residual map, as well as the peak intensity of the current highest brightness observation point in the first residual map and the first Gaussian convolution of the synthetic beam information as point source model components; b. Subtract the second Gaussian convolution of dirty lobe and synthetic beam information from the first residual map to obtain the second residual map, and use the second residual map as the new first residual map.

[0067] Loop iteration process: calculate the position and intensity of the highest brightness point in the image; take it as a "real point source" and record its position and the first Gaussian convolution as a model component; subtract it from the image with a scaled version of the main lobe (second Gaussian convolution) (optional loop gain γ ≈ 0.1-0.2); update the residual image, as shown in the residual dirty map after multiple "box extraction-modeling" iterations, and the plus sign indicates the flux model; if the residual peak value of the updated residual image is lower than the noise threshold or reaches the maximum iteration number, exit the iteration process. Figure 7

[0068] Step 3.5, process the point source model component to obtain a clean model.

[0069] Here, the position coordinates of each highest brightness observation point and the corresponding first Gaussian convolution recorded in the preceding steps are taken as the point source model component, and each point source component is combined to obtain a clean model, which is an initial model at this time.

[0070] Step 3.6, add the point source model component and the first residual image output by the iteration termination to determine a clean image that removes the interference of dirty lobes, as shown in the completed flux model extraction clean image. Figure 8

[0071] Add the point source model component to the residual image to obtain a final FITS format clean image, a CLEAN model file, and a processing log.

[0072] Step S108, based on the target antenna gain factor corresponding to the observation baseline solved by the clean model, correct the observation data of at least one target observation object, and iteratively optimize the clean model to determine the target clean model and the correction parameter of the target clean model.

[0073] In some embodiments, the clean model determined in the preceding steps is taken as an initial model to calibrate the phase and amplitude of the antenna gain, assuming that the gain is stable during this period; group the visibility data on all baselines to solve a set of gain parameters; calculate the visibility data Vmodel output by the initial model; minimize the residual between the visibility data Vmodel and the observation data to update the gain estimate; update the observation visibility by applying gain correction; regenerate a better CLEAN model based on the observation data; recalibrate with the new clean model; iterate until convergence (the residual no longer decreases significantly), which can significantly improve the consistency and signal-to-noise ratio of the subsequent reconstructed image. The step S108 can include: Step 4.1, based on the visibility data calculated by the clean model, construct a gain model for each observation baseline for the observation data of at least one target observation object in a preset time period and the antenna gain factor; ​​The preset time period is the set time interval (solutioninterval) (Δt\Δ~tΔt). Assuming visibility consists of the true value and the error term introduced by the antenna gain, the gain model is set as follows:

[0074] Among them: RV ij The actual observed visibility, i.e., the observation data; V ij : The "true value" of the model calculation output; Gi, Gj: The complex gain (amplitude + phase) of antenna i and antenna j, i.e., antenna gain factors; Thermal noise and residual terms are determined based on receiver parameters and observed frequency bands.

[0075] Step 4.2: Construct the least-squares fitting objective function for the observation baseline based on the gain model, and solve the target antenna gain factor by nonlinear minimization.

[0076] Construct a least-squares fitting objective function for all baselines ( ):

[0077] The target antenna gain factors Gi and Gj are solved by nonlinear minimization, and the phase or complex gain (amp+phase) of all antennas is updated.

[0078] Step 4.3: Correct the observation data using the target antenna gain factor to obtain corrected visibility data; The target antenna gain factor will apply regularization or ignore observation points in observation data with insufficient SNR; the target antenna gain factor will be used to eliminate systematic bad data (such as excessive closure error); for example, the phase 1+2+3 of the three antennas should be 0. If the closure phase is not 0, such as greater than 10°, it is considered that a phase deviation has occurred, and manual judgment is required to determine whether the observation data corresponding to this phase is bad data.

[0079] In addition, a gain table is automatically recorded after each Self-Cal self-calibration to trace the calibration process. The gain table records the complex gain solution for each antenna and each time interval, including amplitude gain and phase gain. It is the antenna calibration history of VLBI data, similar to a separately saved calibration parameter file (GainSolutionTable).

[0080] Step 4.4: Perform image deconvolution based on the dirty image corresponding to the corrected visibility data to determine the current cleaning model, and use the current cleaning model as the new cleaning model.

[0081] Figure 9 Corresponding to the initialization stage of model fitting, located after self-calibration and before model fitting, it describes the action of the subsequent model fitting stage of the CLEAN algorithm, visualizes this process, and shows the process from the point source model to the real structure modeling. Specifically, Figure 9 It can be shown that in the cleaned image after self-calibration, a circularly symmetric Gaussian model component is initialized at the position of the peak value of the target source after removing the original CLEAN model, as the starting point for subsequent structure parameter fitting.

[0082] Repeat steps 4.1-4.4 above until the residual error of the cleaned model converges, and determine the target cleaned model and the correction parameters of the target cleaned model.

[0083] Using the existing source structure cleaned model, the complex gain error / target antenna gain factor (phase and amplitude) of each antenna is iteratively solved by minimizing the deviation between the observed data and the model output visibility data, and these gain corrections are applied to the observed data, and the corrected observed data is used to realize self-calibration of the cleaned model.

[0084] Figure 10 An image result after model fitting is shown-cleaned image, i.e. cleaned image after structure parameter completion, which replaces the delta function model (point source model) used by Clean before with a Gaussian model closer to the real distribution of the source in the sky plane. The parameters of the Gaussian model are more suitable for scientific research.

[0085] Step S110, based on the nonlinear least squares method, the target cleaned model and the correction parameters, determining the preset parameters of at least one target observation object.

[0086] In some embodiments, in view of the characteristics of complex brightness distribution structure and incomplete spatial frequency coverage commonly seen in VLBI observation, source structure modeling and least squares optimization are performed directly in the visibility (uv) domain, so as to accurately estimate the position, size, shape and brightness of the celestial substructure (target observation object) and other preset parameters. This step S110 includes: Using the nonlinear least squares method, the actual observed complex visibility values are fitted to the source structure corresponding to at least one target observation object in the target cleaned image of the target cleaned model in the visibility domain, to determine the preset parameters of at least one target observation object.

[0087] Specifically, the theoretical visibility distribution is calculated by Fourier inversion of the observed visibility, and then the key parameters of the celestial source structure are extracted, such as the preset parameters of position, flux, size, axis ratio and direction angle, etc. Here, taking the identification of the key parameter peak flux in the dirty map as an example: first, find the peak bright spot in the dirty map; record the brightness value of the position; regard it as a "point source" and assign it as flux; then subtract the product of the point and the synthesized beam (PSF) to obtain the residual map; repeat the iteration until the residual peak value is lower than the threshold. All point source model components are added to the flux to obtain the total flux flux of the image.

[0088] The least squares fitting process is to construct a fitting target function; use the nonlinear least squares method to solve the optimal parameters, model each source structure in the image as a point source or Gaussian distribution, fit the actual observed complex visibility Vij to the synthesis result of several ideal model sources in uv space, and extract the structure parameters (position, size, brightness, etc.). The target function is a residual function, which is used to represent the error metric to be minimized, that is, the sum of the squares of all data points. It can be understood that, first, a structure model is defined, such as an elliptical Gaussian; this model corresponds to model parameters such as position x, y, flux Flux, major axis major, minor axis minor, and direction angle pa; the theoretical visibility data in UV space is obtained by FFT change; then compare the observed value and the model value; the sum of the squares of all differences constitutes the target function (residual function); minimizing this value can obtain the optimal model parameters.

[0089] Among them, the preset parameters specifically include: position: (x, y) usually in units of mas (milliarcsecond); total brightness (FluxDensity): S, unit Jy; size: FWHM (Full Width at Half Maximum); shape (if elliptical Gaussian): major axis, minor axis, axis ratio; direction angle (PositionAngle).

[0090] Source structure includes: point source (delta function): used to fit dense sources; Gaussian component (Gaussian component): used to simulate diffuse sources or jet expansion structure; elliptical Gaussian (Elliptical Gaussian): has the ability to model directionality and structural asymmetry.

[0091] Step S112, according to the preset parameters and the target cleaning model, performing automatic image reconstruction.

[0092] Here, the reconstruction step S112 includes: based on the visibility data and the preset parameters calculated by the target cleaning model, reconstructing a cleaning image for at least one target observation object, and the reconstructed cleaning image is as shown in Figure 11 .

[0093] The image reconstruction is realized by fitting the extracted parameters based on the visibility model and the image corresponding to the target cleaning model obtained by the self-calibration optimization; in the image reconstruction stage, all the structural parameters (including position, flux, shape, etc.) extracted in the target cleaning model are converted into the brightness distribution in the image space, each point source model component (such as a Gaussian function) is drawn as a two-dimensional brightness block at its corresponding position, and all the blocks corresponding to the point source model components are superimposed to generate the final cleaning image. The VLBI automatic imaging processing method and the matching software system provided by the embodiment of the present application can read the related original VLBI data to automatically calibrate and map, generate fits format astronomical image data that can be used for specific analysis, so as to adjust or give an early warning to the corresponding antenna equipment, and can be used for batch processing of VLBI data, quality control of data, image deconvolution, self-calibration, model fitting, automatic image reconstruction and the like to improve the processing efficiency and reliability. The method is suitable for large-scale interferometric array (such as EHT, CVN) data processing scenarios.

[0094] In some embodiments, as shown in FIG. 1, Figure 12 The embodiment of the present application provides an image reconstruction device of very long baseline interferometry data, which comprises: A batch processing module is configured to batch process very long baseline interferometry data of at least one target observation object, arrange a structured visibility table, and read visibility data of each observation baseline from the structured visibility table; A preprocessing module is configured to remove observation points with abnormal amplitudes in the visibility data of each observation baseline, delete visibility data with abnormal phases exceeding a preset time threshold, shield visibility data corresponding to abnormal antennas, filter visibility data corresponding to abnormal signal strength time periods, and obtain target visibility data; A deconvolution module is configured to perform image deconvolution on a dirty map corresponding to the target visibility data to obtain a clean image and a clean model of the at least one target observation object without dirty lobe interference; A self-calibration module is configured to correct observation data of the at least one target observation object based on target antenna gain factors of the observation baselines solved by the clean model, and iteratively optimize the clean model to determine a target clean model and a correction parameter of the target clean model; A determination module is configured to determine a preset parameter of the at least one target observation object based on a nonlinear least squares method, the target clean model and the correction parameter; A reconstruction module is configured to perform automatic image reconstruction according to the preset parameter and the target clean model.

[0095] Further, the batch processing module is specifically configured to batch process the very long baseline interferometry data of the at least one target observation object to obtain observation data for characterizing signal strength and phase of each observation baseline in length and direction respectively; wherein the observation data includes spatial frequency coordinates and visibility values in complex form; based on the observation data, visibility data of each observation baseline is calculated; the visibility data includes baseline projection point coordinates, amplitude and phase; the visibility data is arranged according to the observation baseline and time sequence to generate a structured visibility table for the at least one target observation object; based on the time scale of the target observation object and the integral time parameter corresponding to the sampling density of the baseline projection point coordinates, corresponding visibility data is read from the structured visibility table.

[0096] Further, the preprocessing module is specifically configured to generate a time sequence curve of each observation baseline based on the visibility data of each observation baseline read from the structured visibility table; determine the global mean and amplitude standard deviation of each observation baseline according to the number of observation points in the time sequence curve of each observation baseline and the amplitude of each observation point; determine the amplitude threshold range of each observation baseline based on the global mean and the amplitude standard deviation, and remove abnormal observation points in the amplitude of each observation baseline exceeding the amplitude threshold range to obtain first visibility data; process the first visibility data using a central difference method to determine whether the phase gradient of adjacent time points of each observation baseline exceeds a preset gradient threshold; if the duration of the phase gradient exceeding the preset gradient threshold exceeds a time threshold, delete the visibility data corresponding to the duration to obtain second visibility data; shield data corresponding to abnormal antennas whose amplitude standard deviation exceeds a preset standard deviation threshold from the second visibility data to obtain third visibility data; filter data corresponding to a fault time period whose signal strength is lower than a preset strength threshold from the third visibility data to obtain target visibility data.

[0097] Further, the deconvolution module is specifically configured to: perform inverse transformation on the target visibility data by using Fourier transformation to obtain a dirty map of at least one target observation object and dirty lobe information and a synthesized beam of the dirty map; determine the position coordinates and peak intensity of a current highest brightness observation point in the dirty map based on the absolute value of the amplitude, and record the position coordinates of the current highest brightness observation point in the dirty map, the peak intensity of the current highest brightness observation point in the dirty map, and first Gaussian convolution of the synthesized beam as a point source model component; subtract second Gaussian convolution of the dirty lobe and the synthesized beam from the dirty map to obtain a first residual map; repeatedly perform the following steps until the residual peak value of the first residual map is less than a preset noise threshold or the preset number of iterations is reached: determine the position coordinates and peak intensity of a current highest brightness observation point in the first residual map based on the absolute value of the amplitude, and record the position coordinates of the current highest brightness observation point in the first residual map, the peak intensity of the current highest brightness observation point in the first residual map, and first Gaussian convolution of the synthesized beam as a point source model component; subtract second Gaussian convolution of the dirty lobe and the synthesized beam from the first residual map to obtain a second residual map, and take the second residual map as a new first residual map; process the point source model component to obtain a clean model; and add the point source model component and the first residual map output by iteration termination to determine a clean image with the interference of the dirty lobe removed.

[0098] Further, the self-calibration module is specifically configured to: based on the visibility data calculated by the clean model, construct a gain model of each observation baseline for observation data and an antenna gain factor of a preset time period of at least one target observation object; construct a least squares fitting target function of the observation baseline based on the gain model, and solve a target antenna gain factor by nonlinear minimization; correct the observation data by using the target antenna gain factor to obtain corrected visibility data; perform image deconvolution on a dirty map corresponding to the corrected visibility data to determine a current clean model, take the current clean model as a new clean model, and repeat the above steps until the residual of the clean model converges, to determine a target clean model and a correction parameter of the target clean model.

[0099] Further, the determination module is specifically configured to: use a nonlinear least squares method to fit the actual observed visibility value of the correction parameter to a source structure corresponding to at least one target observation object in a target clean image corresponding to the target clean model in the visibility domain, to determine a preset parameter of at least one target observation object.

[0100] Further, the reconstruction module is specifically configured to reconstruct the cleaning image for the at least one target observation object based on the visibility data calculated by the target cleaning model and the preset parameter.

[0101] The electronic device can be, but is not limited to, a personal computer (PC), a notebook computer, a monitoring device, a server, and the like.

[0102] As an exemplary embodiment, refer to Figure 13 The electronic device 110 includes a communication interface 111, a processor 112, a memory 113, and a bus 114, the processor 112, the communication interface 111, and the memory 113 are connected through the bus 114; the memory 113 is configured to store a computer program supporting the processor 112 to execute the above method, and the processor 112 is configured to execute the program stored in the memory 113.

[0103] The machine-readable storage medium mentioned herein can be any electronic, magnetic, optical, or other physical storage device, and can contain or store information such as executable instructions, data, and the like. For example, the machine-readable storage medium can be a RAM (Random Access Memory), a volatile memory, a non-volatile memory, a flash memory, a storage drive (such as a hard disk drive), any type of storage disk (such as an optical disk, a DVD, etc.), or similar storage medium, or a combination thereof.

[0104] The non-volatile medium can be a non-volatile memory, a flash memory, a storage drive (such as a hard disk drive), any type of storage disk (such as an optical disk, a DVD, etc.), or similar non-volatile storage medium, or a combination thereof.

[0105] It can be understood that the specific operation method of each functional module in the embodiment can refer to the detailed description of the corresponding steps in the above method embodiment, which will not be repeated here.

[0106] The computer-readable storage medium provided by the embodiment of the present application stores a computer program, and the computer program code can implement the method described in any of the above embodiments when executed, and the specific implementation can refer to the method embodiment, which will not be repeated here.

[0107] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system and device can refer to the corresponding process in the foregoing method embodiment, which will not be repeated here.

[0108] In addition, in the description of the embodiments of the present application, unless otherwise explicitly specified and limited, the terms "mounting", "connection", "connecting" should be understood in a broad sense, for example, can be fixedly connected, or can be detachably connected, or integrally connected; can be mechanically connected, or can be electrically connected; can be directly connected, or can be indirectly connected through an intermediate medium; can be internal communication of two elements. For those skilled in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances.

[0109] In the description of the present application, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer" and the like indicate the orientation or positional relationship shown in the drawings, and are only for the purpose of facilitating the description of the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, the terms "first", "second", "third" are only for the purpose of description, and cannot be understood as indicating or implying relative importance.

[0110] Finally, it should be noted that: the above-described embodiments are only specific embodiments of the present application, used to illustrate the technical solutions of the present application, and are not limited thereto, the protection scope of the present application is not limited thereto, although the present application 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 within the technical scope disclosed by the present application can still modify or easily think of changes to the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to part of the technical features; and these modifications, changes or replacements do not make the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be covered within the protection scope of the present application.

Claims

1. An image reconstruction method for very long baseline interferometry data, characterized in that, include: Batch processing of very long baseline interferometry data for at least one target observation object, arranging to generate a structured visibility table, and reading the visibility data of each observation baseline from the structured visibility table; The observation points with abnormal amplitude in the visibility data of each observation baseline are removed, the visibility data with abnormal phase exceeding the preset time threshold is deleted, the visibility data corresponding to the abnormal antenna is shielded, and the visibility data corresponding to the fault time period with abnormal signal strength is filtered to obtain the target visibility data. The dirty image corresponding to the target visibility data is deconvolved to obtain a clean image and a clean model for at least one target observation object, which removes dirty lobe interference. Based on the target antenna gain factor corresponding to the observation baseline obtained from the cleaning model, the observation data for at least one target observation object is corrected, and the cleaning model is iteratively optimized to determine the target cleaning model and the correction parameters of the target cleaning model; Based on the nonlinear least squares method, the target cleaning model and the correction parameters determine at least one preset parameter of the target observation object; Automated image reconstruction is performed based on the preset parameters and the target cleaning model.

2. The method according to claim 1, characterized in that, The steps of batch processing very long baseline interferometry data for at least one target observation object, arranging and generating a structured visibility table, and reading the visibility data of each observation baseline from the structured visibility table include: Very long baseline interferometry data for at least one target observation object are batch-processed to obtain observation data characterizing the signal strength and phase of each observation baseline in length and direction, respectively; wherein, the observation data includes spatial frequency coordinates and visibility values ​​in complex form; Based on the observation data, visibility data for each observation baseline is calculated; the visibility data includes the coordinates of the baseline projection point, amplitude, and phase. The visibility data is arranged according to the observation baseline and time series to generate a structured visibility table for the at least one target observation object; Based on the time scale of the target observation object and the integral time parameter corresponding to the sampling density of the baseline projection point coordinates, the corresponding visibility data is read from the structured visibility table.

3. The method according to claim 1, characterized in that, The steps of removing observation points with abnormal amplitude from the visibility data of each observation baseline, deleting visibility data with abnormal phase exceeding a preset time threshold, shielding visibility data corresponding to abnormal antennas, and filtering visibility data corresponding to fault time periods with abnormal signal strength to obtain target visibility data include: Based on the visibility data of each observation baseline read from the structured visibility table, a time series curve for each observation baseline is generated; Based on the number of observation points in the time series curve of each observation baseline and the amplitude of each observation point, determine the global mean and amplitude standard deviation of each observation baseline; Based on the global mean and the amplitude standard deviation, the amplitude threshold range of each observation baseline is determined, and abnormal observation points in each observation baseline whose amplitude exceeds the amplitude threshold range are removed to obtain the first visibility data; The first visibility data is processed using a central difference method to determine whether the phase gradient of each observation baseline at adjacent time points exceeds a preset gradient threshold. If the duration of the phase gradient exceeding the preset gradient threshold exceeds the time threshold, then the visibility data corresponding to the duration is deleted to obtain the second visibility data; The data corresponding to abnormal antennas whose amplitude standard deviation exceeds a preset standard deviation threshold are masked out of the second visibility data to obtain the third visibility data; The data corresponding to the fault time period in the third visibility data where the signal strength is lower than the preset strength threshold is filtered to obtain the target visibility data.

4. The method according to claim 1, characterized in that, The steps of performing image deconvolution on the dirty map corresponding to the target visibility data to obtain a clean image and a clean model for at least one target observation object with dirty lobe interference removed include: The target visibility data is inversely transformed using Fourier transform to obtain a dirty map for at least one target observation object, as well as dirty lobe information and synthetic beam information of the dirty map; Based on the absolute value of the amplitude, the position coordinates and peak intensity of the current highest brightness observation point are determined from the dirty map, and the position coordinates of the current highest brightness observation point in the dirty map, the peak intensity of the current highest brightness observation point in the dirty map, and the first Gaussian convolution of the synthetic beam information are recorded as point source model components. The first residual map is obtained by subtracting the second Gaussian convolution of the dirty lobe information and the synthesized beam information from the dirty map. Repeat the following steps until the peak value of the residual in the first residual map is less than a preset noise threshold, or the preset number of iterations is reached: Determine the position coordinates and peak intensity of the current highest brightness observation point from the first residual map based on the absolute value of the amplitude, and record the position coordinates of the current highest brightness observation point in the first residual map, the peak intensity of the current highest brightness observation point in the first residual map, and the first Gaussian convolution of the synthetic beam information as point source model components; Subtract the second Gaussian convolution of the dirty lobe and the synthetic beam information from the first residual map to obtain the second residual map, and use the second residual map as the new first residual map; The point source model components are processed to obtain a clean model; The point source model components and the first residual map of the iteration termination output are added together to determine a clean image with dirty lobe interference removed.

5. The method according to claim 1, characterized in that, The steps of correcting observation data for at least one target observation object based on the target antenna gain factor corresponding to the observation baseline obtained from the clean model, and iteratively optimizing the clean model to determine the target clean model and the correction parameters of the target clean model include: Based on the visibility data calculated from the clean model, a gain model for each observation baseline is constructed using the observation data and antenna gain factor for a preset time period of at least one target observation object. Based on the gain model, a least-squares fitting objective function for the observation baseline is constructed, and the target antenna gain factor is obtained by nonlinear minimization. The observation data is corrected using the target antenna gain factor to obtain corrected visibility data; Based on the dirty image corresponding to the corrected visibility data, perform image deconvolution to determine the current cleaning model. Use the current cleaning model as the new cleaning model and repeat the above steps until the residual of the cleaning model converges. Then, determine the target cleaning model and the correction parameters of the target cleaning model.

6. The method according to claim 1, characterized in that, Based on the nonlinear least squares method, the step of determining at least one preset parameter of the target observation object using the target cleaning model and the correction parameters includes: Using the nonlinear least squares method, the visibility value of the actual observation under the action of the correction parameter is fitted to the source structure corresponding to at least one target observation object in the target clean image corresponding to the target clean model in the visibility domain, and the preset parameters of at least one target observation object are determined.

7. The method according to claim 1, characterized in that, The steps for automated image reconstruction based on the preset parameters and the target cleaning model include: Based on the visibility data calculated by the target cleaning model and the preset parameters, a cleaning image for at least one target observation object is reconstructed.

8. An image reconstruction apparatus for very long baseline interferometry data, characterized in that, include: The batch processing module processes very long baseline interferometry data for at least one target observation object in batches, arranges them to generate a structured visibility table, and reads the visibility data of each observation baseline from the structured visibility table; The preprocessing module removes observation points with abnormal amplitude from the visibility data of each observation baseline, deletes visibility data with abnormal phase exceeding a preset time threshold, shields visibility data corresponding to abnormal antennas, and filters visibility data corresponding to fault time periods with abnormal signal strength, thereby obtaining target visibility data. The deconvolution module performs image deconvolution on the dirty image corresponding to the target visibility data to obtain a clean image and a clean model for at least one target observation object with dirty lobe interference removed. The self-calibration module, based on the target antenna gain factor corresponding to the observation baseline obtained by the cleaning model, corrects the observation data for at least one target observation object, and iteratively optimizes the cleaning model to determine the target cleaning model and the correction parameters of the target cleaning model; The determination module, based on the nonlinear least squares method, the target cleaning model, and the correction parameters, determines at least one preset parameter of the target observation object; The reconstruction module performs automated image reconstruction based on the preset parameters and the target cleaning model.

9. An electronic device, characterized in that, It includes a memory, a processor, and a program stored in the memory and capable of running on the processor, wherein the processor executes the program to implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The readable storage medium stores a computer program that, when executed, implements the method described in any one of claims 1-7.

Images