CN-122017836-A - Kalman filtering-based multi-source InSAR data space-time fusion method and system

CN122017836ACN 122017836 ACN122017836 ACN 122017836ACN-122017836-A

Abstract

The invention discloses a multi-source InSAR data space-time fusion method and a system based on Kalman filtering, wherein the method comprises the steps of unifying spatial reference standards of InSAR deformation data of different satellite platforms and matching homonymous coherent points; based on the unified space reference standard, utilizing the InSAR time sequence data set to establish a rate displacement matrix of the fusion time sequence and calculate the displacement at each moment, and utilizing a Kalman filtering algorithm to dynamically fuse the multi-source InSAR data to obtain the dynamic evolution time sequence of the observation target. According to the invention, by unifying multisource InSAR space-time references and introducing Kalman filtering dynamic fusion, the limitations of single-platform data deletion, poor coherence, sensor deviation and the like are overcome, and the accuracy, time resolution and continuity of surface deformation monitoring are remarkably improved. The invention realizes the optimal integration of heterogeneous observation information, has self-adaptive weighting and error suppression capability, provides an extensible technical paradigm for engineering deformation monitoring, and has stronger practical value and popularization prospect.

Inventors

SONG CHUANG
WEN FAN
LI ZHENHONG
CHEN YI
ZHANG LIN
XU HONG
YU CHEN
Zou Jinchao

Assignees

长安大学
中铁一局集团有限公司

Dates

Publication Date: 20260512
Application Date: 20260202

Claims (6)

1. A Kalman filtering-based multi-source InSAR data space-time fusion method is characterized by comprising the following steps of: S1, unifying spatial reference benchmarks of InSAR deformation data of different satellite platforms, wherein the spatial reference benchmarks comprise projecting LOS deformation to a vertical direction and matching same-name coherent points; s2, based on the unified space reference standard, a rate displacement matrix of a fusion time sequence is established by collecting InSAR time sequence data sets with time overlapping, and displacement at each moment is calculated; and S3, dynamically fusing the multisource InSAR data by using a Kalman filtering algorithm to obtain a dynamic evolution time sequence of the observation target.
2. The Kalman filtering-based multi-source InSAR data space-time fusion method according to claim 1, wherein the step S1 comprises the following sub-steps: s101, uniformly projecting LOS deformation of each platform radar sight line direction to a vertical direction: based on the characteristic that the spaceborne InSAR is sensitive to vertical deformation compared with horizontal deformation, setting that the horizontal deformation is ignored; Using satellite incidence angle The LOS deformation of all satellite platforms is converted into the vertical deformation through a projection formula, and the expression of the projection formula is as follows: ; Wherein, the Represents the deformation in the vertical direction, Represents LOS-directed deformation; after projection transformation is completed, the deformation data of all satellite platforms are subjected to time sequence fusion by taking the vertical deformation as a reference and utilizing the vertical deformation; S102, constructing a space grid, and matching homonymous coherent points of different satellite platforms based on the space grid: Constructing a regular space grid as a reference frame based on the data of the first satellite platform; matching deformation monitoring results of other satellite platforms to the space grid through geographic coordinates, and counting pixel points of the other satellite platforms falling in the space grid; Only reserving grid units with effective deformation pixels of all satellite platforms so as to ensure space-time consistency and data integrity; And converting the multi-platform deformation monitoring result into a uniform space grid through grid matching, and providing spatially consistent input data for subsequent time sequence fusion processing.
3. The Kalman filtering-based multi-source InSAR data space-time fusion method according to claim 2, wherein the step S2 comprises the following sub-steps: S201, defining multi-source time sequence data parameters: Two InSAR time series data sets with time overlapping are acquired and respectively recorded as a first data set and a second data set, wherein the first data set is recorded in the first data set The cumulative displacement of individual nodes is noted as The second data set is at the first The cumulative displacement of individual nodes is noted as Post-fusion item And the first The time interval and the deformation rate corresponding to the nodes are respectively as follows And Setting the fusion displacement at the first Cumulative deformation of individual nodes becomes ; Establishing an index mapping relation: Representing the first data set Corresponding fusion displacement of each node Is used as an index into the database, Representing the second data set Corresponding fusion displacement of each node Index of (a); S202, constructing an association equation set of accumulated displacement and velocity based on the multi-source time sequence data parameters: (1) (2) Wherein, the formula (1) is a linear relation between the accumulated displacement of each node and the fusion sequence rate, which is established for the first data set, and the formula (2) is a linear relation between the accumulated displacement of each node and the fusion sequence rate, which is established for the second data set; S203, constructing a rate displacement matrix based on an association equation set of accumulated displacement and rate, and matrixing to obtain displacement at each moment: And (3) combining the cumulative displacement and the association equation set of the velocity to construct a velocity displacement matrix fusing the time sequences: ; where L represents the cumulative displacement vector of the first data set, D represents the cumulative displacement vector of the second data set, A coefficient matrix representing the time interval is formed, Representing a rate vector to be solved; solving generalized inverse matrix by Singular Value Decomposition (SVD) Further calculate the rate vector ; Based on the calculated rate vector With time interval Calculating the accumulated deformation value of each node of the fusion sequence time by time An initial displacement estimate is provided for subsequent kalman filter fusion.
4. The Kalman filtering-based multi-source InSAR data spatiotemporal fusion method of claim 3, wherein said step S3 comprises the sub-steps of: S301, constructing a Kalman filtering state space model: Defining Kalman filtered observations as displacements The state vector is defined as The expression of the Kalman filter state space model is: ; Wherein, the Represent the first Based on individual moments The predicted value at time-1, Represent the first The optimal estimate for the 1 moment, Is a state transition matrix; the prediction covariance matrix is represented by a matrix of prediction coefficients, Representing the estimated covariance matrix, Representing a process noise covariance matrix; The gain matrix is represented by a matrix of gains, Representing the observation matrix of the image of the object, Representing an observed noise covariance matrix; indicating that the corrected optimal estimate is updated, Representing the observed value; Representing the updated and corrected covariance matrix; S302, carrying out dynamic estimation and correction by using Kalman filtering state space model, and outputting Optimal estimated value after time correction and correction As a fusion result; s303, repeating S301-S302 until the next time, and realizing dynamic assimilation and continuous fusion of the multisource InSAR observation data to obtain a dynamic evolution time sequence of the observation target.
5. A multi-source InSAR data space-time fusion system based on kalman filtering, which is realized by the multi-source InSAR data space-time fusion method based on kalman filtering as claimed in any one of claims 1-4, and is characterized by comprising: The space-time reference unifying module is used for unifying space reference references of InSAR deformation data of different satellite platforms, and comprises the steps of projecting LOS deformation to the vertical direction and matching with homonymous coherent points; The velocity displacement matrix construction and displacement calculation module is used for establishing a velocity displacement matrix of a fusion time sequence by collecting an InSAR time sequence data set with time overlapping based on a unified space reference standard, and calculating the displacement at each moment; and the Kalman filtering dynamic fusion module utilizes a Kalman filtering algorithm to dynamically fuse the multi-source InSAR data to obtain a dynamic evolution time sequence of the observation target.
6. A computer readable storage medium storing a computer program, characterized in that the computer program when executed by a processor implements the kalman filter based multisource InSAR data spatiotemporal fusion method according to any of claims 1 to 4.

Description

Kalman filtering-based multi-source InSAR data space-time fusion method and system Technical Field The invention relates to the technical field of earth surface deformation monitoring, in particular to a multi-source InSAR data space-time fusion method and system based on Kalman filtering. Background In order to acquire high-precision earth surface deformation information, the prior art mainly relies on SAR data of a single satellite platform, but the inherent limitations are increasingly prominent, namely, firstly, the satellite revisiting period is long, so that serious defects exist in monitoring data in a time domain, the dynamic evolution process of earth surface deformation is difficult to capture, secondly, the observation precision and reliability of a single data source are continuously reduced under the influence of factors such as space-time incoherence, atmospheric delay disturbance, sensor system deviation and the like, and thirdly, the time resolution is limited by the fixed revisiting period, so that the high-frequency monitoring requirement cannot be met. The defects commonly cause discontinuous deformation time sequence and insufficient precision, and limit the practical level of surface deformation monitoring. To overcome the limitations of a single data source, multi-source InSAR data fusion has become a necessary trend. The Kalman Filtering (KF) is used as a dynamic data processing method, and although the data time sequence relevance can be considered, optimal state estimation is extracted from multi-source uncertain data, and the superiority of the method is verified in the fields of navigation positioning and the like, the method still faces serious challenges when the method is directly applied to multi-source InSAR data, firstly, the flight attitude, system parameters and imaging characteristics of different satellite platforms are obviously different, deformation acquired by each platform is radar line of sight (LOS) and the incident angle is different, deformation components are difficult to unify in the spatial direction, secondly, interference patterns generated by different image sets are not strictly corresponding in pixel-level spatial positions, the number of effective deformation pixels of each platform is inconsistent with spatial distribution, and therefore, space-time reference is difficult to align. The problem of space-time inconsistency among the multi-source data is that the existing method cannot fully utilize multi-platform, multi-track and multi-time-phase InSAR observation information, so that the ground surface deformation monitoring precision is insufficient, the time resolution is limited, the continuity is poor, and a new multi-source InSAR data fusion method capable of systematically solving the problems of space-time reference unification and dynamic fusion is needed to be developed so as to make up for the defect of single data and remarkably improve the monitoring precision and the dynamic property. Disclosure of Invention The embodiment of the invention provides a multi-source InSAR data space-time fusion method and system based on Kalman filtering, which are used for solving the technical problems that how to overcome the defects of accuracy, time resolution and continuity of monitoring ground surface deformation by single satellite platform SAR data, and the space-time reference in the multi-source InSAR data fusion is difficult to unify, and the advantages of the multi-source data cannot be fully exerted so as to improve the monitoring performance. A Kalman filtering-based multi-source InSAR data space-time fusion method comprises the following steps: S1, unifying spatial reference benchmarks of InSAR deformation data of different satellite platforms, wherein the spatial reference benchmarks comprise projecting LOS deformation to a vertical direction and matching same-name coherent points; s2, based on the unified space reference standard, a rate displacement matrix of a fusion time sequence is established by collecting InSAR time sequence data sets with time overlapping, and displacement at each moment is calculated; and S3, dynamically fusing the multisource InSAR data by using a Kalman filtering algorithm to obtain a dynamic evolution time sequence of the observation target. A kalman filter based multisource InSAR data space-time fusion system comprising: The space-time reference unifying module is used for unifying space reference references of InSAR deformation data of different satellite platforms, and comprises the steps of projecting LOS deformation to the vertical direction and matching with homonymous coherent points; The velocity displacement matrix construction and displacement calculation module is used for establishing a velocity displacement matrix of a fusion time sequence by collecting an InSAR time sequence data set with time overlapping based on a unified space reference standard, and calculating the displacement at each moment; and