CN113610781B - Method and device for detecting change of time sequence SAR (synthetic aperture radar) graph - Google Patents
Method and device for detecting change of time sequence SAR (synthetic aperture radar) graph Download PDFInfo
- Publication number
- CN113610781B CN113610781B CN202110826661.XA CN202110826661A CN113610781B CN 113610781 B CN113610781 B CN 113610781B CN 202110826661 A CN202110826661 A CN 202110826661A CN 113610781 B CN113610781 B CN 113610781B
- Authority
- CN
- China
- Prior art keywords
- sequence
- data set
- change
- sar
- time
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
- G06T2207/10044—Radar image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The application discloses a method and a device for detecting the change of a time sequence SAR graph, wherein the method comprises the following steps: acquiring an SAR data set, and performing modulo and estimation operation on the SAR data set to obtain sequence data; carrying out hypothesis testing on the intensity sequence to obtain a variation sample, and carrying out interference pair screening on the coherence sequence to obtain a stable sample; fitting the change characteristics of the change samples and the stability characteristics of the stability samples respectively to obtain condition distribution parameters; and acquiring a preset filter coefficient diagram, thresholding the preset filter coefficient diagram by using the conditional distribution parameters as likelihood items and using a preset graph cut algorithm to obtain a change detection result. The method can convert the time sequence SAR decision threshold into two classifications and utilize the two classifications to carry out change detection, so that not only is the spatial relevance of the changes considered, but also complex statistical modeling and clustering processes are avoided, the processing time is shortened, weak changes and noise in data are distinguished, and the accuracy of the decision threshold is improved.
Description
Technical Field
The application relates to the technical field of radar image recognition and detection, in particular to a method and a device for detecting the change of a time sequence SAR graph.
Background
SAR (Synthetic Aperture Radar) synthetic aperture radar is an active earth observation system, and can be installed on flight platforms such as airplanes, satellites, spacecraft and the like to observe the earth all the time and all the weather. Along with the sequential emission of the high-return SAR satellite, the SAR satellite can provide massive basic data for the same region, and the change detection research of the time sequence SAR gradually becomes a research hotspot.
One of the conventional time-sequence SAR variation detection methods is to use hypothesis test as a framework to binarize a time sample or a time covariance matrix one by one, so as to determine a varied and stable ground object target. The test statistics used are mainly Wei Xite distance and likelihood ratio, etc., and the decision boundaries are deduced under multidimensional distribution and analytical solutions are given. The other way is to use the traditional change detection thought to conduct two classifications on the collected difference images. Still other kittler minimum error thresholds, ostu methods, expectation maximization, various types of image segmentation, hypothesis testing, and the like are typical supervised or unsupervised methods.
However, the above-mentioned change detection method has the following technical problems: when the parameter hypothesis test is used, the reject domain of the deduced test statistic has no analytical solution, even the test statistic itself has no analytical solution, so that a user is required to judge a threshold boundary or manually test errors or perform Monte Carlo simulation by experience, the detection accuracy is low, and the operation steps are complicated; the decision threshold based on the time sequence SAR change detection technology is mostly carried out pixel by pixel, so that the association degree between the decision threshold and the space is low; when the traditional two-phase change detection method is used for carrying out statistical modeling on two types (change type and stable type) of ground objects, a complex statistical model is needed to fit SAR images, the processing time is long, weak change and noise are difficult to distinguish in the iterative process of the model, the decision threshold is inaccurate, and the false alarm rate are increased.
Disclosure of Invention
The application provides a method and a device for detecting the change of a time sequence SAR graph, which are used for converting a time sequence SAR decision threshold process into a two-class process, automatically training a change class and a stable class sample according to the information of time sequence SAR data to respectively carry out statistical modeling on two classes of ground object targets on the basis of considering the space relevance, and estimating model parameters, so that weak change and noise can be distinguished, the accuracy of the decision threshold can be improved, and the accuracy of change detection is improved.
A first aspect of an embodiment of the present application provides a method for detecting a change in a time-series SAR pattern, including:
acquiring a SAR data set, and performing modulo and estimation operation on the SAR data set to obtain sequence data, wherein the sequence data comprises an intensity sequence and a coherence sequence;
carrying out hypothesis testing on the intensity sequence to obtain a variation sample, and carrying out interference pair screening on the coherence sequence to obtain a stable sample;
fitting the change characteristics of the change samples and the stability characteristics of the stability samples respectively to obtain condition distribution parameters;
and acquiring a preset filter coefficient diagram, thresholding the preset filter coefficient diagram by using the conditional distribution parameters as likelihood items and using a preset graph cut algorithm to obtain a change detection result.
In a possible implementation manner of the first aspect, the performing a hypothesis test on the intensity sequence to obtain a variation sample includes:
carrying out hypothesis testing on the time sequence of each pixel position in the intensity sequence by using a preset sequence statistical testing mode;
when the time sequence of each pixel position in the intensity sequence does not obey preset index distribution, marking as a change sample;
wherein, the variation class sample is shown as follows:
D N (p)＝sup|F N (I(p))-F(I(p))|
in the above, D N (p) preset sequence statistical test method, F N (I (p)) represents the empirical distribution of the time series of each pixel position p, and F (I (p)) represents the corresponding exponential theoretical distribution.
In a possible implementation manner of the first aspect, the filtering the coherence sequence by using the interference pair to obtain a stable class sample includes:
screening a plurality of interference combinations with highest coherence from the coherence sequence by using a preset single-source shortest path algorithm;
calculating an average threshold value of the coherence time of the interference combination with the highest coherence to obtain a plurality of average values;
and screening the average value larger than a preset threshold value from the average values to obtain a stable sample.
In a possible implementation manner of the first aspect, the fitting the change feature of the change class sample and the stability feature of the stability class sample respectively to obtain the conditional distribution parameter includes:
fitting the variation coefficient histogram of the variation sample and the variation coefficient histogram of the stable sample respectively by using a preset Gaussian mixture model to obtain fitting parameters;
estimating model parameters of a preset Gaussian mixture model based on the fitting parameters, and obtaining conditional distribution parameters based on the model parameters;
wherein, the condition distribution parameters are respectively shown in the following formulas:
in the above formula, g (x|mu) i ,Σ i ) Representing the Gaussian component, pi i Representing the weight of the object to be weighed,representing multidimensional variables and each representing a feature, mu i Sum sigma i Mean and variance are represented, respectively, and M represents the gaussian component.
In a possible implementation manner of the first aspect, the obtaining a preset filter coefficient map includes:
registering the SAR data set to obtain a registration data set;
calculating the variation coefficient of the registration data set to obtain a variation coefficient graph;
and filtering the variation coefficient map to obtain a filtering coefficient map.
In a possible implementation manner of the first aspect, the calculating the coefficient of variation of the registration data set to obtain a coefficient of variation map includes:
calculating the variation coefficient of the registration data set pixel by adopting the following formula to obtain a variation coefficient map;
in the above-mentioned method, the step of,
wherein cv (p) represents the coefficient of variation of position p,and->Respectively representing the variance and the mean of the time samples, I t (p) represents an intensity sample of position p at time t, N represents the number of images of the SAR data set.
In a possible implementation manner of the first aspect, the performing a modulo and estimating operation on the SAR data set to obtain sequence data includes:
registering the SAR data set to obtain a registration data set;
performing a modulus operation on the registration data set to obtain an intensity sequence;
estimating the intensity sequence to obtain a coherence sequence.
A second aspect of an embodiment of the present application provides a device for detecting a change in a time-series SAR pattern, including:
the acquisition module is used for acquiring an SAR data set, and performing modulo and estimation operation on the SAR data set to obtain sequence data, wherein the sequence data comprises an intensity sequence and a coherence sequence;
the sample module is used for carrying out hypothesis test on the intensity sequence to obtain a variation sample, and carrying out interference pair screening on the coherence sequence to obtain a stable sample;
the fitting module is used for respectively fitting the change characteristics of the change samples and the stability characteristics of the stability samples to obtain condition distribution parameters;
the detection module is used for acquiring a preset filter coefficient diagram, taking the conditional distribution parameters as likelihood items, and thresholding the preset filter coefficient diagram by using a preset graph cut algorithm to obtain a change detection result.
Compared with the prior art, the method and the device for detecting the change of the time sequence SAR graph have the beneficial effects that: the method can convert the time sequence SAR decision threshold into two types of variation and stability, consider the space relevance of variation under the condition of no manual intervention, and the samples of the variation and stability consider time, space, interference, backward scattering and context background information, so that the time sequence SAR information quantity can be maximized, and finally, the condition distribution parameters of the two types of variation and stability are utilized for variation detection, thereby avoiding complex statistical modeling and clustering processes, reducing the overlapping area of the two types of distribution, shortening the processing time, accurately determining weak variation and noise in data, improving the accuracy of the decision threshold, and reducing the false alarm rate and the false alarm rate.
Drawings
FIG. 1 is a flow chart of a method for detecting a change of a time-series SAR image according to an embodiment of the present application;
FIG. 2 is a flowchart illustrating an operation of a method for detecting a change in a time-series SAR image according to an embodiment of the present application;
fig. 3 is a schematic structural diagram of a device for detecting a change of a time-series SAR pattern according to an embodiment of the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
The following technical problems exist in several currently used change detection modes: when the parameter hypothesis test is used, the reject domain of the deduced test statistic has no analytical solution, even the test statistic itself has no analytical solution, so that a user is required to judge a threshold boundary or manually test errors or perform Monte Carlo simulation by experience, the detection accuracy is low, and the operation steps are complicated; the decision threshold based on the time sequence SAR change detection technology is mostly carried out pixel by pixel, so that the association degree between the decision threshold and the space is low; when the traditional two-phase change detection method is used for carrying out statistical modeling on two types (change type and stable type) of ground objects, a complex statistical model is needed to fit SAR images, the processing time is long, weak change and noise are difficult to distinguish in the iterative process of the model, the decision threshold is inaccurate, and the false alarm rate are increased.
In order to solve the above-mentioned problems, a method for detecting a change in a time-series SAR pattern according to an embodiment of the present application will be described and illustrated in detail in the following specific embodiments.
Referring to fig. 1, a flow chart of a method for detecting a change of a time-series SAR pattern according to an embodiment of the present application is shown.
As an example, the method for detecting the change of the time sequence SAR pattern may include:
s11, acquiring a SAR data set, and performing modulo and estimation operation on the SAR data set to obtain sequence data, wherein the sequence data comprises an intensity sequence and a coherence sequence.
In this embodiment, the SAR data set may be a massive basic data set acquired by SAR satellites in the same region.
The modulo and estimation operation may be a feature extraction of the data.
In order to divide the acquired SAR data set into two categories, in an alternative embodiment, step S11 may comprise the sub-steps of:
and a substep S111, registering the SAR data set to obtain a registration data set.
The quasi-registration may be a match of geographic coordinates of different image patterns obtained with different imaging means within the same region.
And a substep S112, performing modulo operation on the registration data set to obtain an intensity sequence.
And a substep S113, estimating the intensity sequence to obtain a coherence sequence.
S12, carrying out hypothesis test on the intensity sequence to obtain a variation sample, and carrying out interference pair screening on the coherence sequence to obtain a stable sample.
In the embodiment, the time sequence intensity hypothesis test and the coherence sequence label the change class and the stability class samples, so that the modeling process complexity and the condition distribution estimation inaccuracy caused by the traditional clustering in the original Markov process can be replaced, the modeling efficiency is improved, and the detection accuracy is increased.
In order to obtain the change class samples quickly and accurately, in an alternative embodiment, step S12 may include the sub-steps of:
and S121, performing hypothesis testing on the time sequence of each pixel position in the intensity sequence by using a preset sequence statistical testing mode.
Alternatively, the preset sequence statistical test mode may be a single sample Kolmogorov-Smirnov test mode, or may be other non-parametric statistical test modes.
Sub-step S122, marking as a variation sample when the time sequence of each pixel position in the intensity sequence does not follow a preset exponential distribution;
wherein, the variation class sample is shown as follows:
D N (p)＝sup|F N (I(p))-F(I(p))|
in the above, D N (p) preset sequence statistical test method, F N (I (p)) represents the empirical distribution of the time series of each pixel position p, and F (I (p)) represents the corresponding exponential theoretical distribution.
In order to accurately screen the most relevant stable class samples, in an alternative embodiment, step S12 may comprise the sub-steps of:
and step 123, screening a plurality of interference combinations with highest coherence from the coherence sequence by utilizing a preset single-source shortest path algorithm.
In an alternative embodiment, other methods of interference pair screening may be employed.
And step S124, carrying out average threshold value calculation on the coherence time of the interference combination with the highest coherence to obtain a plurality of average values.
And step S125, screening the average value larger than a preset threshold value from the average values to obtain a stable class sample.
In an embodiment, the preset threshold may be adjusted according to actual needs.
And S13, respectively fitting the change characteristics of the change samples and the stability characteristics of the stability samples to obtain the condition distribution parameters.
In an embodiment, the change feature may be feature data in a change class sample, and the stability feature may be feature data of a stability class sample. The variation characteristic and the stabilization characteristic can be a variation coefficient diagram, a ratio difference diagram of the first scene and the last scene images through time series, a difference diagram or other characteristics, and the like. Alternatively, the dimension of the fit may also be increased by adding feature data or categories.
For the purpose of, in one of the embodiments, step S13 may comprise the sub-steps of:
and S131, respectively fitting the variation coefficient histogram of the variation sample and the variation coefficient histogram of the stable sample by using a preset Gaussian mixture model to obtain fitting parameters.
And a substep S132, estimating model parameters of a preset Gaussian mixture model based on the fitting parameters, and obtaining conditional distribution parameters based on the model parameters.
Wherein, the condition distribution parameters are respectively shown in the following formulas:
in the above formula, g (x|mu) i ,Σ i ) Representing GaussianComponent, pi i Representing the weight of the object to be weighed,representing multidimensional variables and each representing a feature, mu i Sum sigma i Mean and variance are represented, respectively, and M represents the gaussian component.
S14, acquiring a preset filter coefficient diagram, thresholding the preset filter coefficient diagram by using the conditional distribution parameters as likelihood items and using a preset graph cut algorithm to obtain a change detection result.
In an embodiment, the condition distribution parameter may be used as a likelihood term, and a Graph-cut algorithm (Graph cut) may be used to perform thresholding on a preset filter coefficient Graph, so as to obtain a change detection result.
Alternatively, other graph-cut algorithms can be used for processing, and the graph-cut algorithm can be specifically adjusted according to actual needs.
In order to avoid complex statistical modeling and clustering processes and reduce the overlapping area of the two types of distributions, in this embodiment, a preset filter coefficient map is used.
In one of these embodiments, step S14 may include the sub-steps of:
and step S141, registering the SAR data set to obtain a registration data set.
Optionally, the registration is the same as the registration of substep S111.
And a substep S142, calculating the variation coefficient of the registration data set to obtain a variation coefficient graph.
In actual operation, sub-step S142 actually operates as follows:
calculating the variation coefficient of the registration data set pixel by adopting the following formula to obtain a variation coefficient map;
in the above-mentioned method, the step of,
wherein cv (p) represents the coefficient of variation of position p,and->Respectively representing the variance and the mean of the time samples, I t (p) represents an intensity sample of position p at time t, N represents the number of images of the SAR data set.
And S143, filtering the variation coefficient map to obtain a filtering coefficient map.
Referring to fig. 2, an operation flowchart of a method for detecting a change of a time-series SAR pattern according to an embodiment of the present application is shown.
In actual operation, after the SAR data set is acquired, the SAR data set can be registered, after the registration is completed, the variation coefficient of the registered SAR data set can be calculated pixel by pixel to obtain a variation coefficient map, then the registered SAR data is subjected to modulo operation to obtain an intensity sequence, and the coherence sequence is estimated; then, for the intensity sequence, carrying out hypothesis test on the time sequence of the position p one by using a single sample Kolmogorov-Smirnov test, and marking the time sequence as a change sample if the time sequence is not subjected to exponential distribution; for a coherence sequence, after the interference combination with highest coherence is automatically selected by maximizing the coherence by using a single-source shortest path algorithm, the coherence time average is subjected to a threshold value, and when the average value is larger than a preset value, a mark is a stable sample; filtering the variation coefficient diagram, respectively fitting variation coefficient histograms of variation samples and stable samples by using a Gaussian mixture model, and estimating model parameters to obtain conditional distribution parameters; and finally, thresholding the filtered time variation coefficient Graph by using the obtained conditional distribution parameters as likelihood items and using a Graph-cut algorithm (Graph cut) to obtain a variation detection result.
In this embodiment, the embodiment of the present application provides a method for detecting a change of a time sequence SAR pattern, which has the following beneficial effects: the method can convert the time sequence SAR decision threshold into two types of variation and stability, consider the space relevance of variation under the condition of no manual intervention, and the samples of the variation and stability consider time, space, interference, backward scattering and context background information, so that the time sequence SAR information quantity can be maximized, and finally, the condition distribution parameters of the two types of variation and stability are utilized for variation detection, thereby avoiding complex statistical modeling and clustering processes, reducing the overlapping area of the two types of distribution, shortening the processing time, accurately determining weak variation and noise in data, improving the accuracy of the decision threshold, and reducing the false alarm rate and the false alarm rate.
The embodiment of the application also provides a device for detecting the change of the time sequence SAR pattern, and referring to fig. 3, a schematic structural diagram of the device for detecting the change of the time sequence SAR pattern is shown.
As an example, the change detection device of the time sequence SAR pattern may include:
an acquiring module 301, configured to acquire a SAR data set, and perform modulo and estimation operations on the SAR data set to obtain sequence data, where the sequence data includes an intensity sequence and a coherence sequence;
the sample module 302 is configured to perform hypothesis testing on the intensity sequence to obtain a variation sample, and perform interference pair screening on the coherence sequence to obtain a stable sample;
the fitting module 303 is configured to fit the change feature of the change class sample and the stability feature of the stability class sample respectively, so as to obtain a conditional distribution parameter;
the detection module 304 is configured to obtain a preset filter coefficient map, and thresholde the preset filter coefficient map by using the conditional distribution parameter as a likelihood term and using a preset graph-cut algorithm to obtain a change detection result.
Optionally, the sample module is further configured to:
carrying out hypothesis testing on the time sequence of each pixel position in the intensity sequence by using a preset sequence statistical testing mode;
when the time sequence of each pixel position in the intensity sequence does not obey preset index distribution, marking as a change sample;
wherein, the variation class sample is shown as follows:
D N (p)＝sup|F N (I(p))-F(I(p))|
in the above, D N (p) preset sequence statistical test method, F N (I (p)) represents the empirical distribution of the time series of each pixel position p, and F (I (p)) represents the corresponding exponential theoretical distribution.
Optionally, the sample module is further configured to:
screening a plurality of interference combinations with highest coherence from the coherence sequence by using a preset single-source shortest path algorithm;
calculating an average threshold value of the coherence time of the interference combination with the highest coherence to obtain a plurality of average values;
and screening the average value larger than a preset threshold value from the average values to obtain a stable sample.
Optionally, the fitting module is further configured to:
fitting the variation coefficient histogram of the variation sample and the variation coefficient histogram of the stable sample respectively by using a preset Gaussian mixture model to obtain fitting parameters;
estimating model parameters of a preset Gaussian mixture model based on the fitting parameters, and obtaining conditional distribution parameters based on the model parameters;
wherein, the condition distribution parameters are respectively shown in the following formulas:
in the above formula, g (x|mu) i ,Σ i ) Representing the Gaussian component, pi i Representing the weight of the object to be weighed,representing multidimensional variables and each representing a feature, mu i Sum sigma i Mean and variance are represented, respectively, and M represents the gaussian component.
Optionally, the detection module is further configured to:
registering the SAR data set to obtain a registration data set;
calculating the variation coefficient of the registration data set to obtain a variation coefficient graph;
and filtering the variation coefficient map to obtain a filtering coefficient map.
Optionally, the detection module is further configured to:
calculating the variation coefficient of the registration data set pixel by adopting the following formula to obtain a variation coefficient map;
in the above-mentioned method, the step of,
wherein cv (p) represents the coefficient of variation of position p,and->Respectively representing the variance and the mean of the time samples, I t (p) represents an intensity sample of position p at time t, N represents the number of images of the SAR data set.
Optionally, the acquiring module is further configured to:
registering the SAR data set to obtain a registration data set;
performing a modulus operation on the registration data set to obtain an intensity sequence;
estimating the intensity sequence to obtain a coherence sequence.
Further, an embodiment of the present application further provides an electronic device, including: the method for detecting the change of the time sequence SAR graph comprises a memory, a processor and a computer program which is stored in the memory and can run on the processor, wherein the processor realizes the method for detecting the change of the time sequence SAR graph according to the embodiment when executing the program.
Further, an embodiment of the present application also provides a computer-readable storage medium storing computer-executable instructions for causing a computer to perform the method for detecting a change in a time-series SAR pattern according to the above embodiment.
While the foregoing is directed to the preferred embodiments of the present application, it will be appreciated by those skilled in the art that changes and modifications may be made without departing from the principles of the application, such changes and modifications are also intended to be within the scope of the application.
Claims (8)
1. A method for detecting a change in a time-series SAR pattern, comprising:
acquiring a SAR data set, and performing modulo and estimation operation on the SAR data set to obtain sequence data, wherein the sequence data comprises an intensity sequence and a coherence sequence;
carrying out hypothesis testing on the intensity sequence to obtain a variation sample, and carrying out interference pair screening on the coherence sequence to obtain a stable sample;
fitting the change characteristics of the change samples and the stability characteristics of the stability samples respectively to obtain condition distribution parameters;
acquiring a preset filter coefficient diagram, thresholding the preset filter coefficient diagram by using the conditional distribution parameters as likelihood items and a preset graph cut algorithm to obtain a change detection result;
the obtaining a preset filtering coefficient diagram includes:
registering the SAR data set to obtain a registration data set;
calculating the variation coefficient of the registration data set to obtain a variation coefficient graph;
filtering the variation coefficient map to obtain a filtering coefficient map;
the calculating the variation coefficient of the registration data set to obtain a variation coefficient graph comprises the following steps:
calculating the variation coefficient of the registration data set pixel by adopting the following formula to obtain a variation coefficient map;
in the above-mentioned method, the step of,
wherein cv (p) represents the coefficient of variation of position p,and->Respectively representing the variance and the mean of the time samples, I t (p) represents an intensity sample of position p at time t, N represents the number of images of the SAR data set.
2. The method for detecting a change in a time-series SAR pattern according to claim 1, wherein said performing a hypothesis test on the intensity sequence to obtain a change class sample comprises:
carrying out hypothesis testing on the time sequence of each pixel position in the intensity sequence by using a preset sequence statistical testing mode;
when the time sequence of each pixel position in the intensity sequence does not obey preset index distribution, marking as a change sample;
wherein, the variation class sample is shown as follows:
D N (p)＝supF N (I(p))-F(I(p))
in the above, D N (p) preset sequence statistical test method, F N (I (p)) represents the empirical distribution of the time series of each pixel position p, and F (I (p)) represents the corresponding exponential theoretical distribution.
3. The method for detecting a change in a time-series SAR pattern according to claim 1, wherein said filtering the pair of coherence sequences to obtain a stable class sample comprises:
screening a plurality of interference combinations with highest coherence from the coherence sequence by using a preset single-source shortest path algorithm;
calculating an average threshold value of the coherence time of the interference combination with the highest coherence to obtain a plurality of average values;
and screening the average value larger than a preset threshold value from the average values to obtain a stable sample.
4. A method for detecting a change in a time-series SAR pattern according to any one of claims 2 or 3, wherein said fitting the change characteristic of the change class sample and the stability characteristic of the stability class sample, respectively, to obtain the condition distribution parameter comprises:
fitting the variation coefficient histogram of the variation sample and the variation coefficient histogram of the stable sample respectively by using a preset Gaussian mixture model to obtain fitting parameters;
estimating model parameters of a preset Gaussian mixture model based on the fitting parameters, and obtaining conditional distribution parameters based on the model parameters;
wherein, the condition distribution parameters are respectively shown in the following formulas:
in the above formula, g (x|mu) i ,Σ i ) Representing the Gaussian component, pi i Representing the weight of the object to be weighed,representing multidimensional variables and each representing a feature, mu i Sum sigma i Mean and variance are represented, respectively, and M represents the gaussian component.
5. A method for detecting a change in a time-series SAR pattern according to any one of claims 1, 2, and 3, wherein said performing a modulo and estimation operation on the SAR data set to obtain sequence data comprises:
registering the SAR data set to obtain a registration data set;
performing a modulus operation on the registration data set to obtain an intensity sequence;
estimating the intensity sequence to obtain a coherence sequence.
6. A time-series SAR pattern change detection apparatus, comprising:
the acquisition module is used for acquiring an SAR data set, and performing modulo and estimation operation on the SAR data set to obtain sequence data, wherein the sequence data comprises an intensity sequence and a coherence sequence;
the sample module is used for carrying out hypothesis test on the intensity sequence to obtain a variation sample, and carrying out interference pair screening on the coherence sequence to obtain a stable sample;
the fitting module is used for respectively fitting the change characteristics of the change samples and the stability characteristics of the stability samples to obtain condition distribution parameters;
the detection module is used for acquiring a preset filter coefficient diagram, taking the conditional distribution parameters as likelihood items, and thresholding the preset filter coefficient diagram by using a preset graph cut algorithm to obtain a change detection result;
the detection module is further used for:
registering the SAR data set to obtain a registration data set;
calculating the variation coefficient of the registration data set to obtain a variation coefficient graph;
filtering the variation coefficient map to obtain a filtering coefficient map;
the calculating the variation coefficient of the registration data set to obtain a variation coefficient graph comprises the following steps:
calculating the variation coefficient of the registration data set pixel by adopting the following formula to obtain a variation coefficient map;
in the above-mentioned method, the step of,
wherein cv (p) represents the coefficient of variation of position p,and->Respectively representing the variance and the mean of the time samples, I t (p) represents an intensity sample of position p at time t, N represents the number of images of the SAR data set.
7. An electronic device, comprising: memory, a processor and a computer program stored on the memory and executable on the processor, which processor implements the method for detecting a change in a time-series SAR pattern according to claims 1-5 when executing the program.
8. A computer-readable storage medium storing computer-executable instructions for causing a computer to perform the method of time-series SAR pattern change detection method of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110826661.XA CN113610781B (en) | 2021-07-21 | 2021-07-21 | Method and device for detecting change of time sequence SAR (synthetic aperture radar) graph |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110826661.XA CN113610781B (en) | 2021-07-21 | 2021-07-21 | Method and device for detecting change of time sequence SAR (synthetic aperture radar) graph |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113610781A CN113610781A (en) | 2021-11-05 |
| CN113610781B true CN113610781B (en) | 2023-09-15 |
Family
ID=78305078
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110826661.XA Active CN113610781B (en) | 2021-07-21 | 2021-07-21 | Method and device for detecting change of time sequence SAR (synthetic aperture radar) graph |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113610781B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118486468A (en) * | 2024-07-10 | 2024-08-13 | 吉林大学 | Patient care intelligent early warning system and method based on 5G Internet of things technology |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9389311B1 (en) * | 2015-02-19 | 2016-07-12 | Sandia Corporation | Superpixel edges for boundary detection |
| CN111429496A (en) * | 2020-03-05 | 2020-07-17 | 武汉大学 | Statistical characteristic-considered time sequence PolSAR image unsupervised change detection method |
| CN112906514A (en) * | 2021-02-03 | 2021-06-04 | 北京观微科技有限公司 | Time sequence SAR image ground object type change detection method considering different polarizations |
-
2021
- 2021-07-21 CN CN202110826661.XA patent/CN113610781B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9389311B1 (en) * | 2015-02-19 | 2016-07-12 | Sandia Corporation | Superpixel edges for boundary detection |
| CN111429496A (en) * | 2020-03-05 | 2020-07-17 | 武汉大学 | Statistical characteristic-considered time sequence PolSAR image unsupervised change detection method |
| CN112906514A (en) * | 2021-02-03 | 2021-06-04 | 北京观微科技有限公司 | Time sequence SAR image ground object type change detection method considering different polarizations |
Non-Patent Citations (3)
| Title |
|---|
| Sentnel-1 SAR在洪水范围提取与极化分析中的应用研究;陈赛楠 等;《地球信息科学》;第23卷(第6期);1603-1612页 * |
| 时序InSAR同质样本选取算法研究;蒋弥;《地球物理学报》;第61卷(第12期);第4767-4776页 * |
| 时序Sentinel-1 TOPS模式SAR数据精配准;马张烽;《测绘学报》;第50卷(第5期);第634-649页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113610781A (en) | 2021-11-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108229488B (en) | Method and device for detecting key points of object and electronic equipment | |
| Prendes et al. | A new multivariate statistical model for change detection in images acquired by homogeneous and heterogeneous sensors | |
| CN112132042A (en) | SAR image target detection method based on anti-domain adaptation | |
| CN108447057B (en) | SAR image change detection method based on significance and depth convolution network | |
| CN110889843B (en) | SAR image ship target detection method based on maximum stable extremal region | |
| Deng et al. | Cloud detection in satellite images based on natural scene statistics and gabor features | |
| CN105389799B (en) | SAR image object detection method based on sketch map and low-rank decomposition | |
| CN116012364B (en) | SAR image change detection method and device | |
| CN109389062A (en) | Utilize the method for High Resolution Spaceborne SAR image zooming-out lake land and water cut-off rule | |
| CN108171119B (en) | SAR image change detection method based on residual error network | |
| Prendes et al. | Change detection for optical and radar images using a Bayesian nonparametric model coupled with a Markov random field | |
| CN104517124B (en) | SAR image change detection based on SIFT feature | |
| CN113610781B (en) | Method and device for detecting change of time sequence SAR (synthetic aperture radar) graph | |
| Tu et al. | Airport detection in SAR images via salient line segment detector and edge-oriented region growing | |
| CN116543171A (en) | Target detection method and device and electronic equipment | |
| CN113822361B (en) | SAR image similarity measurement method and system based on Hamming distance | |
| CN115082781A (en) | Ship image detection method and device and storage medium | |
| CN111553184A (en) | Small target detection method and device based on electronic purse net and electronic equipment | |
| CN104537384A (en) | SAR (synthetic aperture radar) target identification method combined with likelihood ratio decision | |
| CN104680549A (en) | SAR (synthetic aperture radar) image change detection method based on high-order neighborhood TMF (triplet Markov random field) model | |
| CN104239895B (en) | SAR target identification method based on feature dimension reduction | |
| CN116310913B (en) | Natural resource investigation monitoring method and device based on unmanned aerial vehicle measurement technology | |
| CN107729903A (en) | SAR image object detection method based on area probability statistics and significance analysis | |
| CN112348750A (en) | SAR image change detection method based on threshold fusion and neighborhood voting | |
| CN116719241A (en) | Automatic control method for informationized intelligent gate based on 3D visualization technology |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |