IES87445Y1 - A detection method for mining subsidence area based on the Mask R-CNN model - Google Patents
A detection method for mining subsidence area based on the Mask R-CNN modelInfo
- Publication number
- IES87445Y1 IES87445Y1 IE20230126U IE20230126U IES87445Y1 IE S87445 Y1 IES87445 Y1 IE S87445Y1 IE 20230126 U IE20230126 U IE 20230126U IE 20230126 U IE20230126 U IE 20230126U IE S87445 Y1 IES87445 Y1 IE S87445Y1
- Authority
- IE
- Ireland
- Prior art keywords
- mining
- dataset
- model
- mining subsidence
- mask
- Prior art date
Links
- 238000005065 mining Methods 0.000 title claims abstract description 87
- 238000001514 detection method Methods 0.000 title claims abstract description 47
- 230000000694 effects Effects 0.000 claims abstract description 19
- 238000012549 training Methods 0.000 claims description 17
- 238000011156 evaluation Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 6
- 238000010801 machine learning Methods 0.000 claims description 6
- 230000002452 interceptive effect Effects 0.000 claims description 3
- 238000009827 uniform distribution Methods 0.000 claims description 3
- 238000010200 validation analysis Methods 0.000 claims description 3
- 238000000034 method Methods 0.000 abstract description 8
- 238000005516 engineering process Methods 0.000 description 5
- 238000012544 monitoring process Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005305 interferometry Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/88—Radar or analogous systems specially adapted for specific applications
- G01S13/89—Radar or analogous systems specially adapted for specific applications for mapping or imaging
- G01S13/90—Radar or analogous systems specially adapted for specific applications for mapping or imaging using synthetic aperture techniques, e.g. synthetic aperture radar [SAR] techniques
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/88—Radar or analogous systems specially adapted for specific applications
- G01S13/885—Radar or analogous systems specially adapted for specific applications for ground probing
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/88—Radar or analogous systems specially adapted for specific applications
- G01S13/89—Radar or analogous systems specially adapted for specific applications for mapping or imaging
- G01S13/90—Radar or analogous systems specially adapted for specific applications for mapping or imaging using synthetic aperture techniques, e.g. synthetic aperture radar [SAR] techniques
- G01S13/9021—SAR image post-processing techniques
- G01S13/9023—SAR image post-processing techniques combined with interferometric techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Remote Sensing (AREA)
- Radar, Positioning & Navigation (AREA)
- Physics & Mathematics (AREA)
- Electromagnetism (AREA)
- General Physics & Mathematics (AREA)
- Computer Networks & Wireless Communication (AREA)
- Theoretical Computer Science (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a detection method for mining subsidence area based on the Mask R-CNN model. First, obtaining original SAR images, which are then co-registered to the common master image, and the InSAR wrapped interferograms are obtained. Based on the wrapped interferograms, which are sliced to generate colorful images. After marking the surface subsidence features of mining area on the sliced colorful interferograms to make a dataset, Gaussian surface function combined with GACOS is used to simulate the wrapped interferograms to expand the dataset. Based on the Mask R-CNN model, the dataset is input to train to obtain a mining subsidence area detection model, which is used to detect the mining subsidence area and obtain their location and boundary. After obtaining the historical and latest mining subsidence areas, determining the state of evolution and evaluating the level of surface activity (the activity duration and frequency of each mining subsidence area). The method of the invention is easy to be operated with better effect and saves time and storage space. It is applicable to a wide range without restriction of platform.
Description
Specification
A detection method for mining subsidence area based on the Mask
R-CNN model
Technical field
The invention relates to the technical field of mine monitoring, in particular to a
detection method for mining subsidence area based on the Mask R-CNN model.
Background technology
Synthetic Aperture Radar Interferometry (InSAR) is an advanced tool for
measuring the ground and enjoys the characteristics of all-day and all-weather
working ability, wide space coverage, high spatial resolution, high measurement
accuracy and no need for ground instruments, therefore;,1t has been a powerful
technology for measuring the surface deformation.
In prior art, there are two mining subsidence.area detection methods: (i) setting
the threshold of the deformation rate, and (ii).using machine learning (ML) algorithm.
The first method is mainly to study the average deformation rate obtained by
InSAR stacking or time series algorithm, although it is easy to be obtained, detection
is not possible for areas with small mining subsidence scales. The second method is
one of more algorithms~based on the machine learning to detect the mining
subsidence area using SAR interferogram, but this method is still in the early stage
with immature technology means.
Summary of the invention
A detection method for mining subsidence area based on the Mask R-CNN
model is needed to be provided in view of the technology problem above.
The invention provides a detection method for mining subsidence area based on
the Mask R-CNN model, comprising:
obtaining original SAR images, and the orbital data and topographic data
corresponding to the original SAR images;
in combination with the orbital data and topographic data, two adjacent SAR
images are used to obtain a InSAR wrapped interferogram:
Specification
a sliced colorful interferogram based on the wrapped interferogram is marked
with the surface subsidence features of mining area to make a dataset that contains an
original image, a masked information and a masked image;
based on the Mask R-CNN model, the optimal mining subsidence area detection
model is obtained through dataset training and taking the machine learning evaluation
index as the model training result evaluation index;
inputting the wrapped interferograms into the mining surface subsidence area
detection model to obtain the location and boundary of the mining subsidence area.
Additionally, the determination of the wrapped interferograms comprises the
following steps:
obtaining an original interferogram through interfering the adjacent SAR images;
removing the plain and terrain phase by DEM;
removing the effect of orbital error by a precise orbital data;
operating the interferograms in multi-view to suppress noises;
generating the interferograms into colorful pictures.
Additionally, the determination of the dataset comprises the following steps:
using Gaussian surface function to model the surface deformation caused by
mining, and obtaining the mining subsidence area with different settlements,
boundary and forms by adjusting the parameters;
simulating the atmosphere phase through differentiating the atmospheric delay
map obtained by GACOS;
simulating the noise phase through continuous uniform distribution function;
obtaining the dataset of simulated interferogram data after summing and
wrapping the surface deformation, atmosphere error and random noise.
Additionally, the determination of the mining surface subsidence area detection
model comprises the following steps:
dividing the real dataset into a training dataset and a validation dataset according
to 4:1 to perform training on the Mask R-CNN basic model;
adding an iteration training model and taking comprehensive indexes of
precision, recall, accuracy and Fl-score as the model accuracy evaluation index;
Specification
When the comprehensive accuracy reaches the highest, the mining subsidence area
detection model using real dataset is obtained;
performing the Mask R-CNN basic model by using the real and simulated mixed
dataset to obtain the mining subsidence detection model using mixed dataset;
comparing the accuracy evaluation indexes of mining subsidence detection
models using real dataset and mixed dataset to determine the final mining subsidence
detection model.
Additionally, the obtained location and boundary of the mining subsidence area,
comprising:
performing the same operation as operating the historical interferograms,
generating the wrapped interferogram together with the latest SAR image in the
historical SAR dataset; Inputting the obtained wrapped interferogram to the mining
surface subsidence area detection model to recognize the latest mining subsidence
area.
Additionally, after obtaining the historical and the latest mining subsidence areas
to determine the evolution state thereof and evaluate the level of surface activity;
wherein, the indexes of the level of surface activity comprises: the subsidence
duration and frequency of each mining subsidence area.
Obtaining the subsidence duration of the mining subsidence area through
multiplying the subsidence frequency by the time interval between SAR image
acquisition.
The invention provides a detection method for mining subsidence area based on
the Mask R-CNN model, the beneficial effects thereof are as follows compared with
the prior art:
based on the Mask R-CNN model, the original wrapped interferogram generated
from two adjacent SAR images is used as the dataset of the detection model, a method
to generate the simulated dataset of the mining area is provided, which is helpful to
improve the precision of the model; based on the Mask R-CNN model, the mining
subsidence detection model is molded, which can detect the mining subsidence area
effectively from the wrapped interferogram, saving the calculation time and reducing
Specification
the additional uncertainty caused by the phase unwrapping and time series inversion.
The detection model is easy to be operated with better effect and is resource efficient.
The detection method is a new method for wide area mining subsidence detection,
which has a wide range of application and does not restrict the use of platform. The
rapid detection and long-term monitoring to the mining subsidence area are important
in monitoring the mining activities, which is crucial to the economic development and
personal and property safety. The method can be used to guide the underground
mining work to boost the monitoring for mining activities and discover the illegal
mining in time.
Description of attached drawings
Figure | is a framework diagram of mining subsidence area detection technology
of a detection method for mining subsidence area based on the Mask R-CNN model
provided in an example;
figure 2 is a flow diagram of generating simulated dataset of a detection method
for mining subsidence area based on the Mask R-CNN model provided in an example;
figure 3 is a diagram of surface activity based on the wide-area mining
subsidence detection results of a. detection method for mining subsidence area based
on the Mask R-CNN model provided in an example.
Specific embodiments
In order to describe the object, technical scheme and merit clearer, the further
detailed description is stated in combination with the attached drawings and examples.
It is understood that the specific examples herein are only used to explain, not to limit
the application.
A detection method for mining subsidence area based on the Mask R-CNN
model is provided in an example, comprising:
As shown in figure 1 to 3, in the invention, through the Mask R-CNN basic
model, the mining subsidence is detected using the two adjacent SAR wrapped
interferograms, which aims to operate easily while save time and storage space and to
obtain a better mining subsidence detection model, based on which to evaluate the
level of the surface activity.
Specification
First, obtaining original SAR images and the orbital data and topographic data
corresponding to the original SAR images; and using two adjacent SAR images to
obtain a InSAR wrapped interferogram:
Step 1: obtaining an original interferogram through interfering the adjacent SAR
images;
Step 2: removing the plain and terrain phase by DEM;
Step 3: removing the effect of orbital error by a precise orbital data;
Step 4: operating the interferograms in multi-view to suppress noises;
Step 5: generating the interferograms into colorful pictures.
As shown in figure |, making a dataset based on the wrapped interferograms,
and a sliced colorful interferogram is marked with the surface subsidence features of
mining area. The generated document containing an.original image, a masked
information and a marked masked image as the standard dataset format of training
network model.
According to the activity duration and frequency of each mining subsidence area
and generalization capability of the model, a method of generating InSAR wrapped
interferogram for simulating mining subsidence is provided, as shown in figure 2,
comprising the following steps:
Step 1: using Gaussian surface function to model the surface deformation caused
by mining, and obtaining the mining subsidence area with different settlements,
boundary and forms by adjusting the parameters;
Step 2: simulating the atmosphere phase through differentiating the atmospheric
delay map obtained by GACOS;
Step 3: simulating the noise phase through continuous uniform distribution
function;
Step 4: obtaining the dataset of simulated interferogram data after summing and
wrapping the surface deformation, atmospheric error and random noise.
Based on the Mask R-CNN model and through the training of the dataset, the
optimal detection model of mining subsidence area is obtained through dataset
training and taking the machine learning evaluation index as the model training result
Specification
evaluation index;
Step 1: dividing the real dataset into a training dataset and a validation dataset
according to 4:1 to perform training on the Mask R-CNN basic model;
Step 2: adding an iteration training model, taking comprehensive indexes of
precision, recall, accuracy and F1-score as the model accuracy evaluation index; when
the comprehensive accuracy reaches the highest, the mining subsidence area detection
model using real dataset is obtained;
Step 3: performing the step | and 2 by using the real and simulated mixed dataset
to obtain the mining subsidence detection model using mixed dataset;
Step 4: comparing the accuracy evaluation indexes of mining subsidence
detection models using real dataset and mixed dataset to determine the final mining
subsidence detection model.
As shown in figure 1, inputting the wrapped interferograms into the mining
surface subsidence area detection model to detect the mining subsidence area and
obtain the location and boundary. After obtaining the historical and the latest mining
subsidence areas to determine the evolution state thereof and evaluate the level of
surface activity (the activity duration and frequency of each mining subsidence area).
In figure 3, the mining surface subsidence detection model can be used to detect
the mining subsidence area in the interferograms, the location and boundary of the
mining subsidence area detected are basically consistent with the detection results by
visual identification.
Figure 3 shows the subsidence duration, due to the mining subsidence area
detection is applied to a separate interferogram, there is no overlapped time and the
time span of each interferogram is known. Therefore, the accumulative duration is
defined by summing the time spans of all interferograms for which a particular pixel
is identified as “activity”.
The above are only several embodiments with specific and detailed description
of the invention, can not be understood as the limitation of the invention. It should be
noted that, deformations and improvements can be made by the ordinary technicians
in the field without departing from the conception of the invention and belong to the
Specification
protection scope of the invention. Therefore, the protection scope of the invention is
defined by the attached claims.
Claims (5)
1. A detection method for mining subsidence area based on the Mask R-CNNmodel, characterized by, comprising: obtaining original SAR images, and the orbital data and topographic datacorresponding to the original SAR images; in combination with the orbital data and topographic data, two adjacent SARimages are used to obtain a InSAR wrapped interferogram; a sliced colorful interferogram based on the wrapped interferogram is markedmanually with the surface subsidence features of mining area to make a dataset thatcontains an original image, a masked information and a masked image; based on the Mask R-CNN model, the optimal mining subsidence area detectionmodel is obtained through dataset training and taking the machine learning evaluationindex as the model training result evaluation index; and inputting the wrapped interferogram into.the mining surface subsidence areadetection model to obtain the location and boundary of the mining subsidence area.
2. A detection method for mining subsidence area based on the Mask R-CNNmodel, characterized by: the determination of the wrapped interferogram, whichcomprises the following steps: obtaining an original interferogram through interfering the adjacent SAR images; removing the plain and terrain phase by DEM; removing the effect of orbital error by a precise orbital data; operating the interferograms in multilook to suppress noises; and generating the interferograms into colorful pictures.
3. A detection method for mining subsidence area based on the Mask R-CNNmodel, characterized by: the determination of the dataset, which comprises thefollowing steps: using Gaussian surface function to model the surface deformation caused bymining, and obtaining the mining subsidence area with different settlements,boundary and features by adjusting the parameters; simulating the atmosphere phase through differentiating the atmospheric delay map obtained by GACOS; simulating the noise phase through continuous uniform distribution function; and obtaining the dataset of simulated interferogram after summing and wrapping thesurface deformation, atmospheric error and random noise.
4. A detection method for mining subsidence area based on the Mask R-CNNmodel, characterized by: the determination of the mining surface subsidence areadetection model, which comprises the following steps: dividing the real dataset into a training dataset and a validation dataset accordingto 4:1 to perform training on the Mask R-CNN basic model; adding an iteration training model and taking comprehensive indexes ofprecision, recall, accuracy and Fl-score as the model accuracy evaluation index;When the comprehensive accuracy reaches the highest, the mining subsidence areadetection model using real dataset is obtained; performing the Mask R-CNN basic model by using the real and simulated mixeddataset to obtain the mining subsidence detection model using mixed dataset; and comparing the accuracy evaluation indexes of mining subsidence detectionmodels using real dataset and mixed dataset to determine the final mining subsidencedetection model.
5. A detection method for mining subsidence area based on the Mask R-CNNmodel, characterized by, the obtained location and boundary of the mining subsidencearea, comprising: performing the same operation as operating the historicalinterferogram, generating the wrapped interferogram together with the latest SARimage in the historical SAR dataset; inputting the obtained wrapped interferogram tothe mining surface subsidence area detection model to recognize the latest miningsubsidence area; comprising: after obtaining the historical and the latest mining subsidence areasto determine the evolution state thereof and evaluate the level of surface activity;wherein, the indexes of the level of surface activity comprises: the subsidenceduration and frequency of each mining subsidence area; and obtaining the subsidence duration of the mining subsidence area through multiplying the subsidence frequency by the time interval between SAR image acquisition
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310148628.5A CN116819521A (en) | 2023-02-22 | 2023-02-22 | Mining subsidence area detection method based on Mask R-CNN model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| IE20230126U1 IE20230126U1 (en) | 2023-10-25 |
| IES87445Y1 true IES87445Y1 (en) | 2023-10-25 |
Family
ID=88111558
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| IE20230126U IES87445Y1 (en) | 2023-02-22 | 2023-04-26 | A detection method for mining subsidence area based on the Mask R-CNN model |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN116819521A (en) |
| IE (1) | IES87445Y1 (en) |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103091676A (en) * | 2013-01-22 | 2013-05-08 | 中国矿业大学 | Mining area surface subsidence synthetic aperture radar interferometry monitoring and calculating method |
| CN107506953A (en) * | 2017-10-12 | 2017-12-22 | 北京蓝尊科技有限公司 | A kind of Collapse by Mining ground environment remediation decision-making assistant information acquisition methods |
| CN113192086B (en) * | 2021-05-11 | 2022-01-28 | 中国自然资源航空物探遥感中心 | Generation method of geological disaster hidden danger deformation intensity distribution diagram and storage medium |
| CN113378945B (en) * | 2021-06-17 | 2021-12-03 | 首都师范大学 | Method for reconstructing high-spatial-temporal-resolution ground settlement information based on machine learning |
| CN114660598A (en) * | 2022-02-07 | 2022-06-24 | 安徽理工大学 | InSAR and CNN-AFSA-SVM fused mining subsidence basin automatic detection method |
| CN115661004B (en) * | 2022-12-26 | 2023-03-21 | 武汉天际航信息科技股份有限公司 | Three-dimensional terrain model and road DEM updating method, device and storage medium |
-
2023
- 2023-02-22 CN CN202310148628.5A patent/CN116819521A/en active Pending
- 2023-04-26 IE IE20230126U patent/IES87445Y1/en unknown
Also Published As
| Publication number | Publication date |
|---|---|
| IE20230126U1 (en) | 2023-10-25 |
| CN116819521A (en) | 2023-09-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Cahill et al. | Modeling sea-level change using errors-in-variables integrated Gaussian processes | |
| CN103114848B (en) | A kind of formation fracture Space Reconstruction method measured based on rock core | |
| Wilson et al. | Surf zone bathymetry and circulation predictions via data assimilation of remote sensing observations | |
| Brodie et al. | Evaluation of video-based linear depth inversion performance and applications using altimeters and hydrographic surveys in a wide range of environmental conditions | |
| Kang et al. | Topographic mapping on large-scale tidal flats with an iterative approach on the waterline method | |
| CN108387899B (en) | Ground control point automatic selection method in synthetic aperture radar interferometry | |
| Fan et al. | Error in target-based georeferencing and registration in terrestrial laser scanning | |
| Jiang et al. | Modeling minimum and maximum detectable deformation gradients of interferometric SAR measurements | |
| CN111323776B (en) | Method for monitoring deformation of mining area | |
| Favalli et al. | LIDAR strip adjustment: Application to volcanic areas | |
| Liu et al. | Bare‐Earth DEM generation in urban areas for flood inundation simulation using global digital elevation models | |
| CN104122553A (en) | Regional ground settlement monitoring method based on multiple track and long strip CTInSAR (coherent target synthetic aperture radar interferometry) | |
| CN110388986B (en) | Land surface temperature inversion method based on TASI data | |
| Hengl et al. | Mapping efficiency and information content | |
| CN109444879A (en) | A kind of nearly tomography coseismic deformation extracting method of DInSAR | |
| Cui et al. | The accurate estimation of GPR migration velocity and comparison of imaging methods | |
| Li et al. | Land subsidence modelling using a long short-term memory algorithm based on time-series datasets | |
| CN116338607B (en) | Time domain and space domain two-step InSAR troposphere delay correction method | |
| CN114812491A (en) | Power transmission line earth surface deformation early warning method and device based on long-time sequence analysis | |
| IES87445Y1 (en) | A detection method for mining subsidence area based on the Mask R-CNN model | |
| Zheng et al. | On the potential of commercial microwave link networks for high spatial resolution rainfall monitoring in urban areas | |
| CN103018728A (en) | Laser radar real-time imaging and building characteristic extracting method | |
| CN116068511A (en) | Deep learning-based InSAR large-scale system error correction method | |
| CN112859073B (en) | Road damage assessment method based on PSInSAR technology | |
| Jahanmard et al. | Retrieval of directional power spectral density and wave parameters from airborne LiDAR point cloud |