CN115908954A - Geological disaster hidden danger identification system and method based on artificial intelligence and electronic equipment - Google Patents
Geological disaster hidden danger identification system and method based on artificial intelligence and electronic equipment Download PDFInfo
- Publication number
- CN115908954A CN115908954A CN202310182148.0A CN202310182148A CN115908954A CN 115908954 A CN115908954 A CN 115908954A CN 202310182148 A CN202310182148 A CN 202310182148A CN 115908954 A CN115908954 A CN 115908954A
- Authority
- CN
- China
- Prior art keywords
- image
- elevation
- layer
- geological disaster
- sub
- 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.)
- Granted
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
- Y02A10/40—Controlling or monitoring, e.g. of flood or hurricane; Forecasting, e.g. risk assessment or mapping
Abstract
The application provides a geological disaster hidden danger identification system and method based on artificial intelligence and electronic equipment, wherein an elevation image set of a monitoring area is obtained through an elevation image set obtaining unit, a first elevation difference image and a second elevation difference image are further determined, the first elevation difference image and the second elevation difference image are input into a geological disaster hidden danger identification model, intelligent identification is carried out, and a geological disaster hidden danger identification result is obtained. In such a way, the intelligent identification of the geological disaster can be realized by using the neural network model according to the first elevation difference image (reflecting the short-term deformation condition of the monitored area) and the second elevation difference image (reflecting the medium-term deformation condition of the monitored area) of the height Cheng Tuxiang of the monitored area, so that the dependence on expert experience can be reduced, the identification of the hidden danger area is made before the geological disaster occurs, corresponding precautionary measures can be made in advance, for example, operation is carried out far away from the hidden danger area, human activities are avoided being carried out in the hidden danger area, and the like.
Description
Technical Field
The application relates to the technical field of geological disaster monitoring, in particular to a geological disaster hidden danger identification system and method based on artificial intelligence and electronic equipment.
Background
Geological disasters are major disaster events occurring in natural processes, such as collapse, landslide, debris flow, ground collapse, ground subsidence, and the like.
The existing landslide and debris flow monitoring technology is developed, casualties can be reduced as far as possible, but for long-term identification of hidden dangers of geological disasters, the existing landslide and debris flow monitoring technology mainly depends on expert experience, is difficult to realize reliable intelligent identification, and is not beneficial to making identification and precautionary measures of the hidden dangers of the geological disasters in advance.
Disclosure of Invention
An object of the embodiment of the application is to provide a geological disaster hidden danger identification system and method based on artificial intelligence and an electronic device, so that intelligent identification of a geological disaster hidden danger area is realized, and prevention is made in advance.
In order to achieve the above object, embodiments of the present application are implemented as follows:
in a first aspect, an embodiment of the present application provides a geological disaster hidden danger identification system based on artificial intelligence, including: an elevation image set obtaining unit, configured to obtain an elevation image set of a monitored area, where the elevation image set includes a current elevation image, a first reference image, and a second reference image of the monitored area, and the first reference image and the current elevation image are separated by an interval
Month, the second reference image being spaced ∑ from the current elevation image>
Moon,. Sup.,>
，
(ii) a An elevation difference image processing unit, configured to determine a first elevation difference image based on the current elevation image and the first reference image, and determine a second elevation difference image based on the current elevation image and the second reference image; and the geological disaster hidden danger identification unit is preset with a geological disaster hidden danger identification model and used for inputting the first elevation difference image and the second elevation difference image into the geological disaster hidden danger identification model to obtain a geological disaster hidden danger identification result output by the geological disaster hidden danger identification model.
In the embodiment of the present application, an elevation image set (a current elevation image, a first reference image, and a second reference image, where the first reference image is spaced from the current elevation image) of a monitored area is acquired by an elevation image set acquisition unit
The second reference image is used as a basis for observing the deformation of the monitoring area in the short and medium periods, and the second reference image is separated from the current elevation image>
And the image is used as a basis for observing long-term deformation in the monitored area), further determining a first elevation difference image and a second elevation difference image, inputting the first elevation difference image and the second elevation difference image into a geological disaster hidden danger identification model, and carrying out intelligent identification to obtain a geological disaster hidden danger identification result. In such a way, the intelligent identification of the geological disaster can be realized by the neural network model according to the first elevation difference image (reflecting the short-term deformation condition of the monitored area) and the second elevation difference image (reflecting the medium-term deformation condition of the monitored area) of the height Cheng Tuxiang of the monitored area, so that the experience of experts can be reducedDepending on the situation, the hidden danger areas are identified before geological disasters occur, and corresponding precautionary measures can be taken in advance, such as operation far away from the hidden danger areas, human activities in the hidden danger areas are avoided, and the like.
With reference to the first aspect, in a first possible implementation manner of the first aspect, the geological disaster potential risk identification model includes a preprocessing layer, a first trunk network, a second trunk network, a full connection layer, and an output layer, where the preprocessing layer is configured to generate a first input image based on an elevation difference corresponding to each pixel in the first elevation difference image, and is configured to generate a second input image based on an elevation difference corresponding to each pixel in the second elevation difference image; the first trunk network is connected with the preprocessing layer and used for carrying out feature extraction on the input first input image to obtain a first feature extraction result; the second trunk network is connected with the preprocessing layer and is used for performing feature extraction on the input second input image to obtain a second feature extraction result; the full connection layer is respectively connected with the first trunk network and the second trunk network and is used for integrating the first feature extraction result and the second feature extraction result; the output layer is connected with the full-connection layer through a classifier, the classifier classifies the integrated feature result to obtain a geological disaster hidden danger identification result, and the geological disaster hidden danger identification result is output through the output layer.
In this implementation, the geological disaster hidden danger identification model includes a preprocessing layer, a first trunk network, a second trunk network, a full connection layer and an output layer, the first trunk network, the second trunk network is used for respectively carrying out feature extraction on the input first input image and the input second input image, and corresponding first feature extraction result and second feature extraction result are obtained, so that identification of the hidden danger region is carried out on the basis of the features of the monitoring region in short-medium deformation and the features of the monitoring region in medium-long deformation, the features of the geological disaster hidden danger region in short-medium and medium-long periods can be considered, and the accuracy of identification of the geological disaster hidden danger region is favorably improved.
With reference to the first possible implementation manner of the first aspect, in a second possible implementation manner of the first aspect, the preprocessing layer is specifically configured to: calculating a first gray value corresponding to each pixel based on the height difference corresponding to each pixel in the first height difference image, and constructing a first input image corresponding to the first height difference image based on the first gray value corresponding to each pixel; and calculating a second gray value corresponding to each pixel based on the height difference corresponding to each pixel in the second height difference image, and constructing a second input image corresponding to the second height difference image based on the second gray value corresponding to each pixel.
With reference to the second possible implementation manner of the first aspect, in a third possible implementation manner of the first aspect, the preprocessing layer is specifically configured to: calculating a first gray value corresponding to each pixel in the first height difference image by adopting the following formula:
wherein the content of the first and second substances,
is the ^ th or greater in the first difference of elevation image>
A first gray value corresponding to each pixel->
Is the ^ th or greater in the first difference of elevation image>
The elevation difference corresponding to each pixel is greater or less>
For a maximum value of the elevation difference in the first elevation difference image->
For a minimum value of the elevation difference in the first elevation difference image->
；
Calculating a second gray value corresponding to each pixel in the second elevation difference image by adopting the following formula:
wherein the content of the first and second substances,
is a second elevation difference image>
A second gray value corresponding to an individual pixel->
Is the ^ th or greater in the second elevation difference image>
The elevation difference corresponding to each pixel is greater or less>
For a maximum value of the elevation difference in the second elevation difference image->
For a minimum value of elevation difference in the second elevation difference image, based on the elevation difference value>
。
In the implementation mode, by using the conversion mode, the elevation difference of each pixel in the first elevation difference image (or the second elevation difference image) can be converted into a gray value, so that a corresponding first input image (or the second input image) is obtained, the elevation difference is reflected in the gray value, the neural network model can conveniently perform processing such as feature extraction and feature mapping, and finally, identification of the hidden danger of the geological disaster is realized.
With reference to the first possible implementation manner of the first aspect, in a fourth possible implementation manner of the first aspect, the first backbone network includes convolutional layers
The pooling layer is used for collecting and storing the blood>
And a convolution layer>
And pooling layer>
Said convolutional layer->
For performing convolution processing on the first input image to obtain a characteristic sub-picture->
(ii) a The pooling layer->
For sub-picture->
Performing maximum pooling treatment to obtain characteristic sub-map->
(ii) a The convolutional layer->
For sub-picture->
Convolution processing is carried out to obtain a characteristic sub-picture->
(ii) a The pooling layer->
For sub-picture->
Where the maximum pooling is performedProcessing to obtain a characteristic sub-picture>
。
With reference to the first possible implementation manner of the first aspect, in a fifth possible implementation manner of the first aspect, the second backbone network includes a convolutional layer
The pooling layer is used for collecting and storing the blood>
And the convolutional layer->
And pooling layer>
Said convolutional layer->
Is used for carrying out convolution processing on the second input image to obtain a characteristic sub image->
(ii) a The pooling layer->
For sub-picture->
Performing maximal pooling to obtain characteristic sub-map->
(ii) a The convolutional layer->
For sub-picture->
Convolution processing is carried out to obtain a characteristic sub-picture->
(ii) a The pooling layer->
For sub-picture->
Performing maximum pooling treatment to obtain characteristic sub-map->
。
With reference to the first possible implementation manner of the first aspect, in a sixth possible implementation manner of the first aspect, the full connection layer includes a sub connection layer
Sub-connecting layer->
Sub-connecting layer->
And sub-connecting layer>
Said sub-junction layer->
The first intermediate feature extraction module is connected with the first trunk network and used for integrating the first feature extraction result to obtain a first intermediate feature; said sub-connecting layer +>
The second intermediate feature extraction module is connected with the second trunk network and used for integrating the second feature extraction result to obtain a second intermediate feature; said sub-connecting layer->
Are connected to the sub-layer->
And said sub-junction layer->
A connection for integrating the first intermediate feature and the second intermediate feature to obtain a third intermediate feature; the sub-connection layer
Is connected to the said sub-layer->
And the full connection is used for integrating the third intermediate features to obtain an integrated feature result so as to map the integrated feature result to the classifier.
In a second aspect, an embodiment of the present application provides an artificial intelligence-based method for identifying hidden dangers of a geological disaster, which is applied to any one of the first aspect or possible implementation manners of the first aspect, and the method includes: acquiring an elevation image set of a monitored area, wherein the elevation image set comprises a current elevation image, a first reference image and a second reference image of the monitored area, and the first reference image and the current elevation image are separated by an interval
A month, the second reference image being spaced from the current elevation image >>
Moon,. Sup.,>
，
(ii) a Determining a first elevation difference image based on the current elevation image and the first reference image, and determining a second elevation difference image based on the current elevation image and the second reference image; inputting the first elevation difference image and the second elevation difference image into a preset geological disaster hidden danger identification model to obtain the geological disaster hidden danger identification modelAnd (5) identifying the hidden danger of the geological disaster.
In a third aspect, an embodiment of the present application provides a storage medium, where the storage medium includes a stored program, and when the program runs, the device where the storage medium is located is controlled to execute the artificial intelligence based geological disaster potential identification method according to the second aspect.
In a fourth aspect, an embodiment of the present application provides an electronic device, which includes a memory and a processor, where the memory is configured to store information including program instructions, and the processor is configured to control execution of the program instructions, where the program instructions are loaded and executed by the processor to implement the artificial intelligence based geological disaster potential identification method according to the second aspect.
In order to make the aforementioned objects, features and advantages of the present application more comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments of the present application will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and that those skilled in the art can also obtain other related drawings based on the drawings without inventive efforts.
Fig. 1 is a schematic diagram of a geological disaster hidden danger identification model provided in an embodiment of the present application.
Fig. 2 is a schematic diagram of a geological disaster hidden danger identification system based on artificial intelligence provided in an embodiment of the present application.
Fig. 3 is a flowchart of a geological disaster hidden danger identification method based on artificial intelligence according to an embodiment of the present application.
Fig. 4 is a block diagram of an electronic device according to an embodiment of the present disclosure.
An icon: 10-a geological disaster hidden danger identification system based on artificial intelligence; 11-an elevation image set acquisition unit; 12-a height difference image processing unit; 13-a geological disaster hidden danger identification unit; 20-an electronic device; 21-a memory; 22-a communication module; 23-a bus; 24-a processor.
Detailed Description
The technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application.
In order to facilitate understanding of the scheme, a geological disaster hidden danger identification model is introduced in the embodiment.
Referring to fig. 1, fig. 1 is a schematic view of a geological disaster hidden danger identification model provided in an embodiment of the present application. The geological disaster hidden danger identification model can comprise a preprocessing layer, a first main network, a second main network, a full connection layer and an output layer.
Illustratively, the preprocessing layer is mainly used for preprocessing the input elevation difference images (such as the first elevation difference image and the second elevation difference image) so as to obtain normalized input images (the first input image and the second input image).
Specifically, the preprocessing layer may calculate a gray value corresponding to each pixel by using an elevation difference corresponding to each pixel in the elevation difference image, so as to construct an input image corresponding to the elevation difference image. The quantity of the pixels in the constructed input image is consistent with that of the pixels in the elevation difference image.
For example, the preprocessing layer may calculate the gray value corresponding to each pixel in the height difference image by using the following formula:
wherein the content of the first and second substances,
is the ^ th or greater in the elevation difference image>
The gray value corresponding to each pixel is->
Is the ^ th or greater in the elevation difference image>
Elevation difference corresponding to each pixel>
For the maximum value of the elevation difference in the elevation difference image->
Is the minimum value of the elevation difference in the elevation difference image,
。
by using the conversion mode, the elevation difference of each pixel in the elevation difference image can be converted into a gray value, so that a corresponding input image is obtained, the elevation difference is reflected to the gray value, and the neural network model can conveniently perform processing such as feature extraction and feature mapping.
Illustratively, the first backbone network is connected to the preprocessing layer, and is configured to perform feature extraction on an input image (first input image) input to the first backbone network to obtain a first feature extraction result.
Specifically, the first backbone network may include a plurality of convolutional layers and pooling layers, which are staggered. For example, the first backbone network may include convolutional layers
The pooling layer is used for collecting and storing the blood>
And the convolutional layer->
And pooling layer>
。
Convolutional layer
For convolution processing a first input image resulting in a characteristic sub-picture->
(ii) a Layer in pool->
For characteristic sub-picture->
Performing maximum pooling treatment to obtain characteristic sub-map->
(ii) a Convolutional layer>
For selecting characteristic sub-picture->
Convolution processing is carried out to obtain a characteristic sub-picture->
(ii) a Layer in pool->
For characteristic sub-picture->
Performing maximal pooling to obtain characteristic sub-map->
。/>
Illustratively, the second backbone network is connected to the preprocessing layer, and is configured to perform feature extraction on an input image (second input image) input to the second backbone network, so as to obtain a second feature extraction result.
Specifically, the second backbone network may also include a plurality of convolutional layers and pooling layers, and the convolutional layers and the pooling layers are arranged in an interlaced manner. For example, the second backbone network may include convolutional layers
The pooling layer is used for collecting and storing the blood>
And convolution of the twoLayer->
And pooling layer>
。
Convolutional layer
Is used for carrying out convolution processing on the second input image to obtain a characteristic sub image->
(ii) a Layer in pool->
For selecting characteristic sub-picture->
Performing maximum pooling treatment to obtain characteristic sub-map->
(ii) a Entrapment layer->
For selecting characteristic sub-picture->
Convolution processing is carried out to obtain a characteristic sub-picture->
(ii) a Pooling layer>
For characteristic sub-picture->
Performing maximal pooling to obtain characteristic sub-map->
。
Based on the size of the input image, the first backbone network and the second backbone network can also design more convolution layers and pooling layers so as to extract more features and improve the identification accuracy.
Illustratively, the full connectivity layer is connected to the first backbone network and the second backbone network, respectively, and is configured to integrate the first feature extraction result and the second feature extraction result.
In particular, the fully-connected layer may include a sub-connected layer
Sub-connecting layer->
Sub-connecting layer->
And sub-connecting layer>
。
Sub-connection layer
And the first intermediate feature extraction module is connected with the first backbone network and used for integrating the first feature extraction result to obtain a first intermediate feature. Sub-connecting layer->
And the second intermediate feature extraction module is connected with the second backbone network and used for integrating the second feature extraction result to obtain a second intermediate feature. Sub-connecting layer->
Are respectively connected with the sub-connecting layer>
And the sub-junction layer->
And the connection is used for integrating the first intermediate characteristic and the second intermediate characteristic to obtain a third intermediate characteristic. Sub-connecting layer->
Is connected with the sonLayer->
And the full connection is used for integrating the third intermediate features to obtain an integrated feature result.
Illustratively, the output layer is connected to the fully-connected layer via a classifier (e.g., softmax, facilitating multi-classification), a sub-connected layer of the fully-connected layer
The integrated feature result can be mapped into a classifier, and the classifier classifies the integrated feature result to obtain a geological disaster hidden danger identification result and outputs the result through an output layer.
The above is an introduction of a model architecture of the geological disaster hidden danger identification model, and here, training of the geological disaster hidden danger identification model will be further introduced so as to obtain an applicable geological disaster hidden danger identification model.
First, a training set needs to be constructed. The training image may be selected from a verified area (for example, an area where an accident such as collapse or landslide occurs, and an elevation image set previously stored in the area is selected) or an elevation image set of an expert-labeled hidden danger area, where the elevation image set of the hidden danger area may include a plurality of elevation images, for example, an elevation image at a first time point (as a current elevation image), a plurality of elevation images at a second time point (as a first reference image), and a plurality of elevation images at a third time point (as a second reference image), and the elevation image at the second time point (the first reference image) is separated from the elevation image at the first time point (the current elevation image)
Monthly, third time point elevation image (second reference image) and first time point elevation image (current elevation image) are separated>
Moon,. Sup.,>
，/>
。
selecting an elevation image set of 200-400 hidden danger areas, and performing the following steps: 2: the scale of 1 is divided into a training set, a validation set, and a test set.
And utilizing an elevation image set in a training set (one elevation image at a first time point, one elevation image at a second time point and one elevation image at a third time point in the elevation image set are selected each time) to obtain a first elevation difference image (determined based on the elevation image at the first time point and the elevation image at the second time point) and a second elevation difference image (determined based on the elevation image at the first time point and the elevation image at the third time point) as input of the geological disaster hidden danger identification model, and training the model. The training process of the model is approximately equivalent to the training process of the existing neural network model, and the parameters of the model are optimized through a designed loss function (such as square absolute error loss, ioU loss or CIoU loss), so that the training of the model is completed.
And then, the model can be verified by using the elevation image set in the verification set, the model is evaluated by using indexes such as accuracy, recall rate and accuracy, and the model is tested by using the test set, so that the trained geological disaster hidden danger identification model can be obtained.
Based on the geological disaster hidden danger identification model, the embodiment provides a geological disaster hidden danger identification system 10 based on artificial intelligence. Referring to fig. 2, fig. 2 is a schematic diagram of a geological disaster risk identification system 10 based on artificial intelligence according to an embodiment of the present disclosure. The artificial intelligence based geological disaster risk identification system 10 may include an elevation image set acquisition unit 11, an elevation difference image processing unit 12, and a geological disaster risk identification unit 13 (in which a geological disaster risk identification model is built).
For ease of description of the operation of the artificial intelligence based geological disaster risk identification system 10, it is described herein in conjunction with an artificial intelligence based geological disaster risk identification method (which may be performed by the electronic device 20). Referring to fig. 3, the method for identifying potential hazards of geological disasters based on artificial intelligence may include step S10, step S20 and step S30.
In order to identify the potential hazards of the geological disaster, the electronic device 20 may execute step S10 to implement the function of the high-range image set obtaining unit 11 in the artificial intelligence based geological disaster potential hazard identification system 10.
Step S10: acquiring an elevation image set of a monitored area, wherein the elevation image set comprises a current elevation image, a first reference image and a second reference image of the monitored area, and the first reference image and the current elevation image are separated by an interval
Month, the second reference image being spaced ∑ from the current elevation image>
Moon->
，/>
。
In this embodiment, the electronic device 20 may acquire an elevation image set of the monitored area. The elevation images of the monitored area may be oversized elevation images (a current elevation image, a first reference image, and a second reference image). In order to facilitate the processing of the geological disaster hidden danger identification model, the original oversized elevation image may be divided into a plurality of elevation images with standard sizes, for example, a plurality of elevation images (the size of which represents the grid number of the elevation images) divided into 128 × 128 (or 256 × 256, 64 × 64, etc.), each of the divided elevation images corresponds to one number, three elevation images corresponding to the same number and derived from different original elevation images (one of the current elevation image, the first reference image, and the second reference image) form a new elevation image set, and the same number is used to obtain a plurality of elevation image sets, where the plurality of elevation image sets are all elevation image sets of the monitored area (different areas).
The first reference image is separated from the current elevation image
Month, second reference image spaced from current elevation image>
The time of the month is,
，/>
. Here, in order to +>
，/>
By way of example, but not limitation.
After obtaining the elevation image set of the monitored area, the electronic device 20 may execute step S20 to implement the function of the elevation difference image processing unit 12.
Step S20: determining a first elevation difference image based on the current elevation image and the first reference image, and determining a second elevation difference image based on the current elevation image and the second reference image.
In this embodiment, after the electronic device 20 may register the current elevation image with the first reference image, the elevation difference corresponding to each pixel is obtained (whether the elevation difference is positive or negative or positive or negative is distinguished by the sign "-") by using the elevation of each pixel in the current elevation image and the first reference image, so that the first elevation difference image may be determined. Similarly, the electronic device 20 may determine a second elevation difference image using the current elevation image and the second reference image.
After the first elevation difference image and the second elevation difference image are determined, the electronic device 20 may execute step S30 to realize the function of the geological disaster hidden danger identifying unit 13, and a geological disaster hidden danger identifying model is preset in the geological disaster hidden danger identifying unit 13.
Step S30: and inputting the first elevation difference image and the second elevation difference image into a preset geological disaster hidden danger identification model to obtain a geological disaster hidden danger identification result output by the geological disaster hidden danger identification model.
In this embodiment, the electronic device 20 may input the first elevation difference image and the second elevation difference image into a preset geological disaster risk identification model.
Based on this, the electronic device 20 may generate the first input image based on the elevation difference corresponding to each pixel in the first elevation difference image and generate the second input image based on the elevation difference corresponding to each pixel in the second elevation difference image by using the preprocessing layer of the identification model for identifying potential hazards of geological disasters.
For example, the electronic device 20 may calculate a first gray value corresponding to each pixel based on the height difference corresponding to each pixel in the first height difference image by using the preprocessing layer.
For example, the following formula may be used to calculate the first gray-scale value corresponding to each pixel in the first height difference image:
wherein, the first and the second end of the pipe are connected with each other,
is the fifth or fifth letter in the first altitude difference image>
A first gray value corresponding to each pixel->
Is the fifth or fifth letter in the first altitude difference image>
Elevation difference corresponding to each pixel>
Is the maximum elevation difference in the first elevation difference imageValue,. Or>
Is the minimum value of the elevation difference in the first elevation difference image->
。
After the first gray value corresponding to each pixel in the first elevation difference image is calculated, a first input image corresponding to the first elevation difference image may be constructed based on the first gray value corresponding to each pixel. That is, a gray image (e.g., 128 × 128) with a corresponding size is constructed as the first input image according to the first gray value corresponding to each pixel.
Similarly, the electronic device 20 may calculate, by using a preprocessing layer of the geological disaster hidden danger identification model, a second gray scale value corresponding to each pixel according to the elevation difference corresponding to each pixel in the second elevation difference image.
For example, the following formula may be used to calculate the second gray scale value corresponding to each pixel in the second elevation difference image:
wherein the content of the first and second substances,
is the ^ th or greater in the second elevation difference image>
A second gray value corresponding to individual pixel>
Is the ^ th or greater in the second elevation difference image>
The elevation difference corresponding to each pixel is greater or less>
Is the maximum value of the elevation difference in the second elevation difference image->
Is the minimum value of the elevation difference in the second elevation difference image->
。
After the second gray scale value corresponding to each pixel in the second elevation difference image is calculated, a second input image corresponding to the second elevation difference image can be constructed based on the second gray scale value corresponding to each pixel. That is, a gray image (e.g., 128 × 128) with a corresponding size is constructed as the second input image according to the second gray value corresponding to each pixel.
By using the conversion mode, the elevation difference of each pixel in the first elevation difference image (or the second elevation difference image) can be converted into a gray value, so that a corresponding first input image (or the second input image) is obtained, the elevation difference is reflected to the gray value, a neural network model can conveniently perform processing such as feature extraction and feature mapping, and finally recognition of the hidden danger of the geological disaster is realized.
After the first input image and the second input image are obtained, the first input image may be input to the first backbone network, and the second input image may be input to the second backbone network.
The first trunk network is connected with the preprocessing layer and used for carrying out feature extraction on an input first input image to obtain a first feature extraction result.
Illustratively, the first backbone network comprises convolutional layers
The pooling layer is used for collecting and storing the blood>
And the convolutional layer->
And pooling layer>
For example: entrapment layer->
A convolution processing of the first input image can be carried out, resulting in a characteristic sub-picture->
(ii) a Layer in pool->
Can be used for characteristic subgraph
Performing maximal pooling to obtain characteristic sub-map->
(ii) a Entrapment layer->
Can be taken over a characteristic sub-picture>
Convolution processing is carried out to obtain a characteristic sub-picture->
(ii) a Layer in pool->
Can be taken over a characteristic sub-picture>
Performing maximum pooling treatment to obtain characteristic sub-map->
. Here, the characteristic sub-picture->
I.e. representing the first feature extraction result.
And the second trunk network is connected with the preprocessing layer and is used for carrying out feature extraction on the input second input image to obtain a second feature extraction result.
Illustratively, the second backbone network comprises convolutional layers
The pooling layer is used for collecting and storing the blood>
And the convolutional layer->
And pooling layer>
For example, the following steps are carried out: entrapment layer->
The second input image may be convolved resulting in a characteristic sub-picture->
(ii) a Layer in pool->
Can map features
Performing maximum pooling treatment to obtain characteristic sub-map->
(ii) a Entrapment layer->
Can be taken over a characteristic sub-picture>
Convolution processing is carried out to obtain a characteristic sub-picture->
(ii) a Layer in pool->
Can be taken over a characteristic sub-picture>
Performing maximum pooling treatment to obtain characteristic sub-map->
. Here, the characteristic sub-picture->
I.e. representing the second feature extraction result.
The full connection layer is respectively connected with the first trunk network and the second trunk network and is used for integrating the first feature extraction result and the second feature extraction result.
Illustratively, the full connection layer comprises a sub-connection layer
Sub-connecting layer->
Sub-connecting layer->
And the sub-junction layer->
For example, the following steps are carried out:
sub-connection layer
And the first intermediate feature extraction module is connected with the first backbone network and used for integrating the first feature extraction result to obtain a first intermediate feature. The first intermediate feature is obtained by integrating the extracted features of the first input image after the first trunk network extracts the features of the first input image.
Sub-connection layer
And the second intermediate feature extraction module is connected with the second backbone network and used for integrating the second feature extraction result to obtain a second intermediate feature. The second intermediate feature is obtained by integrating the extracted features of the second input image after the second input image is subjected to feature extraction by the second backbone network.
Sub-connection layer
Respectively connected with sub-connecting layer->
And the sub-junction layer->
In conjunction with the first intermediate feature and the second intermediate feature, a third intermediate feature may be integrated.
And sub-connection layer
And sub-connecting layer>
And the full connection is used for integrating the third intermediate features to obtain an integrated feature result so as to map the integrated feature result to a classifier (for example, a softmax function, and the set categories may include multiple geological disaster hidden danger categories such as landslide hidden danger, collapse hidden danger, debris flow hidden danger, collapse hidden danger and the like).
The output layer can be connected with the full-connection layer through the classifier, the classifier classifies the integrated feature result to obtain a geological disaster hidden danger identification result (the final geological disaster hidden danger category is determined through the probability of each category), and the geological disaster hidden danger identification result is output through the output layer.
The hidden danger areas are identified on the basis of the characteristics of the short-term deformation and the medium-term deformation of the monitoring areas, the characteristics of the geological disaster hidden danger areas in the short-term and the medium-term can be considered, and the accuracy of identification of the geological disaster hidden danger areas is improved.
Because each elevation image set can be a small part of the monitoring area, the hidden danger areas with the geological disaster hidden dangers can be identified from the monitoring areas, the corresponding parts can be determined from the monitoring areas through the numbers of the elevation image sets to serve as the hidden danger areas, and accurate hidden danger area positioning can be achieved.
Referring to fig. 4, fig. 4 is a block diagram of an electronic device 20 according to an embodiment of the present disclosure. The electronic device 20 may include: a communication module 22 connected to the outside world via a network, one or more processors 24 for executing program instructions, a bus 23, and a different form of memory 21, such as a disk, ROM, or RAM, or any combination thereof. The memory 21, the communication module 22, and the processor 24 may be connected by a bus 23.
Illustratively, the memory 21 has stored therein a program. The processor 24 may call and run these programs from the memory 21, so as to implement a method for identifying potential hazards of a geological disaster based on artificial intelligence by running the programs, and further implement identification of potential hazards of a geological disaster in a monitored area.
The embodiment of the application also provides a storage medium which comprises a stored program, wherein when the program runs, the equipment where the storage medium is located is controlled to execute the artificial intelligence-based geological disaster hidden danger identification method.
To sum up, the embodiments of the present application provide a system, a method and an electronic device for identifying hidden dangers of geological disasters based on artificial intelligence, in which an elevation image set (a current elevation image, a first reference image and a second reference image, the first reference image and the current elevation image being separated by a gap) of a monitored area is obtained through an elevation image set obtaining unit 11
The second reference image is used as a basis for observing the deformation of the monitoring area in the short and medium periods, and the second reference image is separated from the current elevation image>
And the image is used as a basis for observing long-term deformation in the monitored area), further determining a first elevation difference image and a second elevation difference image, inputting the first elevation difference image and the second elevation difference image into a geological disaster hidden danger identification model, and carrying out intelligent identification to obtain a geological disaster hidden danger identification result. In such a way, the intelligent identification of the geological disaster can be realized by the neural network model according to the first elevation difference image (reflecting the short-term deformation condition of the monitored area) and the second elevation difference image (reflecting the medium-term deformation condition of the monitored area) of the height Cheng Tuxiang of the monitored area, so that the dependence on expert experience can be reduced, the identification of the hidden danger area is made before the geological disaster occurs, corresponding precautionary measures can be made in advance, for example, the operation is far away from the hidden danger area, and the human activities are avoided in the hidden conditionIn the affected area, etc.
In the embodiments provided in the present application, it should be understood that the disclosed system and method may be implemented in other ways. The above-described system embodiments are merely illustrative, and for example, the division of the units is only one logical functional division, and there may be other divisions in actual implementation, and for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed coupling or direct coupling or communication connection between each other may be through some communication interfaces, indirect coupling or communication connection between devices or units, and may be in an electrical, mechanical or other form.
In addition, units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In this document, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions.
The above description is only an example of the present application and is not intended to limit the scope of the present application, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims (10)
1. The utility model provides a geological disaster hidden danger identification system based on artificial intelligence which characterized in that includes:
an elevation image set acquisition unit, configured to acquire an elevation image set of a monitored area, where the elevation image set includes the elevation image setMonitoring a current elevation image, a first reference image, and a second reference image of an area, wherein the first reference image is spaced apart from the current elevation image
Month, the second reference image being spaced ∑ from the current elevation image>
Moon,. Sup.,>
，
；
an elevation difference image processing unit, configured to determine a first elevation difference image based on the current elevation image and the first reference image, and determine a second elevation difference image based on the current elevation image and the second reference image;
and the geological disaster hidden danger identification unit is preset with a geological disaster hidden danger identification model and used for inputting the first elevation difference image and the second elevation difference image into the geological disaster hidden danger identification model to obtain a geological disaster hidden danger identification result output by the geological disaster hidden danger identification model.
2. The artificial intelligence based geological disaster potential identification system of claim 1, wherein the geological disaster potential identification model comprises a preprocessing layer, a first backbone network, a second backbone network, a full connection layer and an output layer,
the preprocessing layer is used for generating a first input image based on the elevation difference corresponding to each pixel in the first elevation difference image and generating a second input image based on the elevation difference corresponding to each pixel in the second elevation difference image;
the first trunk network is connected with the preprocessing layer and used for performing feature extraction on the input first input image to obtain a first feature extraction result;
the second trunk network is connected with the preprocessing layer and is used for performing feature extraction on the input second input image to obtain a second feature extraction result;
the full connection layer is respectively connected with the first trunk network and the second trunk network and is used for integrating the first feature extraction result and the second feature extraction result;
the output layer is connected with the full connection layer through a classifier, the classifier classifies the integrated feature result to obtain a geological disaster hidden danger identification result, and the geological disaster hidden danger identification result is output through the output layer.
3. The artificial intelligence based geological disaster hidden danger identification system according to claim 2, wherein said preprocessing layer is specifically configured to:
calculating a first gray value corresponding to each pixel based on the height difference corresponding to each pixel in the first height difference image, and constructing a first input image corresponding to the first height difference image based on the first gray value corresponding to each pixel;
and calculating a second gray value corresponding to each pixel based on the height difference corresponding to each pixel in the second height difference image, and constructing a second input image corresponding to the second height difference image based on the second gray value corresponding to each pixel.
4. The artificial intelligence based geological disaster hidden danger identification system according to claim 3, wherein said preprocessing layer is specifically configured to:
calculating a first gray value corresponding to each pixel in the first height difference image by adopting the following formula:
wherein the content of the first and second substances,
is the ^ th or greater in the first difference of elevation image>
A first gray value corresponding to each pixel +>
Is the ^ th or greater in the first difference of elevation image>
Elevation difference corresponding to each pixel>
For a maximum value of the elevation difference in the first elevation difference image->
For a minimum value of the elevation difference in the first elevation difference image->
；
Calculating a second gray value corresponding to each pixel in the second elevation difference image by adopting the following formula:
wherein the content of the first and second substances,
is the ^ th or greater in the second elevation difference image>
A second gray value corresponding to individual pixel>
Is a second elevation difference image>
Elevation difference corresponding to each pixel>
For a maximum value of the elevation difference in the second elevation difference image>
For a minimum value of the elevation difference in the second elevation difference image->
。
5. The artificial intelligence-based geological disaster potential identification system as claimed in claim 2, wherein said first backbone network comprises convolutional layers
The pooling layer is used for collecting and storing the blood>
And a convolution layer>
And pooling layer>
，
The convolutional layer
Is used for carrying out convolution processing on the first input image to obtain a characteristic sub image->
；
The pooling layer
For sub-picture->
Performing maximum pooling treatment to obtain characteristic sub-map->
；
The convolutional layer
For sub-picture->
Convolution processing is carried out to obtain a characteristic sub-picture->
；
The pooling layer
For sub-picture->
Performing maximum pooling treatment to obtain characteristic sub-map->
。
6. The artificial intelligence-based geological disaster potential identification system as claimed in claim 2, wherein said second backbone network comprises convolutional layers
And pooling layer>
And the convolutional layer->
And pooling layer>
，
The convolutional layer
For performing convolution processing on the second input image to obtain a characteristic sub-image->
；
The pooling layer
For sub-picture->
Performing maximum pooling treatment to obtain characteristic sub-map->
；
The convolutional layer
For sub-picture->
Convolution processing is carried out to obtain a characteristic sub-picture->
；
The pooling layer
For sub-picture->
Performing maximum pooling treatment to obtain characteristic sub-map->
。
7. The artificial intelligence-based geological disaster potential risk identification system according to claim 2, wherein said full-link layer comprises sub-link layers
Sub-connecting layer->
Sub-connecting layer->
And the sub-junction layer->
，
The sub-connection layer
The first intermediate feature extraction module is connected with the first trunk network and used for integrating the first feature extraction result to obtain a first intermediate feature;
the sub-connection layer
The second intermediate feature extraction module is connected with the second trunk network and used for integrating the second feature extraction result to obtain a second intermediate feature;
the sub-connection layer
Are connected to the sub-layer->
And said sub-connecting layer +>
A connection for integrating the first intermediate feature and the second intermediate feature to obtain a third intermediate feature;
the sub-connection layer
Is connected to the said sub-layer->
And the full connection is used for integrating the third intermediate features to obtain an integrated feature result so as to map the integrated feature result to the classifier.
8. An artificial intelligence based geological disaster hidden danger identification method is applied to the artificial intelligence based geological disaster hidden danger identification system of any one of claims 1 to 7, and the method comprises the following steps:
acquiring an elevation image set of a monitored area, wherein the elevation image set comprises a current elevation image, a first reference image and a second reference image of the monitored area, and the first reference image and the current elevation image are separated by a gap
A month, the second reference image being spaced from the current elevation image >>
Moon->
，/>
；
Determining a first elevation difference image based on the current elevation image and the first reference image, and determining a second elevation difference image based on the current elevation image and the second reference image;
and inputting the first elevation difference image and the second elevation difference image into a preset geological disaster hidden danger identification model to obtain a geological disaster hidden danger identification result output by the geological disaster hidden danger identification model.
9. A storage medium comprising a stored program, wherein when the program is executed, the apparatus on which the storage medium is located is controlled to execute the method for identifying potential hazards in geological disasters based on artificial intelligence according to claim 8.
10. An electronic device comprising a memory for storing information including program instructions and a processor for controlling execution of the program instructions, the program instructions being loaded and executed by the processor to implement the artificial intelligence based geological disaster potential identification method of claim 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310182148.0A CN115908954B (en) | 2023-03-01 | 2023-03-01 | Geological disaster hidden danger identification system and method based on artificial intelligence and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310182148.0A CN115908954B (en) | 2023-03-01 | 2023-03-01 | Geological disaster hidden danger identification system and method based on artificial intelligence and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115908954A true CN115908954A (en) | 2023-04-04 |
| CN115908954B CN115908954B (en) | 2023-07-28 |
Family
ID=86495068
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310182148.0A Active CN115908954B (en) | 2023-03-01 | 2023-03-01 | Geological disaster hidden danger identification system and method based on artificial intelligence and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115908954B (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180158210A1 (en) * | 2015-08-26 | 2018-06-07 | Digitalglobe, Inc. | Synthesizing training data for broad area geospatial object detection |
| CN109376777A (en) * | 2018-10-18 | 2019-02-22 | 四川木牛流马智能科技有限公司 | Cervical cancer tissues pathological image analysis method and equipment based on deep learning |
| CN110057344A (en) * | 2019-04-30 | 2019-07-26 | 北京交通大学 | Settle detection method and platform |
| CN112464807A (en) * | 2020-11-26 | 2021-03-09 | 北京灵汐科技有限公司 | Video motion recognition method and device, electronic equipment and storage medium |
| CN112732092A (en) * | 2021-01-22 | 2021-04-30 | 河北工业大学 | Surface electromyogram signal identification method based on double-view multi-scale convolution neural network |
| CN113159174A (en) * | 2021-04-21 | 2021-07-23 | 云南师范大学 | DEM (digital elevation model) matching and deformation quantity detecting method and device without control points |
| CN114219687A (en) * | 2021-11-02 | 2022-03-22 | 三峡大学 | Intelligent identification method for potential construction safety hazards by fusing human-computer vision |
| CN114267155A (en) * | 2021-11-05 | 2022-04-01 | 国能大渡河革什扎水电开发有限公司 | Geological disaster monitoring and early warning system based on video recognition technology |
| CN114821376A (en) * | 2022-06-27 | 2022-07-29 | 中咨数据有限公司 | Unmanned aerial vehicle image geological disaster automatic extraction method based on deep learning |
| CN115019145A (en) * | 2022-06-15 | 2022-09-06 | 交通运输部公路科学研究所 | Method for monitoring settlement deformation of road surface of Qinghai-Tibet plateau highway |
| CN115272874A (en) * | 2022-09-28 | 2022-11-01 | 中国科学院、水利部成都山地灾害与环境研究所 | Debris flow disaster identification and frequency calculation method based on remote sensing image |
-
2023
- 2023-03-01 CN CN202310182148.0A patent/CN115908954B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180158210A1 (en) * | 2015-08-26 | 2018-06-07 | Digitalglobe, Inc. | Synthesizing training data for broad area geospatial object detection |
| CN109376777A (en) * | 2018-10-18 | 2019-02-22 | 四川木牛流马智能科技有限公司 | Cervical cancer tissues pathological image analysis method and equipment based on deep learning |
| CN110057344A (en) * | 2019-04-30 | 2019-07-26 | 北京交通大学 | Settle detection method and platform |
| CN112464807A (en) * | 2020-11-26 | 2021-03-09 | 北京灵汐科技有限公司 | Video motion recognition method and device, electronic equipment and storage medium |
| CN112732092A (en) * | 2021-01-22 | 2021-04-30 | 河北工业大学 | Surface electromyogram signal identification method based on double-view multi-scale convolution neural network |
| CN113159174A (en) * | 2021-04-21 | 2021-07-23 | 云南师范大学 | DEM (digital elevation model) matching and deformation quantity detecting method and device without control points |
| CN114219687A (en) * | 2021-11-02 | 2022-03-22 | 三峡大学 | Intelligent identification method for potential construction safety hazards by fusing human-computer vision |
| CN114267155A (en) * | 2021-11-05 | 2022-04-01 | 国能大渡河革什扎水电开发有限公司 | Geological disaster monitoring and early warning system based on video recognition technology |
| CN115019145A (en) * | 2022-06-15 | 2022-09-06 | 交通运输部公路科学研究所 | Method for monitoring settlement deformation of road surface of Qinghai-Tibet plateau highway |
| CN114821376A (en) * | 2022-06-27 | 2022-07-29 | 中咨数据有限公司 | Unmanned aerial vehicle image geological disaster automatic extraction method based on deep learning |
| CN115272874A (en) * | 2022-09-28 | 2022-11-01 | 中国科学院、水利部成都山地灾害与环境研究所 | Debris flow disaster identification and frequency calculation method based on remote sensing image |
Non-Patent Citations (3)
| Title |
|---|
| GUANGPENG FAN 等: "Recognizing Multiple Types of Rocks Quickly and Accurately Based on Lightweight CNNs Model", 《IEEE ACCESS》, vol. 8, pages 55269, XP011780456, DOI: 10.1109/ACCESS.2020.2982017 * |
| 田凡: "时序InSAR技术在南水北调中线形变监测中的应用研究", 《中国优秀硕士学位论文全文数据库工程科技Ⅱ辑》, no. 1, pages 037 - 169 * |
| 韩少帅: "2010-2019年南极冰盖的高程变化与物质平衡", 《中国优秀硕士学位论文全文数据库基础科学辑》, no. 2, pages 012 - 49 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115908954B (en) | 2023-07-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108537191B (en) | Three-dimensional face recognition method based on structured light camera | |
| CN109543526B (en) | True and false facial paralysis recognition system based on depth difference characteristics | |
| CN111754396B (en) | Face image processing method, device, computer equipment and storage medium | |
| CN110222787A (en) | Multiscale target detection method, device, computer equipment and storage medium | |
| CN111738044B (en) | Campus violence assessment method based on deep learning behavior recognition | |
| CN112561876B (en) | Image-based water quality detection method and system for ponds and reservoirs | |
| JP2016538671A (en) | Method and system for extracting facial features from facial image data | |
| CN112950780B (en) | Intelligent network map generation method and system based on remote sensing image | |
| CN115620384B (en) | Model training method, fundus image prediction method and fundus image prediction device | |
| CN108710950A (en) | A kind of image quantization analysis method | |
| CN112686184A (en) | Remote sensing house change detection method based on neural network | |
| CN114299082A (en) | New coronary pneumonia CT image segmentation method, device and storage medium | |
| CN115424056A (en) | Model training method and device, image classification method, equipment and medium | |
| CN114998667A (en) | Multispectral target detection method, multispectral target detection system, computer equipment and storage medium | |
| CN115908954A (en) | Geological disaster hidden danger identification system and method based on artificial intelligence and electronic equipment | |
| CN111325756A (en) | Three-dimensional image artery and vein segmentation method and system based on deep learning network | |
| CN113361494B (en) | Self-service method and self-service system based on face recognition | |
| CN115457366A (en) | Chinese herbal medicine multi-label recognition model based on graph convolution neural network | |
| CN115731620A (en) | Method for detecting counter attack and method for training counter attack detection model | |
| CN113657293A (en) | Living body detection method, living body detection device, electronic apparatus, medium, and program product | |
| CN115346091A (en) | Method and device for generating Mura defect image data set | |
| CN115223157A (en) | Power grid equipment nameplate optical character recognition method based on recurrent neural network | |
| CN115083006A (en) | Iris recognition model training method, iris recognition method and iris recognition device | |
| Michel et al. | Rapid change detection algorithm for disaster management | |
| CN114582012A (en) | Skeleton human behavior recognition method, device and equipment |
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 |