CN114169404A - Method for intelligently acquiring quantitative information of slope diseases based on images - Google Patents
Method for intelligently acquiring quantitative information of slope diseases based on images Download PDFInfo
- Publication number
- CN114169404A CN114169404A CN202111357743.0A CN202111357743A CN114169404A CN 114169404 A CN114169404 A CN 114169404A CN 202111357743 A CN202111357743 A CN 202111357743A CN 114169404 A CN114169404 A CN 114169404A
- Authority
- CN
- China
- Prior art keywords
- disease
- slope
- image
- area
- neural network
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a method for acquiring a slope disease based on an image, and S1, a slope image is acquired; s2, identifying the slope image by using a preset target positioning neural network to obtain a first identification area, wherein the first identification area is a disease area; s3, identifying the disease information in the first identification area by using a preset pixel-level classification neural network to obtain a second identification result; and S4, calculating the disease number, area and length information of the slope disease according to the first identification area and the second identification result. According to the method, the disease position information in the slope image is acquired by using the target positioning neural network, and the slope disease information in the slope image is identified by using the pixel-level classification neural network, so that the pixel information of the disease can be acquired more accurately. The invention calculates the length information of the disease by using the length of the major axis of the ellipse with the same normalized second-order central moment as that of a part of pixel areas, and can more accurately acquire the length and area equivalent information of the disease.
Description
Technical Field
The invention relates to the technical field of images, in particular to a method and a device for acquiring a slope disease based on an image and electronic equipment.
Background
The apparent disease detection and degree analysis have important significance in the field of slope health monitoring. Currently, in the context of slopes, the existing overall technical situation is assessed by: workers regularly patrol the overall appearance of the slope, record the disease types of key parts (such as lattice beams, retaining walls, vegetation conditions and the like) of the slope, and judge the damage degree of the parts according to quantitative information such as the area, the length, the width and the like of the diseases. And finally, counting the number and the degree of the diseases, and obtaining the overall safety condition of the slope through quantitative summary information of the diseases. However, for large civil slopes such as slopes, the height difference between the slope surface of the slope and the road surface is large, the number of the slopes is large, the generated data volume is huge, and the method relying on manual observation is long, low in efficiency and subjective. Therefore, how to quickly detect the apparent diseases of the side slope (such as the side slope) and accurately obtain the quantitative information of each disease is a key problem faced by the current evaluation of the whole technical condition of the side slope.
With the rapid development of the slope disease detection technology in recent years and the continuous progress of the computer technology, the disease of the slope is detected by using an artificial intelligent method. However, the artificial intelligence based approach requires a good and robust intelligent model, which does not leave a large amount of training data sets. In terms of the input end of the artificial intelligence method, numerical parameters (such as structural modal strain energy, acceleration signals, shape and position information and the like) or image data can be used as a training data source for slope disease identification. However, for such large slopes (e.g. a side slope), the amount of work required to evaluate the overall technical condition of the slope using data from the sensors arranged is enormous. Therefore, whether from the technical level of the collection worker or the intuitive result, the image data is more easily obtained by comparing the numerical parameters. However, it is difficult and time-consuming to capture an apparent image of the slope by merely manually climbing the slope with respect to the captured image data itself. Therefore, how to quickly obtain a clear slope appearance image is a problem which needs to be solved urgently at present.
In addition, in addition to solving the problem of image data sets, artificial intelligence methods (models) for image recognition are diverse and varied in nature. Therefore, it is important to find an artificial intelligence model suitable for the slope diseases. In terms of outputting results from image-based artificial intelligence methods, there are typically: image classification, target positioning and pixel level classification. The image classification is to classify all contents in the whole image into a single category, the target positioning can classify and position a plurality of objects in the whole image by using a positioning frame, and the pixel-level classification is to classify each pixel in the whole image. For slopes, it is common that a plurality of different diseases appear in one image. If an artificial intelligence model of an image classification level is used, a plurality of diseases of different types in the image cannot be identified. If an artificial intelligence model of a target positioning layer is used, different disease types of a single slope disease image and the position of the disease in the image can be obtained, but the degree of the disease cannot be quantified (namely information such as the coverage area and the length of the disease). If an artificial intelligence model of a pixel-level classification level is used, the artificial intelligence model can inherit the advantages of the former two (can identify various different diseases in a single slope image and identify the outline positions (positioning) of the diseases), and can also calculate the pixel number of a certain disease so as to obtain the quantitative information of the diseases. However, the existing pixel-level classification artificial intelligence model has the defects that: when the pixel of an object in an image is in a small proportion in the whole image, the model has poor effect of classifying the pixel of the object, and the object is often mistakenly judged as the object with the largest pixel proportion in the image. Therefore, how to accurately identify the diseases (small diseases) with small pixel proportion in the slope disease quantitative information obtained by selecting a proper artificial intelligence model is a problem which needs to be solved urgently at present.
The background description provided herein is for the purpose of generally presenting the context of the disclosure. Unless otherwise indicated herein, the material described in this section is not prior art to the claims in this application and is not admitted to be prior art by inclusion in this section.
Disclosure of Invention
Aiming at the technical problems in the related art, the invention provides a method for acquiring a slope disease based on an image, which comprises the following steps:
s1, acquiring a slope image;
s2, identifying the slope image by using a preset target positioning neural network to obtain a first identification area, wherein the first identification area is a disease area;
s3, identifying the disease information in the first identification area by using a preset pixel-level classification neural network to obtain a second identification result;
and S4, calculating the disease number, area and length information of the slope disease according to the first identification area and the second identification result.
Specifically, a preset target positioning neural network is used for identifying a slope image output positioning frame, and the positioning frame is used for identifying the position of the first identification area on the slope image.
Specifically, the number of the slope apparent diseases is determined by counting the number of the prediction positioning frames of the target recognition neural network.
Specifically, the length information of the side slope disease is obtained as follows:
dividing the slope diseases into n sections, and acquiring the ellipse major axis d of each section by a method of summing n ellipse major axes with pixels in partial areas having the same normalized second-order center distanceiAnd calculating and identifying the length d of the disease according to the major axis of the ellipse of each section,
specifically, the method further comprises the following steps: s5, acquiring the disease degree of the side slope according to the disease quantity, area and length information of the side slope diseases, and pushing corresponding alarm information according to the disease degree.
In a second aspect, another embodiment of the present invention discloses a slope disease device based on image acquisition, which includes the following units:
the side slope image acquisition unit is used for acquiring side slope images;
the first identification area acquisition unit is used for identifying a slope image by using a preset target positioning neural network to acquire a first identification area, wherein the first identification area is a disease area;
the second identification result acquisition unit is used for identifying the disease information in the first identification area by using a preset pixel-level classification neural network to acquire a second identification result;
and the slope disease calculation unit is used for calculating the disease number, area and length information of the slope disease according to the first identification area and the second identification result.
Specifically, a preset target positioning neural network is used for identifying a slope image output positioning frame, and the positioning frame is used for identifying the position of the first identification area on the slope image.
Specifically, the number of the slope apparent diseases is determined by counting the number of the prediction positioning frames of the target recognition neural network.
Specifically, the length information of the side slope disease is obtained as follows:
dividing the slope diseases into n sections, and acquiring the ellipse major axis d of each section by a method of summing n ellipse major axes with pixels in partial areas having the same normalized second-order center distanceiAnd calculating and identifying the length d of the disease according to the major axis of the ellipse of each section,
in a third aspect, another embodiment of the present invention discloses an electronic device, where the electronic device includes a processor and a memory, where the memory stores instructions, and the instructions, when executed by the processor, are used to implement the above method for obtaining a slope disease based on an image.
According to the method, the disease position information in the slope image is obtained by using the target positioning neural network, and the slope disease information in the slope image is identified by using the pixel-level classification neural network according to the disease position. The invention finally obtains the disease information of the slope by using the pixel-level recognition result and the target positioning neural network result. In addition, the invention uses the length of the ellipse long axis with the same normalized second-order central moment as that of a part of pixel areas to calculate the length of the disease and other information, and can more accurately acquire the length and area information of the disease.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
FIG. 1 is a flowchart of a method for acquiring a slope disease based on an image according to an embodiment of the present invention;
FIG. 2 is a schematic view of a flight path of an unmanned aerial vehicle provided by an embodiment of the invention;
fig. 3 is a schematic diagram illustrating a target classification and identification network positioning a slope disease according to an embodiment of the present invention;
FIG. 4 is a schematic view of a crack in a positioning frame provided by an embodiment of the invention;
fig. 5 is a schematic diagram illustrating a pixel-level classification network according to an embodiment of the present invention for identifying a slope disease; FIG. 6 is a schematic illustration of a calculated fracture length provided by an embodiment of the present invention;
FIG. 7 is a schematic diagram comparing the effect of the disease identification scheme provided by the embodiment of the present invention with that of a neural network identified by target classification alone, and a pixel-level classification neural network;
FIG. 8 is a schematic diagram of a slope disease device based on image acquisition provided by an embodiment of the present invention;
fig. 9 is a schematic view of a slope disease device based on image acquisition according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments that can be derived by one of ordinary skill in the art from the embodiments given herein are intended to be within the scope of the present invention.
Example one
Referring to fig. 1, in order to solve the problem of rapidly detecting an apparent disease of a slope and obtaining quantitative information of the area, length, width, and the like of the disease, the embodiment provides a method for obtaining a disease of the slope based on an image, which includes the following steps:
s1, acquiring a slope image;
the traditional method for acquiring the information of the surface of the side slope is to use a fixed camera to shoot images near the side slope, and the method is limited by the measuring environment to a great extent. Therefore this embodiment uses the unmanned aerial vehicle who carries the high resolution camera to gather the surface information of side slope.
Referring to fig. 2, the present embodiment uses DJI drones to acquire the slope image. And (3) carrying out path planning in advance for the unmanned aerial vehicle by using the DJI GS Pro ground station professional version of the IPAD end, and setting a plurality of acquisition points as flight waypoints in a mode of passing through intelligent waypoint flight. Meanwhile, continuous waypoint actions are set for each waypoint, including adjusting the yaw angle of the aircraft, adjusting the pitch angle of the holder, starting or stopping video recording and the like, so that more accurate measurement images are collected.
The circled points in fig. 2 are positions that need to be reached by the drone when the flight path is manually planned. When a plurality of circle points appear on the planning flight path diagram, the unmanned aerial vehicle sequentially executes linear flight tasks according to the point circle sequence, and performs video or image shooting on related objects through a preset angle.
S2, identifying the slope image by using a preset target positioning neural network to obtain a first identification area, wherein the first identification area is a disease area;
the diseases forming the side slope are different, and the influence on the safety and stability of the side slope is different. When slope diseases (such as vegetation exposure) with more image pixels are detected, a pixel-level classification neural network can be directly used for identification. However, because the unmanned aerial vehicle still has a certain height from the slope surface when shooting, the contour features of some diseases (small diseases, such as cracks) with few pixel ratios in the image are not easy to learn by the neural network of pixel classification. Therefore, if the neural network model of the pixel level classification is directly used to identify the small diseases, the accuracy of the network model will be greatly reduced.
In order to improve the identification accuracy of the artificial intelligence method, the embodiment proposes to use a target positioning neural network model for the input slope disease image. Referring to fig. 3, in the embodiment, the slope disease is first located by the target location identification neural network, and the target location identification neural network outputs a location frame to calibrate the disease position, and records the position of the image where the location frame is located. Referring to fig. 3, the left image in fig. 3 is an input image, and the right image in fig. 3 is an output image, where the output image has a positioning frame, and the positioning frame marks a damaged portion of a slope, such as a crack in the positioning frame in the right image in fig. 4. Locating the pixel p within the frame with reference to FIG. 3bMuch smaller than the pixel p of the whole image, the small defective pixel pbIs in a ratio of r' ═ pd/pbGreater than r ═ pdAnd/p, the neural network model is easier to learn and recognize by the pixel-level classification neural network model.
The present embodiment identifies a neural network using existing object classes such as the series of RCNN convolutional neural networks, the series of YOLO convolutional neural networks, the series of SSD, and the like.
S3, identifying the disease information in the first identification area by using a preset pixel-level classification neural network to obtain a second identification result;
the present embodiment uses existing pixel-level classification neural networks such as U-Net, FCN, deep Lab v3 plus.
Referring to fig. 5, the present embodiment uses a pixel-level classification neural network to identify the disease information from the first identification region output in step S2, and outputs a second identification result, where the second identification result outputs a binary image information in which the disease region is represented by "1".
The input image of the pixel-level classification neural network is the first region output in step S2, where the first region may be the first identification region obtained from the location box information output in step S2, and the first identification region is obtained from the slope image according to the location box information.
And S4, calculating the disease number, area and length information of the slope disease according to the first identification area and the second identification result.
The number of the slope apparent diseases can be determined by counting the number of the prediction positioning boxes of the target recognition neural network. The area of the disease, in this embodiment, the area of the disease refers to the area of the pixel, and this embodiment is calculated by MATLAB function "regionprops".
Referring to FIG. 6, the approximate length d of the crack in the region is obtained by taking x (the size of x can be determined according to the situation) pixels as a segment in the vertical direction of the image, wherein the smaller x is generally the closer the calculated approximate length is to the real length, but the longer the required calculation time is, and the invention integrates the precision and time considerations and takes x as 20 into consideration, and using the measurement type of "MajorAxisLength" in the MATLAB function "regionprops" (which is based on the principle of calculating the length of the major axis of the ellipse with the same normalized second-order center distance as the region, and the length of the major axis is approximately equal to the length of the crack in the region), the approximate length d of the crack in the region is obtainedi. And then summing the lengths obtained from each segment to obtain the approximate length of the whole crack in the image
Where n is the height of the image divided by x. Since the width and length of each pixel are 1 unit. The (pixel) area of the crack is thus directly obtained by counting the sum of pixels of the pixel-level classification results. And dividing the area of each segment by the length of each segment to obtain the width of each segment.
In the embodiment, the disease position information in the slope image is acquired by using the target positioning neural network, and the slope disease information in the slope image is identified by using the pixel-level classification neural network according to the disease position. In the embodiment, the disease information of the slope is obtained by using the pixel-level recognition result and the target positioning neural network result. In addition, in the embodiment, the length of the disease and other information are calculated by using the length of the major axis of the ellipse with the same normalized second-order central moment as that of a part of the pixel area, so that the length and area information of the disease can be acquired more accurately.
Referring to fig. 7, a schematic diagram of a comparison between the target classification, the pixel-level classification and the method of the embodiment is shown, where (a) is an input image, (b) is a target classification image, (c) is a pixel-level image of a crack obtained by directly using a pixel-level classification network, and (d) is an acquired second recognition result of the embodiment, that is, an obtained pixel-level image of the crack. The target classification is to locate the position of the disease in the image, and the result of using the pixel-level classification alone may result in that the disease cannot be identified or the disease identification is incomplete. The method provided by the invention can more accurately identify the pixel position of the disease (the white part is the background, and the blue part is the disease), and further carry out quantitative analysis on the disease.
In order to push the corresponding disease information to the corresponding administrator, the embodiment further includes the following steps:
s5, acquiring the disease degree of the side slope according to the disease quantity, area and length information of the side slope diseases, and pushing corresponding alarm information according to the disease degree.
Specifically, in this embodiment, in order to remind a slope manager, after the disease quantity, area, and length information of a slope disease is identified, the disease degree of the slope is evaluated according to the disease quantity, area, and length information of the slope disease, and corresponding warning information is pushed according to the disease degree.
Specifically, the present embodiment is illustrated only with or without severity.
Specifically, a disease number threshold is set in this embodiment, and when the disease number is greater than the disease number threshold, the disease degree is considered to be serious. Otherwise, the disease degree is not considered to be serious.
Specifically, in this embodiment, a plurality of disease number thresholds may also be set, and different disease degrees are divided according to the plurality of disease number thresholds, for example, two disease thresholds are set, so that the disease degree can be divided into normal, general, and serious. Otherwise, the modes are analogized in turn, and the description of this embodiment is omitted.
Specifically, this embodiment may also set a disease area threshold, and when the disease area is greater than the disease area threshold, the disease degree is considered to be serious. Otherwise, the disease degree is not considered to be serious.
Specifically, this embodiment may also set a disease length threshold, and when the disease length is greater than the disease length threshold, the disease degree is considered to be serious. Otherwise, the disease degree is not considered to be serious.
The alarm information of the embodiment can be pushed to the manager in the form of a short message or an email, or the system can directly remind the manager in a highlight form or a sound form.
Example two
The embodiment discloses a method for acquiring a slope disease based on an image, which comprises the following steps:
s1, acquiring a slope image;
s10, judging whether the side slope image has small diseases or not; if the small diseases exist, executing step S2, and if the small diseases do not exist, executing step S30;
and for the condition that whether the small diseases exist in the slope image or not, the workers can preliminarily judge the diseases in the image. The small diseases in this embodiment may be defined in advance by the staff, and the staff may define that the disease of a certain area belongs to the small diseases, for example, the disease belongs to the small diseases with less than 100 pixels.
In this embodiment, whether a small disease exists may also be determined by using the target localization neural network, and if the number value output of the localization frame is zero after using the target localization neural network, it indicates that there is no "predefined small disease" in the image, otherwise, there is a small disease.
S2, identifying the slope image by using a preset target positioning neural network to obtain a first identification area, wherein the first identification area is a disease area;
s3, identifying the disease information in the first identification area by using a preset pixel-level classification neural network to obtain a second identification result;
s30, recognizing the disease information in the slope image by using a preset pixel-level classification neural network to obtain a second recognition result;
and S4, calculating the disease number, area and length information of the slope disease according to the first identification area and the second identification result.
The steps S1, S2, S3, and S4 in this embodiment are the same as those in the first embodiment, and are not repeated herein.
The implementation uses the target-locating neural network not only to obtain the position of the disease in the image. The more important reason is that the diseases in the image (small image) extracted according to the target positioning frame can be more easily identified by the pixel-level classification neural network. The defect that a pixel-level classification neural network is difficult to identify small objects in the image is overcome. However, for large diseases (such as vegetation exposure), the pixel-level classification neural network can be directly used for identification. The embodiment of the present invention uses the target localization neural network + the pixel-level classification neural network to obtain the correct pixel-level classification result of the whole image.
EXAMPLE III
Referring to fig. 8, the present embodiment discloses a slope disease device based on image acquisition, which includes the following units:
the side slope image acquisition unit is used for acquiring side slope images;
the traditional method for acquiring the information of the surface of the side slope is to use a fixed camera to shoot images near the side slope, and the method is limited by the measuring environment to a great extent. Therefore this embodiment uses the unmanned aerial vehicle who carries the high resolution camera to gather the surface information of side slope.
Referring to fig. 2, the present embodiment uses DJI drones to acquire the slope image. And (3) carrying out path planning in advance for the unmanned aerial vehicle by using the DJI GS Pro ground station professional version of the IPAD end, and setting a plurality of acquisition points as flight waypoints in a mode of passing through intelligent waypoint flight. Meanwhile, continuous waypoint actions are set for each waypoint, including adjusting the yaw angle of the aircraft, adjusting the pitch angle of the holder, starting or stopping video recording and the like, so that more accurate measurement images are collected.
The first identification area acquisition unit is used for identifying a slope image by using a preset target positioning neural network to acquire a first identification area, wherein the first identification area is a disease area;
the diseases forming the side slope are different, and the influence on the safety and stability of the side slope is different. When slope diseases (such as vegetation exposure) with more image pixels are detected, a pixel-level classification neural network can be directly used for identification. However, because the unmanned aerial vehicle still has a certain height from the slope surface when shooting, the contour features of some diseases (small diseases, such as cracks) with few pixel ratios in the image are not easy to learn by the neural network of pixel classification. Therefore, if the neural network model of the pixel level classification is directly used to identify the small diseases, the accuracy of the network model will be greatly reduced.
In order to improve the identification accuracy of the artificial intelligence method, the embodiment proposes to use a target positioning neural network model for the input slope disease image. Referring to fig. 3, in the embodiment, the slope disease is first located by the target location identification neural network, and the target location identification neural network outputs a location frame to calibrate the disease position, and records the position of the image where the location frame is located. Referring to fig. 3, the left image in fig. 3 is an input image, and the right image in fig. 3 is an output image, where the output image has a positioning frame, and the positioning frame marks a damaged portion of a slope, such as a crack in the positioning frame in the right image in fig. 4. Locating the pixel p within the frame with reference to FIG. 3bMuch smaller than the pixel p of the whole image, the small defective pixel pbIs in a ratio of r' ═ pd/pbGreater than r ═ pdAnd/p, the neural network model is easier to learn and recognize by the pixel-level classification neural network model.
The second identification result acquisition unit is used for identifying the disease information in the first identification area by using a preset pixel-level classification neural network to acquire a second identification result;
referring to fig. 5, the present embodiment uses a pixel-level classification neural network to identify the disease information for the first identification region output by the first identification region acquisition unit, and outputs a second identification result, where the second identification result outputs binary image information in which the disease region is represented by "1".
The input image of the pixel-level classification neural network is a first area output by a first identification area acquisition unit, wherein the first area can be the positioning frame information output by the first identification area acquisition unit, and the first identification area is acquired in the slope image according to the positioning frame information.
And the slope disease calculation unit is used for calculating the disease number, area and length information of the slope disease according to the first identification area and the second identification result.
The number of the slope apparent diseases can be determined by counting the number of the prediction positioning boxes of the target recognition neural network. The area of the disease, in this embodiment, the area of the disease refers to the area of the pixel, and this embodiment is calculated by MATLAB function "regionprops".
Referring to FIG. 6, the approximate length d of the crack in the region is obtained by taking x (the size of x can be determined according to the situation) pixels as a segment in the vertical direction of the image, wherein the smaller x is generally the closer the calculated approximate length is to the real length, but the longer the required calculation time is, and the invention integrates the precision and time considerations and takes x as 20 into consideration, and using the measurement type of "MajorAxisLength" in the MATLAB function "regionprops" (which is based on the principle of calculating the length of the major axis of the ellipse with the same normalized second-order center distance as the region, and the length of the major axis is approximately equal to the length of the crack in the region), the approximate length d of the crack in the region is obtainedi. And then summing the lengths obtained from each segment to obtain the approximate length of the whole crack in the image
Where n is the height of the image divided by x. Since the width and length of each pixel are 1 unit. The (pixel) area of the crack is thus directly obtained by counting the sum of pixels of the pixel-level classification results. And dividing the area of each segment by the length of each segment to obtain the width of each segment.
In the embodiment, the disease position information in the slope image is acquired by using the target positioning neural network, and the slope disease information in the slope image is identified by using the pixel-level classification neural network according to the disease position. In the embodiment, the disease information of the slope is obtained by using the pixel-level recognition result and the target positioning neural network result. In addition, in the embodiment, the length of the disease and other information are calculated by using the length of the major axis of the ellipse with the same normalized second-order central moment as that of a part of the pixel area, so that the length and area information of the disease can be acquired more accurately.
In order to push the corresponding disease information to the corresponding administrator, the embodiment further includes the following units:
and the alarm information pushing unit is used for acquiring the disease degree of the side slope according to the disease quantity, area and length information of the side slope diseases and pushing corresponding alarm information according to the disease degree.
Specifically, in this embodiment, in order to remind a slope manager, after the disease quantity, area, and length information of a slope disease is identified, the disease degree of the slope is evaluated according to the disease quantity, area, and length information of the slope disease, and corresponding warning information is pushed according to the disease degree.
Specifically, the present embodiment is illustrated only with or without severity.
Specifically, a disease number threshold is set in this embodiment, and when the disease number is greater than the disease number threshold, the disease degree is considered to be serious. Otherwise, the disease degree is not considered to be serious.
Specifically, in this embodiment, a plurality of disease number thresholds may also be set, and different disease degrees are divided according to the plurality of disease number thresholds, for example, two disease thresholds are set, so that the disease degree can be divided into normal, general, and serious. Otherwise, the modes are analogized in turn, and the description of this embodiment is omitted.
Specifically, this embodiment may also set a disease area threshold, and when the disease area is greater than the disease area threshold, the disease degree is considered to be serious. Otherwise, the disease degree is not considered to be serious.
Specifically, this embodiment may also set a disease length threshold, and when the disease length is greater than the disease length threshold, the disease degree is considered to be serious. Otherwise, the disease degree is not considered to be serious.
The alarm information of the embodiment can be pushed to the manager in the form of a short message or an email, or the system can directly remind the manager in a highlight form or a sound form.
EXAMPLE III
Referring to fig. 9, fig. 9 is a schematic structural diagram of a slope disease device based on image acquisition according to the embodiment. The image acquisition-based slope disease device 20 of this embodiment includes a processor 21, a memory 22, and a computer program stored in the memory 22 and executable on the processor 21. The processor 21 realizes the steps in the above-described method embodiments when executing the computer program. Alternatively, the processor 21 implements the functions of the modules/units in the above-described device embodiments when executing the computer program.
Illustratively, the computer program may be divided into one or more modules/units, which are stored in the memory 22 and executed by the processor 21 to accomplish the present invention. The one or more modules/units may be a series of computer program instruction segments capable of performing specific functions, which are used to describe the execution process of the computer program in the image-based acquisition highwall disease device 20. For example, the computer program may be divided into the modules in the second embodiment, and for the specific functions of the modules, reference is made to the working process of the apparatus in the foregoing embodiment, which is not described herein again.
The image-based slope disease acquisition device 20 may include, but is not limited to, a processor 21 and a memory 22. Those skilled in the art will appreciate that the schematic diagram is merely an example of the image-based acquisition slope disease device 20, and does not constitute a limitation of the image-based acquisition slope disease device 20, and may include more or less components than those shown in the drawings, or combine some components, or different components, for example, the image-based acquisition slope disease device 20 may further include an input/output device, a network access device, a bus, and the like.
The Processor 21 may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic device, discrete hardware component, etc. The general processor may be a microprocessor or the processor may be any conventional processor, and the processor 21 is a control center of the image-based acquisition slope disease device 20, and various interfaces and lines are used to connect various parts of the entire image-based acquisition slope disease device 20.
The memory 22 may be configured to store the computer program and/or module, and the processor 21 implements various functions of the image-based acquisition slope disease device 20 by running or executing the computer program and/or module stored in the memory 22 and calling data stored in the memory 22. The memory 22 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required by at least one function (such as a sound playing function, an image playing function, etc.), and the like; the storage data area may store data (such as audio data, a phonebook, etc.) created according to the use of the cellular phone, and the like. In addition, the memory 22 may include high speed random access memory, and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), at least one magnetic disk storage device, a Flash memory device, or other volatile solid state storage device.
The module/unit integrated with the image-based slope disease equipment 20 may be stored in a computer-readable storage medium if it is implemented in the form of a software functional unit and sold or used as an independent product. Based on such understanding, all or part of the flow of the method according to the above embodiments may be implemented by a computer program, which may be stored in a computer readable storage medium and used by the processor 21 to implement the steps of the above embodiments of the method. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, usb disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-Only Memory (ROM), Random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution medium, and the like. It should be noted that the computer readable medium may contain content that is subject to appropriate increase or decrease as required by legislation and patent practice in jurisdictions, for example, in some jurisdictions, computer readable media does not include electrical carrier signals and telecommunications signals as is required by legislation and patent practice.
It should be noted that the above-described device embodiments are merely illustrative, where the units described as separate parts may or may not be physically separate, and the parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. In addition, in the drawings of the embodiment of the apparatus provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, and may be specifically implemented as one or more communication buses or signal lines. One of ordinary skill in the art can understand and implement it without inventive effort.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (10)
1. A method for acquiring a slope disease based on an image comprises the following steps:
s1, acquiring a slope image;
s2, identifying the slope image by using a preset target positioning neural network to obtain a first identification area, wherein the first identification area is a disease area;
s3, identifying the disease information in the first identification area by using a preset pixel-level classification neural network to obtain a second identification result;
and S4, calculating the disease number, area and length information of the slope disease according to the first identification area and the second identification result.
2. The method of claim 1, wherein the identifying the slope image using the preset target location neural network outputs a location box for identifying a location of the first identification area on the slope image.
3. The method of claim 2, wherein the number of slope apparent diseases is determined by the number of predicted location boxes of a statistical target recognition neural network.
4. The method according to claim 3, wherein the length information of the slope disease is obtained as follows:
dividing the slope diseases into n sections, and enabling the pixels in partial areas to have the same normalization second orderThe method for summing the long axes of n ellipses at the center distance obtains the long axis d of the ellipse of each sectioniAnd calculating and identifying the length d of the disease according to the major axis of the ellipse of each section,
5. the method of claim 1, further comprising the steps of: s5, acquiring the disease degree of the side slope according to the disease quantity, area and length information of the side slope diseases, and pushing corresponding alarm information according to the disease degree.
6. A slope disease device based on image acquisition comprises the following units:
the side slope image acquisition unit is used for acquiring side slope images;
the first identification area acquisition unit is used for identifying a slope image by using a preset target positioning neural network to acquire a first identification area, wherein the first identification area is a disease area;
the second identification result acquisition unit is used for identifying the disease information in the first identification area by using a preset pixel-level classification neural network to acquire a second identification result;
and the slope disease calculation unit is used for calculating the disease number, area and length information of the slope disease according to the first identification area and the second identification result.
7. The apparatus of claim 6, the identifying a slope image using a preset target location neural network outputting a location box, the location box identifying a location of the first identified region in the slope image.
8. The apparatus of claim 7, the number of slope apparent diseases is determined by the number of predicted location boxes of a statistical target recognition neural network.
9. The apparatus according to claim 8, wherein the length information of the slope disease is obtained as follows:
dividing the slope diseases into n sections, and acquiring the ellipse major axis d of each section by a method of summing n ellipse major axes with pixels in partial areas having the same normalized second-order center distanceiAnd calculating and identifying the length d of the disease according to the major axis of the ellipse of each section,
10. an electronic device comprising a processor, and a memory having stored thereon instructions that, when executed by the processor, are to implement the method of any of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111357743.0A CN114169404A (en) | 2021-11-16 | 2021-11-16 | Method for intelligently acquiring quantitative information of slope diseases based on images |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111357743.0A CN114169404A (en) | 2021-11-16 | 2021-11-16 | Method for intelligently acquiring quantitative information of slope diseases based on images |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114169404A true CN114169404A (en) | 2022-03-11 |
Family
ID=80479301
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111357743.0A Pending CN114169404A (en) | 2021-11-16 | 2021-11-16 | Method for intelligently acquiring quantitative information of slope diseases based on images |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114169404A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114841995A (en) * | 2022-06-02 | 2022-08-02 | 西南交通大学 | Deep learning-based railway roadbed fender equipment service state evaluation method |
-
2021
- 2021-11-16 CN CN202111357743.0A patent/CN114169404A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114841995A (en) * | 2022-06-02 | 2022-08-02 | 西南交通大学 | Deep learning-based railway roadbed fender equipment service state evaluation method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ukhwah et al. | Asphalt pavement pothole detection using deep learning method based on YOLO neural network | |
| WO2022083402A1 (en) | Obstacle detection method and apparatus, computer device, and storage medium | |
| CN109087510B (en) | Traffic monitoring method and device | |
| CN110084165B (en) | Intelligent identification and early warning method for abnormal events in open scene of power field based on edge calculation | |
| US10853943B2 (en) | Counting objects in images based on approximate locations | |
| US20210192227A1 (en) | Method and apparatus for detecting parking space usage condition, electronic device, and storage medium | |
| CN113822247B (en) | Method and system for identifying illegal building based on aerial image | |
| CN113537049B (en) | Ground point cloud data processing method and device, terminal equipment and storage medium | |
| CN114565895B (en) | Security monitoring system and method based on intelligent society | |
| CN112184773A (en) | Helmet wearing detection method and system based on deep learning | |
| CN109684986B (en) | Vehicle analysis method and system based on vehicle detection and tracking | |
| CN108229473A (en) | Vehicle annual inspection label detection method and device | |
| WO2022126522A1 (en) | Object recognition method, apparatus, movable platform, and storage medium | |
| CN113255444A (en) | Training method of image recognition model, image recognition method and device | |
| CN114169404A (en) | Method for intelligently acquiring quantitative information of slope diseases based on images | |
| CN110909674A (en) | Traffic sign identification method, device, equipment and storage medium | |
| CN112784494B (en) | Training method of false positive recognition model, target recognition method and device | |
| CN112529836A (en) | High-voltage line defect detection method and device, storage medium and electronic equipment | |
| US20230314169A1 (en) | Method and apparatus for generating map data, and non-transitory computer-readable storage medium | |
| CN115690747A (en) | Vehicle blind area detection model test method and device, electronic equipment and storage medium | |
| CN112686162B (en) | Method, device, equipment and storage medium for detecting clean state of warehouse environment | |
| CN114973326A (en) | Fall early warning method, device, equipment and readable storage medium | |
| CN114445326A (en) | Photovoltaic panel abnormity detection method, detection device and computer readable storage device | |
| CN114005041A (en) | Road disease identification control method and equipment based on UAVRS and BIM | |
| CN113569954A (en) | Intelligent wild animal classification and identification method |
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 |