CN111861982A - Visual image monitoring and identifying system - Google Patents
Visual image monitoring and identifying system Download PDFInfo
- Publication number
- CN111861982A CN111861982A CN202010509305.0A CN202010509305A CN111861982A CN 111861982 A CN111861982 A CN 111861982A CN 202010509305 A CN202010509305 A CN 202010509305A CN 111861982 A CN111861982 A CN 111861982A
- Authority
- CN
- China
- Prior art keywords
- charging
- image
- monitoring
- aerial vehicle
- unmanned aerial
- 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
- 238000012544 monitoring process Methods 0.000 title claims abstract description 90
- 230000000007 visual effect Effects 0.000 title claims abstract description 19
- 239000011435 rock Substances 0.000 claims abstract description 108
- 238000001514 detection method Methods 0.000 claims description 30
- 230000008859 change Effects 0.000 claims description 26
- 238000006073 displacement reaction Methods 0.000 claims description 24
- 238000000034 method Methods 0.000 claims description 11
- 238000000605 extraction Methods 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 7
- 238000003702 image correction Methods 0.000 claims description 5
- 230000009471 action Effects 0.000 claims description 3
- 230000000630 rising effect Effects 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 abstract description 5
- 238000010586 diagram Methods 0.000 description 6
- 238000013527 convolutional neural network Methods 0.000 description 5
- 239000004020 conductor Substances 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 230000014509 gene expression Effects 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 239000002689 soil Substances 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 230000000994 depressogenic effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
-
- G06T5/70—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
- G06T2207/30184—Infrastructure
Abstract
The invention relates to the field of dangerous rock monitoring by utilizing an image recognition technology, and provides a visual image monitoring and recognizing system.
Description
Technical Field
The invention relates to the field of dangerous rock monitoring by utilizing an image recognition technology, in particular to a visual image monitoring and recognizing system.
Background
China is vast in breadth, two thirds of China's soil belongs to mountain landforms, and is one of the most serious countries in the world suffering from mountain collapse disasters, and dangerous rock collapse can cause great loss to the property of people and threaten personal safety. At present, dangerous rock monitoring is mainly adopted as means for treating dangerous rocks, cracks with different degrees can be generated on the surface of the dangerous rocks before collapse, manual monitoring modes are adopted more, or the dangerous rocks are judged after being monitored by an unmanned aerial vehicle and laser, more manpower and material resources are consumed, and real-time monitoring on the dangerous rocks cannot be realized. With the development of image recognition technology, dangerous rocks can be recognized and monitored, and the method and the system for recognizing the rock mass fracture of the Chinese patent CN104700407B establish a rock mass model according to the acquired rock mass surface data, and cut the surface of the rock mass model into triangular units, so that fracture lines are determined according to the triangular units. However, the accuracy of the technology for identifying the fractures of the dangerous rock is not high, misjudgment is easy to occur, and the dangerous rock cannot be monitored in real time. Chinese patent CN109584240A "displacement image recognition method of trailing edge of landslide crack" converts an original color image into a single-channel gray image by using a graying method through the vegetation on the surface of a mountain and the color of exposed rock soil of the landslide, converts the image into a binarized image after noise reduction, and then recognizes the trailing edge of landslide crack through image corrosion and expansion.
Disclosure of Invention
The invention aims to solve the technical problem of providing a visual image monitoring and identifying system, which can solve the problems of low dangerous rock monitoring precision and incapability of real-time monitoring.
In order to solve the technical problems, the technical scheme adopted by the invention is as follows: visual image monitoring identification system carries out the dangerous rock monitoring through the image to acquireing, including image acquisition module, crack identification module, displacement identification module and alarm module, the step of realizing is:
s1, the image acquisition module acquires a dangerous rock monitoring image and demarcates a dangerous rock monitoring area;
s2, the crack recognition module performs feature recognition on the crack by using the trained crack recognition model;
s3, identifying dangerous rock displacement change by a displacement identification module;
s4, the alarm module gives an alarm when the change of the size of the crack and the displacement of the critical edge exceeds the threshold value.
In a preferred embodiment, step S1 is implemented as follows:
s1a, acquiring a monitoring image by shooting through an unmanned aerial vehicle or a fixed camera;
s1b, defining a dangerous rock monitoring area according to the datum points;
and S1c, calculating the inclination angle and the reference point distance between the current frame image and a preset image, and performing image correction.
In a preferred embodiment, step S2 is implemented as follows:
S2a, establishing a fracture recognition image training set and training a fracture recognition model;
s2b, acquiring the target frame in the image corrected in the step S1c, and selecting characteristic points to perform fracture identification;
s2c, calculating the length and the width of the crack.
In a preferred embodiment, step S3 is implemented as follows:
s3a, extracting the feature points of the target area corresponding to the two frames of images acquired in the step S1 c;
s3b, calculating the distance from the characteristic point to the datum point;
and S3c, judging whether the dangerous rock is subjected to displacement change.
In a preferred scheme, in step S1a, the non-attention points are excluded by the image features of different viewing angles acquired by the unmanned aerial vehicle.
Preferably, in step S2a, the image related to the attention point is subjected to feature extraction as a positive sample, and then the subsequent image is subjected to corresponding feature extraction as a negative sample.
In the preferred scheme, the feature extraction mode adopts a maximum inter-class variance method, and the specific implementation steps are as follows:
ST1, acquiring the rising and falling rules of the sun and the change of the shadow at the position of the monitoring place to obtain the optimal shooting time and angle of the monitoring image and ensure that the color difference of the crack and the bottom surface reaches the maximum;
ST2, acquiring a dangerous rock monitoring image by the unmanned aerial vehicle according to a preset track;
ST3, acquiring a dangerous rock monitoring image;
and ST4, identifying the dangerous rock fracture.
In a preferred embodiment, the specific implementation steps in step ST2 are:
ST21, setting one or more near-field charging devices on a path from a control site to the dangerous rock mass according to the secondary navigation distance of the unmanned aerial vehicle;
the unmanned aerial vehicle flies to a preset near field charging device according to the planned path and the GPS positioning device, identifies a charging platform of the near field charging device through a camera arranged at the bottom of the unmanned aerial vehicle, adjusts the position according to the identification result, lands on the charging platform and charges;
ST23, after charging, the unmanned aerial vehicle takes off to collect the image of the dangerous rock mass again, and after collecting, the unmanned aerial vehicle lands on the charging platform to charge again and then returns;
realize utilizing unmanned aerial vehicle to carry out remote monitoring to the dangerous rock through above step.
In the preferred scheme, one or more relay airships are also provided, and a GPS positioning device and a height instrument are arranged on each relay airship;
a near-field charging device is arranged on the relay airship;
each relay airship is arranged along the flight path of the unmanned aerial vehicle, and a relay is arranged on each relay airship and used for amplifying the control signal of the unmanned aerial vehicle so as to control the action of the unmanned aerial vehicle;
The working path for the dangerous rock mass is preset in the unmanned aerial vehicle, the working path is a flight path which takes a near-field charging device arranged near the dangerous rock mass as a starting point and an end point and takes the near-field charging device arranged near the dangerous rock mass as a target, and the working path is executed after the unmanned aerial vehicle reaches the near-field charging device arranged near the dangerous rock mass and finishes charging.
In a preferred scheme, a plurality of charging electrodes arranged in an array are arranged on the charging platform and are used for being electrically connected with a connecting electrode at the bottom of the unmanned aerial vehicle to realize charging;
elastic conductive contacts are arranged at the positions of the charging electrodes, so that the row lead and the column lead of the current charging electrode are communicated when the connecting electrode is pressed down, and the row lead and the column lead of the current charging electrode are separated when the connecting electrode is separated;
a plurality of row detection switches are arranged along one side of the charging electrodes arranged in an array, one ends of the row detection switches are electrically connected with row conducting wires of all rows, and the other ends of the row detection switches are electrically connected with detection signals;
a plurality of column detection switches are arranged along one side of the charging electrodes arranged in an array, one end of each column detection switch is electrically connected with a column lead of each column, and the other end of each column detection switch is electrically connected with a detection signal source;
A plurality of row charging switches are arranged along one side of the charging electrodes arranged in an array, one ends of the row charging switches are electrically connected with the row conducting wires of the rows, and the other ends of the row charging switches are electrically connected with a charging power supply;
and a plurality of column charging switches are arranged on one side of the charging electrodes arranged along the array, one ends of the column charging switches are electrically connected with column conducting wires of all the columns, and the other ends of the column charging switches are electrically connected with a charging power supply.
The invention provides a visual image monitoring and identifying system, which has the following beneficial effects by adopting the scheme:
1. the labor cost is greatly reduced. According to the invention, dangerous rocks are monitored in real time by using an image recognition technology, manpower is not needed for monitoring, and only processing is needed according to the sent alarm, so that manpower and material resources are greatly reduced, and social resources are saved.
2. The dangerous rock monitoring is accurate and timely. The collected image data volume is very large, the difficulty of manual real-time identification is high, but the unmanned aerial vehicle continuous navigation monitoring system can perform real-time monitoring according to a fixed camera or an image obtained by the unmanned aerial vehicle, and when the unmanned aerial vehicle is used for monitoring, the problem of the unmanned aerial vehicle continuous navigation can be solved by adopting the scheme of the arranged near-field charging device, the monitoring time is accurately selected, and the monitoring efficiency is high, accurate and timely.
3. The observation error is small. Compared with the traditional manual monitoring or laser monitoring, the method is more accurate, high in accuracy and small in error.
Drawings
The invention is further illustrated by the following examples in conjunction with the accompanying drawings:
FIG. 1 is a flow chart of feature image registration of the present invention.
Fig. 2 is a dangerous rock monitoring image in the invention.
Fig. 3 is a gray scale image of a dangerous rock monitoring image in the invention.
Fig. 4 is a schematic diagram of a dangerous rock monitoring image transformation result in the invention.
Fig. 5 is a schematic view of feature recognition of a monitored image in the present invention.
FIG. 6 is a schematic diagram of a datum point in the present invention.
FIG. 7 is a diagram illustrating the distance difference between a feature point and a reference point according to the present invention.
Fig. 8 is a schematic view of the overall structure of the near-field arrangement near-field charging device according to the present invention.
Fig. 9 is a schematic structural diagram of a near-field charging device according to the present invention.
Fig. 10 is a schematic structural view of the unmanned aerial vehicle landing on the charging platform according to the present invention.
Fig. 11 is a schematic top view of the charging platform according to the present invention.
FIG. 12 is a schematic diagram of a control circuit according to the present invention.
Fig. 13 is a schematic structural view of the charging electrode in the present invention.
In the figure: an unmanned aerial vehicle 1; connecting the electrodes 11; a near-field charging device 2; a solar cell panel 21; a charging platform 22; a charging electrode 221; a column detection switch 222; a row detection switch 223; column charge switch 224; row charge switch 225; a first interlock relay switch 226; column conductor 227; row conductors 228; conductive contacts 229; a second interlock relay switch 230; an elastic bowl 220; a drain structure 23; a dangerous rock mass 3; rock mass fracture 4; a relay airship 5; a main control chip 6; and a relay 7.
Detailed Description
Example 1:
the embodiment performs fracture recognition on the image acquired by the fixed camera. From a mechanical mechanism, the dangerous rock instability mode is divided into falling type dangerous rock, slipping type dangerous rock and dumping type dangerous rock, as shown in figures 1-7, cracks of the dangerous rock are monitored, and the method is realized by the following steps:
s1, an image acquisition module acquires a dangerous rock monitoring image, a dangerous rock monitoring area is defined, a reference point is set in the dangerous rock monitoring area, the monitoring image is acquired by shooting through a fixed camera, the dangerous rock monitoring area is defined according to the reference point, the inclination angle and the reference point distance between a current frame image and a preset image are calculated, and image correction is carried out.
S2, the crack recognition module performs feature recognition on the crack by using the trained crack recognition model;
firstly, a fracture recognition model is trained, and a data set for model training is prepared, wherein the data set is an image of dangerous rocks needing to be monitored, which is shot by a fixed camera in an earlier stage and comprises training data and test data. And (5) making a label, and enabling the image to correspond to the label. And converting the image into a computer readable format, starting training, verifying the accuracy of the model, and exporting the model to identify a single dangerous rock image. As shown in fig. 5, the current frame image is input into the CNN, and Feature Map of the crisis rock fracture is obtained. And inputting the convolution characteristics into the RPN to obtain the characteristic information of the candidate frame, and adding the RPN when the characteristic frame is selected to generate a candidate area in order to improve the speed of the algorithm. And judging whether the fracture features extracted from the candidate frame belong to the fractures or not by using a classifier, comparing the extracted fracture features with the fracture features extracted from the trained image, and judging whether the fracture features belong to the suspected fractures or not according to a preset label in the classifier. And for the candidate frames belonging to the fracture, further adjusting the position of the fracture by using a regressor, wherein the system shares a convolution characteristic diagram, and the extraction of the candidate region and the CNN classification are combined into a unified network. And the RPN establishes a neural network separately for carrying out fracture feature classification and frame position regression, and compared with the CNN and the R-CNN, the fast R-CNN adds a layer of neural convolution network RPN in the selection of the feature frame, thereby greatly improving the monitoring rate.
And S3, the alarm module gives an alarm when the change of the fracture size exceeds a threshold value and a new dangerous rock fracture is generated.
Example 2:
the embodiment identifies the crack of the image acquired by the unmanned aerial vehicle.
S1a, shooting through an unmanned aerial vehicle to obtain a monitoring image, wherein the image obtained by the unmanned aerial vehicle has the problems of large target rotation change, large angle change, large scene light and shade change and the like, and the flight path of the image obtained by the unmanned aerial vehicle has at least 60% course overlapping and 40% lateral overlapping, so that the posture and the track of the unmanned aerial vehicle are required to be planned, and the unmanned aerial vehicle is ensured to run the same path to obtain images at the same position and the same inclination angle. When obtaining the monitoring image for the first time, obtain unmanned aerial vehicle and shoot position and angle parameter, when obtaining the monitoring image through unmanned aerial vehicle again, set up deflection value and counterforce, when unmanned aerial vehicle skew target track is more, increase the counterforce and control unmanned aerial vehicle toward the target track deflection, carry out the integral according to the time that deflection error and error experience, then apply again for a power of target direction, let unmanned aerial vehicle get back to on the target track, unmanned aerial vehicle acquires the dangerous rock monitoring image at the target point according to the target track.
S1b, a dangerous rock monitoring area is defined according to the reference points, as shown in fig. 7, the reference points are selected, the reference points cannot change along with the change of dangerous rocks, and the reference points do not exceed the maximum visual angle range of the images acquired by the unmanned aerial vehicle.
And S1c, calculating the inclination angle and the reference point distance between the current frame image and a preset image, and performing image correction. And comparing the inclination angle and the reference point distance of the current frame of the dangerous rock monitoring image acquired in the step S1a with the preset image to realize the rotation and the equal-scale reduction of the monitoring image for correction.
And S2, carrying out crack identification on the corrected image, searching an extreme point in the monitored image, extracting the position, the scale and the rotation invariant of a reference point, taking the position, the scale and the rotation invariant as a characteristic point, and generating a characteristic vector by utilizing the neighborhood of the characteristic point. As shown in fig. 5, the crack is identified, the red circle represents the characteristic point of the crack matching success, the size of the circle represents the matching degree, the larger the radius of the red circle is, the higher the matching degree is, and the green square represents the identified dangerous rock crack.
Example 3:
this embodiment carries out displacement monitoring to the image that fixed camera obtained.
S1, an image acquisition module acquires a dangerous rock monitoring image, a dangerous rock monitoring area is defined, a reference point is set in the dangerous rock monitoring area, the monitoring image is acquired by shooting through a fixed camera, the dangerous rock monitoring area is defined according to the reference point, the inclination angle and the reference point distance between a current frame image and a preset image are calculated, and image correction is carried out.
S2, selecting a current frame image of the monitoring video acquired by the fixed camera, selecting fracture characteristic points on the current frame image, calculating to obtain a spatial displacement vector of a dangerous rock fracture expansion area, and comparing expansion characteristics and displacement relational expressions corresponding to various possible deformation damage modes, so that the actual dangerous rock fracture expansion amount, expansion area and dangerous rock displacement change are identified.
And S3, calculating the distance between the crack characteristic points and the reference points, and calculating the displacement change of the crack and the displacement change of the dangerous rock. As shown in fig. 7, the distance difference (Δ x, Δ y) from the fracture characteristic point (x 1, y 1) to the reference point (x 2, y 2) is calculated, and compared with the distance difference from the image fracture characteristic point obtained for the first time to the reference point, the displacement change of the fracture is calculated, and similarly, the change of the dangerous rock is calculated.
Example 4: displacement monitoring is carried out on images acquired by unmanned aerial vehicle
S3a, extracting two frames of images acquired by the unmanned aerial vehicle to correspond to the feature points of the target area, and removing the non-attention points through the image features of different visual angles acquired by the unmanned aerial vehicle. The image relating to the point of interest is feature extracted as a positive sample, and then the subsequent images are correspondingly feature extracted as negative samples.
And S3b, calculating the distance from the characteristic points to the reference point, selecting fracture characteristic points on the current frame image, calculating to obtain a space displacement vector of a dangerous rock fracture expansion area, and comparing expansion characteristics and displacement relational expressions corresponding to various possible deformation damage modes, so that the actual dangerous rock fracture expansion amount, expansion area and displacement change of dangerous rock are identified.
And S3c, judging whether the crack is expanded or not and whether the dangerous rock is subjected to displacement change or not. And calculating the distance difference (delta x and delta y) from the fracture characteristic points (x 1 and y 1) to the reference points (x 2 and y 2), comparing the distance difference (delta x and delta y) with the distance difference from the fracture characteristic points of the image acquired for the first time to the reference points, calculating the displacement change of the fracture, and calculating the change of the dangerous rock in the same way.
Example 5:
in this example, the maximum inter-class variance method is used for fracture identification.
The feature extraction method can adopt a maximum inter-class variance method. When the unmanned aerial vehicle acquires an image, the color difference between the crack and the ground is not large due to the light or the weather, and the optimal shooting position is obtained by acquiring the illumination information and the weather information of the monitoring place. The method comprises the following steps:
ST 1: acquiring the rising and falling rules of the sun and the change of the shadow at the monitoring place to obtain the optimal shooting time and angle of the monitoring image and ensure that the color difference of the crack and the bottom surface is maximum.
ST 2: unmanned aerial vehicle acquires dangerous rock monitoring image according to predetermined orbit, obtain unmanned aerial vehicle when obtaining monitoring image for the first time and shoot position and angle parameter, when acquireing monitoring image through unmanned aerial vehicle once more, set up deflection value and counterforce, when unmanned aerial vehicle skew target track is more, increase the control unmanned aerial vehicle of counterforce and deflect toward the target track, carry out the integral according to the time that deflection error and error experience, then apply a power for the target direction again, let unmanned aerial vehicle get back to on the target track, unmanned aerial vehicle acquires dangerous rock monitoring image at the target point according to the target track.
ST 3: dividing the obtained dangerous rock monitoring image into a foreground image and a background image, recording t as a segmentation threshold value of the foreground and the background, wherein the ratio of the number of foreground points to the image is w0, and the average gray level is u 0; the number of background points is w1 in the image scale, and the average gray scale is u 1. The total average gray scale of the image is: u = w0 u0+ w1 u1, variance of foreground and background images: g = w0 (u0-u) (u0-u) + w1 (u1-u) (u1-u) = w 0) (w 1 (u0-u1) (u0-u1), this formula is a variance formula, and when the variance g is maximum, it can be considered that the difference between the foreground and the background at this time is maximum, that is, the gray level at this time is the optimum threshold.
ST 4: and identifying the fracture of the dangerous rock, carrying out median filtering on the gray level images obtained in the three steps to remove discrete noise in the images, then filtering out low-frequency information by using a filter to enhance the edge of the fracture, obtaining a binary image according to the optimal threshold segmentation obtained in the three steps, comparing the binary result with the binary result of the gray level image obtained for the first time to extract the information of the fracture, and judging the change of the fracture.
Example 6:
this embodiment is the course of operation of unmanned aerial vehicle near field continuation of the journey, ensures that the colour difference of crack and bottom surface reaches the biggest, need carry out the accuracy to the time of acquireing the monitoring image and control, when utilizing unmanned aerial vehicle to acquire the image, near field continuation of the journey system can ensure that unmanned aerial vehicle takes off and acquires the monitoring image according to the time of needs, as shown in fig. 8~13, unmanned aerial vehicle near field continuation of the journey and the step of acquiring the monitoring image as follows:
ST21, one or more near-field charging devices 2 are arranged on a path from a control site to the dangerous rock mass 3 according to the secondary navigation distance of the unmanned aerial vehicle 1; be equipped with the charging platform 22 that supplies unmanned aerial vehicle 1 to descend and charge on the near field charging device 2, charging platform 22 arrange with the angle slope of 1~19, one side of charging platform 22 still is equipped with solar cell panel 21, solar cell panel 21 and battery electricity are connected. A drainage structure 23 is provided at the periphery of the charging platform 22. From this structure, can avoid charging platform 22 ponding under the prerequisite that does not influence unmanned aerial vehicle 1 to descend, improve near field charging device 2's practicality. Be equipped with camera or camera on the unmanned aerial vehicle 1 for gather dangerous rock mass 3, especially the image of rock mass crack 4 on the dangerous rock mass 3, through the data acquisition of a plurality of time quantum, compare the difference on the image of different time quantum, thereby monitor dangerous rock mass 3 to and the change of rock mass crack 4.
The method comprises the following steps that ST22, the unmanned aerial vehicle 1 flies to a preset near-field charging device 2 according to a planned path and a GPS positioning device, a camera arranged at the bottom of the unmanned aerial vehicle 1 is used for identifying a charging platform 22 of the near-field charging device 2, the position of the charging platform is adjusted according to an identification result, and the charging platform 22 is landed for charging;
as shown in fig. 11, the bottom of the unmanned aerial vehicle 1 is provided with at least two groups of connecting electrodes 11, which are respectively a positive electrode and a negative electrode; the connecting electrode 11 is arranged at the bottom of the landing gear of the unmanned aerial vehicle 1. The connection electrodes 11 are provided so that each of the two sets of connection electrodes 11 can cover at least 2 charging electrodes 221. At this time, the connection electrode 11 presses down the charge electrode 221, and the column wire 227 and the row wire 228 of the pressed charge electrode 221 are electrically connected.
As shown in fig. 11 to 13, a plurality of charging electrodes 221 arranged in an array are disposed on the charging platform 22, and are used for electrically connecting with the connecting electrode 11 at the bottom of the unmanned aerial vehicle 1 to realize charging, an elastic conductive contact 229 is disposed at each position of the charging electrode 221, so that when the connecting electrode 11 is pressed down, the row lead 228 and the column lead 227 of the current charging electrode 221 are communicated, and when the connecting electrode 11 is separated, the row lead 228 and the column lead 227 of the current charging electrode 221 are separated, a plurality of row detection switches 223 are disposed along one side of the charging electrode 221 arranged in an array, one end of each row detection switch 223 is electrically connected with the row lead 228 of each row, the other end of each row detection switch 223 is electrically connected with a detection signal, a plurality of column detection switches 222 are disposed along one side of the charging electrode 221 arranged in an array, one end of each column detection switch 222 is electrically connected with the column lead 227 of each column, and the other end of each column, a plurality of row charging switches 225 are disposed along one side of the charging electrodes 221 arranged in an array, one end of each row charging switch 225 is electrically connected to a row lead 228 of each row, the other end of each row charging switch 225 is electrically connected to a charging power source, a plurality of column charging switches 224 are disposed along one side of the charging electrodes 221 arranged in an array, one end of each column charging switch 224 is electrically connected to a column lead 227 of each column, and the other end of each column charging switch 224 is electrically connected to a charging power source.
During charging, the relay 7 is powered on, the second interlock relay switch 230 is turned on, and the first interlock relay switch 226 is turned off; the main control chip 6 controls the column detection switch 222 and the row detection switch 223 to be turned on in a scanning manner, and detects the depressed charging electrode 221. The main control chip 6 may optionally adopt an STM32F series main control chip. The column detection switch 222 and the row detection switch 223 may optionally adopt a thyristor, a relay or a current control type pulse width modulator UC3842 chip. After the detection is completed, the charging electrodes 221 that are short-circuited with each other are grouped, and the charging electrodes 221 can be quickly grouped by adopting a scheme that the profiles of the charging electrodes 221 to be pressed are fitted, and then the fitted profiles are overlapped with the profile of the connecting electrode 11. When the relay 7 is powered off, the second interlock relay switch 230 is turned off, the first interlock relay switch 226 is turned on, and different rows or columns of the grouped charging electrodes 221 are selected to supply power. Taking fig. 11 as an example, the angle between the connection electrode 11 and the horizontal line is determined first, if the angle is smaller than 45 °, different rows are selected for power supply, and if the angle is larger than 45 °, different columns are selected for power supply, for example, in fig. 11, the charging switch 225 for the row 3 and the row 4 from top to bottom is turned on, and the row detection switch 223 for the row 5 and the row 6 from top to bottom is turned on, so that the unmanned aerial vehicle 1 can be charged. By adopting the scheme, the difficulty of remote control or automatic charging control is greatly reduced, and the charging success rate is improved.
The charging power supply in this example is preferably 36V, and is provided by the solar panel 21 on the charging platform 22, and the electric energy collected by the solar panel 21 is stored in the storage battery.
ST23, after the charging, the unmanned aerial vehicle 1 takes off to collect the image of the dangerous rock mass 3 again, and after the collection is finished, the unmanned aerial vehicle lands on the charging platform 22 to charge again and then returns;
realize utilizing unmanned aerial vehicle to carry out remote monitoring to the dangerous rock through above step.
In a preferred scheme, as shown in fig. 8, one or more relay airships 5 are further provided, and a GPS positioning device and a height instrument are arranged on the relay airships 5;
the relay airship 5 is provided with a near-field charging device 2;
each relay airship 5 is arranged along the flight path of the unmanned aerial vehicle, and a relay is arranged on each relay airship 5 and is used for amplifying the control signal of the unmanned aerial vehicle 1 so as to control the action of the unmanned aerial vehicle 1;
the working path to dangerous rock mass 3 is preset in the unmanned aerial vehicle 1, the working path is a flight path which takes the near-field charging device 2 arranged near the dangerous rock mass 3 as a starting point and an end point and takes the near-field charging device 2 arranged near the dangerous rock mass 3 as a target to completely collect images around the dangerous rock mass 3, and the working path is executed after the unmanned aerial vehicle 1 reaches the near-field charging device 2 arranged near the dangerous rock mass 3 and completes charging. From this structure, when the signal is not smooth, perhaps when control distance is far away, unmanned aerial vehicle 1 also can accomplish the job task automatically. The unmanned aerial vehicle carries out image acquisition to dangerous rock mass 3. By adopting the scheme, the workload of the dangerous rock mass 3 and the rock mass fracture 4 is greatly improved, the labor intensity is reduced, the monitoring cost is reduced, and the safety is improved.
The above-described embodiments are merely preferred embodiments of the present invention, and should not be construed as limiting the present invention, and features in the embodiments and examples in the present application may be arbitrarily combined with each other without conflict. The protection scope of the present invention is defined by the claims, and includes equivalents of technical features of the claims. I.e., equivalent alterations and modifications within the scope hereof, are also intended to be within the scope of the invention.
Claims (10)
1. Visual image monitoring identification system, characterized by: through carrying out dangerous rock monitoring to the image that obtains, including image acquisition module, crack identification module, displacement identification module and alarm module, the step of realization is:
s1, the image acquisition module acquires a dangerous rock monitoring image and demarcates a dangerous rock monitoring area;
s2, the crack recognition module performs feature recognition on the crack by using the trained crack recognition model;
s3, identifying dangerous rock displacement change by a displacement identification module;
s4, the alarm module gives an alarm when the change of the size of the crack and the displacement of the critical edge exceeds the threshold value.
2. The visual image monitoring and recognition system according to claim 1, wherein: the step S1 is specifically realized by:
S1a, acquiring a monitoring image by shooting through an unmanned aerial vehicle or a fixed camera;
s1b, defining a dangerous rock monitoring area according to the datum points;
and S1c, calculating the inclination angle and the reference point distance between the current frame image and a preset image, and performing image correction.
3. The visual image monitoring and recognition system according to claim 2, wherein: the step S2 is specifically realized by:
s2a, establishing a fracture recognition image training set and training a fracture recognition model;
s2b, acquiring the target frame in the image corrected in the step S1c, and selecting characteristic points to perform fracture identification;
s2c, calculating the length and the width of the crack.
4. The visual image monitoring and recognition system according to claim 2, wherein: the step S3 is specifically realized by:
s3a, extracting the feature points of the target area corresponding to the two frames of images acquired in the step S1 c;
s3b, calculating the distance from the characteristic point to the datum point;
and S3c, judging whether the dangerous rock is subjected to displacement change.
5. The visual image monitoring and recognition system according to claim 2, wherein: in step S1a, the non-attention points are excluded by the image features of different viewing angles acquired by the unmanned aerial vehicle.
6. The visual image monitoring and recognition system according to claim 3, wherein: in step S2a, the image relating to the point of interest is subjected to feature extraction as a positive sample, and then the subsequent images are subjected to corresponding feature extraction as a negative sample.
7. The visual image monitoring and recognition system according to claim 2, wherein: the feature extraction mode adopts a maximum inter-class variance method, and the specific implementation steps are as follows:
ST1, acquiring the rising and falling rules of the sun and the change of the shadow at the position of the monitoring place to obtain the optimal shooting time and angle of the monitoring image and ensure that the color difference of the crack and the bottom surface reaches the maximum;
ST2, acquiring a dangerous rock monitoring image by the unmanned aerial vehicle according to a preset track;
ST3, acquiring a dangerous rock monitoring image;
and ST4, identifying the dangerous rock fracture.
8. The visual image monitoring and recognition system according to claim 7, wherein: the specific implementation steps in step ST2 are:
ST21, one or more near-field charging devices (2) are arranged on a path from a control site to the dangerous rock mass (3) according to the secondary navigation distance of the unmanned aerial vehicle (1);
ST22, the unmanned aerial vehicle (1) flies to a preset near-field charging device (2) according to the planned path and the GPS positioning device, identifies a charging platform (22) of the near-field charging device (2) through a camera arranged at the bottom of the unmanned aerial vehicle (1), adjusts the position according to the identification result, lands on the charging platform (22) and charges;
ST23, after charging, the unmanned aerial vehicle (1) takes off to collect the image of the dangerous rock mass (3) again, and after collecting, the unmanned aerial vehicle lands on the charging platform (22) to charge again and then returns;
realize utilizing unmanned aerial vehicle to carry out remote monitoring to the dangerous rock through above step.
9. The visual image monitoring and recognition system of claim 8, wherein: one or more relay airships (5) are also arranged, and a GPS positioning device and a height instrument are arranged on the relay airships (5);
a near-field charging device (2) is arranged on the relay airship (5);
each relay airship (5) is arranged along the flight path of the unmanned aerial vehicle, and a relay is arranged on each relay airship (5) and is used for amplifying the control signal of the unmanned aerial vehicle (1) so as to control the action of the unmanned aerial vehicle (1);
the working path to dangerous rock mass (3) is preset in unmanned aerial vehicle (1), the working path is the near field charging device (2) that sets up near using being close to dangerous rock mass (3) as starting point and terminal to the flight path who gathers the image around dangerous rock mass (3) completely as the purpose, reach near field charging device (2) that set up near dangerous rock mass (3) and accomplish the back of charging at unmanned aerial vehicle (1), carry out the working path promptly.
10. The visual image monitoring and recognition system according to claim 9, wherein: the charging platform (22) is provided with a plurality of charging electrodes (221) which are arranged in an array and used for being electrically connected with a connecting electrode (11) at the bottom of the unmanned aerial vehicle (1) to realize charging;
an elastic conductive contact (229) is provided at a position of each of the charging electrodes (221) so as to connect the row conductive line (228) of the current charging electrode (221) to the column conductive line (227) when the connection electrode (11) is pressed down, and to separate the row conductive line (228) of the current charging electrode (221) from the column conductive line (227) when the connection electrode (11) is separated;
a plurality of row detection switches (223) are arranged on one side of the charging electrodes (221) arranged along the array, one ends of the row detection switches (223) are electrically connected with row leads (228) of each row, and the other ends of the row detection switches (223) are electrically connected with detection signals;
a plurality of column detection switches (222) are arranged on one side of the charging electrodes (221) arranged along the array, one end of each column detection switch (222) is electrically connected with a column conducting wire (227) of each column, and the other end of each column detection switch (222) is electrically connected with a detection signal source;
a plurality of row charging switches (225) are arranged on one side of the charging electrodes (221) arranged along the array, one ends of the row charging switches (225) are electrically connected with row conducting wires (228) of each row, and the other ends of the row charging switches (225) are electrically connected with a charging power supply;
A plurality of column charging switches (224) are arranged on one side of the charging electrodes (221) arranged along the array, one ends of the column charging switches (224) are electrically connected with column conducting wires (227) of all columns, and the other ends of the column charging switches (224) are electrically connected with a charging power supply.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010509305.0A CN111861982B (en) | 2020-06-07 | 2020-06-07 | Visual image monitoring and identifying system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010509305.0A CN111861982B (en) | 2020-06-07 | 2020-06-07 | Visual image monitoring and identifying system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111861982A true CN111861982A (en) | 2020-10-30 |
| CN111861982B CN111861982B (en) | 2024-03-12 |
Family
ID=72986411
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010509305.0A Active CN111861982B (en) | 2020-06-07 | 2020-06-07 | Visual image monitoring and identifying system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111861982B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113743227A (en) * | 2021-08-06 | 2021-12-03 | 山东大学 | Rock mass fracture intelligent identification method and system based on proximity photography |
| CN115600034A (en) * | 2022-12-12 | 2023-01-13 | 深圳酷源数联科技有限公司（Cn） | Target image detection method and system |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN202929383U (en) * | 2012-09-11 | 2013-05-08 | 深圳一电科技有限公司 | Unmanned plane and automatic charging system thereof |
| CN105336119A (en) * | 2015-11-27 | 2016-02-17 | 中国地质调查局水文地质环境地质调查中心 | Collapse mass crack identification alarm device and identification method thereof |
| CN106056593A (en) * | 2016-05-26 | 2016-10-26 | 招商局重庆交通科研设计院有限公司 | Dangerous rock displacement monitoring method based on image modeling |
| US20180202266A1 (en) * | 2017-01-13 | 2018-07-19 | Baker Hughes, A Ge Company, Llc | Near wellbore discrete fracture networks |
| CN108492281A (en) * | 2018-03-06 | 2018-09-04 | 陕西师范大学 | A method of fighting Bridge Crack image detection of obstacles and the removal of network based on production |
| CN108725820A (en) * | 2018-04-08 | 2018-11-02 | 北京领航智能科技发展有限公司 | Unmanned plane automatic charging platform |
| CN109840483A (en) * | 2019-01-11 | 2019-06-04 | 深圳大学 | A kind of method and device of landslide fissure detection and identification |
| US20190197340A1 (en) * | 2016-01-15 | 2019-06-27 | Wuhan Wuda Zoyon Science And Technology Co., Ltd. | Object surface deformation feature extraction method based on line scanning three-dimensional point cloud |
| CN110065645A (en) * | 2019-04-23 | 2019-07-30 | 河南三和航空工业有限公司 | High-precision UAV Flight Control System |
| CN110175658A (en) * | 2019-06-26 | 2019-08-27 | 浙江大学 | A kind of distress in concrete recognition methods based on YOLOv3 deep learning |
| CN110211097A (en) * | 2019-05-14 | 2019-09-06 | 河海大学 | A kind of crack image detecting method based on the migration of Faster R-CNN parameter |
| CN110569730A (en) * | 2019-08-06 | 2019-12-13 | 福建农林大学 | Road surface crack automatic identification method based on U-net neural network model |
| CN111091178A (en) * | 2019-11-22 | 2020-05-01 | 国网山西省电力公司大同供电公司 | Porcelain bottle crack identification method based on convolutional neural network |
| CN111223086A (en) * | 2020-01-13 | 2020-06-02 | 南京理工大学 | Building crack identification and identification effect optimization method based on deep learning |
-
2020
- 2020-06-07 CN CN202010509305.0A patent/CN111861982B/en active Active
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN202929383U (en) * | 2012-09-11 | 2013-05-08 | 深圳一电科技有限公司 | Unmanned plane and automatic charging system thereof |
| CN105336119A (en) * | 2015-11-27 | 2016-02-17 | 中国地质调查局水文地质环境地质调查中心 | Collapse mass crack identification alarm device and identification method thereof |
| US20190197340A1 (en) * | 2016-01-15 | 2019-06-27 | Wuhan Wuda Zoyon Science And Technology Co., Ltd. | Object surface deformation feature extraction method based on line scanning three-dimensional point cloud |
| CN106056593A (en) * | 2016-05-26 | 2016-10-26 | 招商局重庆交通科研设计院有限公司 | Dangerous rock displacement monitoring method based on image modeling |
| US20180202266A1 (en) * | 2017-01-13 | 2018-07-19 | Baker Hughes, A Ge Company, Llc | Near wellbore discrete fracture networks |
| CN108492281A (en) * | 2018-03-06 | 2018-09-04 | 陕西师范大学 | A method of fighting Bridge Crack image detection of obstacles and the removal of network based on production |
| CN108725820A (en) * | 2018-04-08 | 2018-11-02 | 北京领航智能科技发展有限公司 | Unmanned plane automatic charging platform |
| CN109840483A (en) * | 2019-01-11 | 2019-06-04 | 深圳大学 | A kind of method and device of landslide fissure detection and identification |
| CN110065645A (en) * | 2019-04-23 | 2019-07-30 | 河南三和航空工业有限公司 | High-precision UAV Flight Control System |
| CN110211097A (en) * | 2019-05-14 | 2019-09-06 | 河海大学 | A kind of crack image detecting method based on the migration of Faster R-CNN parameter |
| CN110175658A (en) * | 2019-06-26 | 2019-08-27 | 浙江大学 | A kind of distress in concrete recognition methods based on YOLOv3 deep learning |
| CN110569730A (en) * | 2019-08-06 | 2019-12-13 | 福建农林大学 | Road surface crack automatic identification method based on U-net neural network model |
| CN111091178A (en) * | 2019-11-22 | 2020-05-01 | 国网山西省电力公司大同供电公司 | Porcelain bottle crack identification method based on convolutional neural network |
| CN111223086A (en) * | 2020-01-13 | 2020-06-02 | 南京理工大学 | Building crack identification and identification effect optimization method based on deep learning |
Non-Patent Citations (3)
| Title |
|---|
| MOHAN A等: "Crack detection using image processing: A critical review and analysis", 《ALEXANDRIA ENGINEERING JOURNAL》, vol. 57, no. 2, pages 787 - 798, XP055692282, DOI: 10.1016/j.aej.2017.01.020 * |
| 张茂省等: "地质灾害风险调查的方法与实践", 《地质通报》, vol. 27, no. 8, pages 1205 - 1216 * |
| 徐建强等: "公路边坡危岩智能安全监测及现场灾害报警技术", 《土工基础》, vol. 31, no. 2, pages 241 - 242 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113743227A (en) * | 2021-08-06 | 2021-12-03 | 山东大学 | Rock mass fracture intelligent identification method and system based on proximity photography |
| CN115600034A (en) * | 2022-12-12 | 2023-01-13 | 深圳酷源数联科技有限公司（Cn） | Target image detection method and system |
| CN115600034B (en) * | 2022-12-12 | 2023-03-10 | 深圳酷源数联科技有限公司 | Target image detection method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111861982B (en) | 2024-03-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108765404B (en) | A kind of road damage testing method and device based on deep learning image classification | |
| CN111861982A (en) | Visual image monitoring and identifying system | |
| CN102590823B (en) | Rapid extraction and reconstruction method for data power line of airborne LIDAR | |
| CN111342391B (en) | Power transmission line insulator and line fault inspection method and inspection system | |
| CN113284124B (en) | Photovoltaic panel defect detection method based on unmanned aerial vehicle vision | |
| CN112101088B (en) | Unmanned aerial vehicle electric power automatic inspection method, device and system | |
| CN108981924B (en) | On-site detection system for early damage of blades of large-scale in-service wind turbine | |
| CN106200640A (en) | A kind of automatic charging system based on image recognition location technology and charging method thereof | |
| CN103345637A (en) | On-line detecting system and method of hydrophobicity of composite insulator of electric transmission line | |
| CN111855501B (en) | Automatic water spraying composite insulator hydrophobicity detection system and method based on unmanned aerial vehicle | |
| CN111831430A (en) | Electrical equipment defect identification system based on edge calculation | |
| CN111931559A (en) | Method for classifying tree species in corridor area of power transmission line | |
| Zhang et al. | Aerial image analysis based on improved adaptive clustering for photovoltaic module inspection | |
| CN104573650B (en) | A kind of electric wire detection sorting technique based on filter response | |
| CN110703661A (en) | Crop disease and insect pest monitoring system and control method thereof | |
| CN114549998A (en) | Satellite image-based anti-interference evaluation method for photovoltaic potential of roof | |
| CN112947567A (en) | Method for inspecting power transmission line by using multi-rotor unmanned aerial vehicle | |
| CN115642877A (en) | Photovoltaic module abnormal shielding detection method and system based on deep learning | |
| CN112528979A (en) | Transformer substation inspection robot obstacle distinguishing method and system | |
| CN111762040B (en) | Near-field relay unmanned aerial vehicle dangerous rock monitoring system and method | |
| CN112270234B (en) | Power transmission line insulation sub-target identification method based on aerial image | |
| CN106149577A (en) | A kind of wedge shape chargeable electrical network shaft tower unmanned plane airplane parking area and control method thereof | |
| CN117036825A (en) | Solar cell panel detection method, medium and system | |
| CN115102287A (en) | Intelligent management and control system for new energy power station centralized area | |
| CN209938953U (en) | Pier detection unmanned aerial vehicle device |
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 |