CN109902668A - The pavement detection system and detection method of UAV system - Google Patents
The pavement detection system and detection method of UAV system Download PDFInfo
- Publication number
- CN109902668A CN109902668A CN201910305803.0A CN201910305803A CN109902668A CN 109902668 A CN109902668 A CN 109902668A CN 201910305803 A CN201910305803 A CN 201910305803A CN 109902668 A CN109902668 A CN 109902668A
- Authority
- CN
- China
- Prior art keywords
- image
- unmanned plane
- gray
- data set
- value
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
- H04N7/183—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast for receiving images from a single remote source
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Signal Processing (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses the pavement detection system of UAV system and detection method, including unmanned plane, GPS system and drone flying height control system are provided on the unmanned plane；Equipped with camera on the unmanned plane, the camera connects imaging sensor；The output end of described image sensor is connected to the input terminal of signal amplification circuit, and the output end of the signal amplification circuit is connected to the input terminal of signal conversion circuit；The output end of the signal conversion circuit is connected to wireless digital broadcasting station；The input terminal connection signal of the signal amplification circuit detects and the output end of processing circuit, and the input terminal of signal detection and processing circuit connects controller；Terrestrial wireless data radio station connects ground-based server by level pinboard, and the wireless digital broadcasting station being mounted on unmanned plane is wirelessly communicated with terrestrial wireless data radio station.Different information of road surface can be acquired and carry out image procossing respectively, establish corresponding model, dynamic road pavement can be detected in real time.
Description
Technical field
This disclosure relates to traffic detection technique field, pavement detection system and detection more particularly to UAV system
Method.
Background technique
With the development of road traffic, road traffic plays important role in economic development, also goes on a journey to people
Bring great convenience.However, the problem of it is supervened is also especially prominent, e.g., road surface is sunk, and road surface fracture is dropped object, applied
A series of problems, such as solid waste generated in work is cleared up not in time, the toy stopped among road, is seriously affecting vehicle just
Often traveling, causes a large amount of traffic accidents.Some Domestic pavement detection system is served only for solving the problems, such as some of them, Wu Faquan at present
The problem that face is quick, efficient solving road.If dropped object on highway, an engineering is driven dependent on two people
Vehicle, 13.6 ten thousand kilometers of visual inspection whole nation highway mileage for completing 50 kilometers along highway daily expend a large amount of manpower objects
Power.
Summary of the invention
In order to solve the deficiencies in the prior art, embodiment of the present disclosure provides the pavement detection system of UAV system, energy
Enough real-time dynamic road pavements are detected.
To achieve the goals above, the application uses following technical scheme:
The pavement detection system of UAV system, including unmanned plane, GPS system is provided on the unmanned plane and unmanned plane flies
Row height control system；
Equipped with camera on the unmanned plane, the camera connects imaging sensor；Described image sensor it is defeated
Outlet is connected to the input terminal of signal amplification circuit, and the output end of the signal amplification circuit is connected to the defeated of signal conversion circuit
Enter end；The output end of the signal conversion circuit is connected to wireless digital broadcasting station；
The input terminal connection signal of the signal amplification circuit detects and the output end of processing circuit, signal detection and processing
The input terminal of circuit connects controller；
Terrestrial wireless data radio station connects ground-based server by level pinboard, the wireless data sending electricity being mounted on unmanned plane
Platform is wirelessly communicated with terrestrial wireless data radio station.
Further technical solution carries out the processing of image, including ash in ground-based server for the image to acquisition
Degreeization, image smoothing and sharpening, image gray-scale transformation, image segmentation；
Picture after image procossing is extracted into corresponding object features；
Image procossing is carried out respectively for the different images of shooting, obtains image data set, and the data set of acquisition is carried out
Classification processing carries out pre-training using image data set in convolutional network, optimizes to model.
Further technical solution carries out pre-training in convolutional network: by each width figure characteristic parameter in data set
Using back-propagation algorithm and stochastic gradient descent method, according to the size of the loss value of propagated forward, Lai Jinhang backpropagation
Iteration updates each layer of weight, and when the loss value of model is intended to convergence, deconditioning model obtains deep learning mould
Type extracts deep learning feature in the full convolutional layer of the layer second from the bottom of image；
For giving any one image to be identified, it is input in trained deep learning model, extracts sample
Deep learning feature, differentiates which classification the image belongs to.
Further technical solution, image type included by image data set include but is not limited to that road surface is sunk image,
Road surface is broken image, drops subject image, and the solid waste generated in construction is cleared up image not in time, stopped among road small
Animal painting.
Embodiment of the disclosure also discloses the detection method of the pavement detection system of UAV system, comprising:
The position where unmanned plane is positioned, judges unmanned plane road detected, and control unmanned plane along detected road
Road flight；
Drone flying height control system is according to the electric wave emitted vertically downward and the electric wave returned by ground launch
The output power of unmanned plane is controlled, to adjust the flying height of unmanned plane；
According to the detection road surface where unmanned plane, image is shot by camera, keeps the road image front and back of shooting right
Continue in succession, and image is sent to controller after treatment, controller sends out image digital signal by wireless digital broadcasting station
It send to terrestrial wireless data radio station, ground-based server is reached by terrestrial wireless data radio station；
In ground-based server carry out image processing, including gray processing, image smoothing and sharpening, image gray-scale transformation,
Image segmentation；
Picture after image procossing is extracted into corresponding object features；
Image procossing is carried out respectively for the different images of shooting, obtains image data set, and the data set of acquisition is carried out
Classification processing carries out pre-training using self-control data set in convolutional network, optimizes to model.
Further technical solution carries out pre-training in convolutional network: by each width figure characteristic parameter in data set
Using back-propagation algorithm and stochastic gradient descent method, according to the size of the loss value of propagated forward, Lai Jinhang backpropagation
Iteration updates each layer of weight, and when the loss value of model is intended to convergence, deconditioning model obtains deep learning mould
Type extracts deep learning feature in the full convolutional layer of the layer second from the bottom of image；
For giving any one image to be identified, it is input in trained deep learning model, extracts sample
Deep learning feature, differentiates which classification the image belongs to.
Further technical solution, image type included by image data set include but is not limited to that road surface is sunk image,
Road surface is broken image, drops subject image, and the solid waste generated in construction is cleared up image not in time, stopped among road small
Animal painting.
Further technical solution, gray processing specifically:
The image that video camera collects computer is rgb format, and in the gray level image after conversion, a pixel indicates it
Gray value, then according to formula:
Y=0.299R+0.587G+0.114B
R, G, B are respectively the value of every bit pixel value red, green, blue, and range is 0 to 255, and R, G, B are assigned to Y in this way,
Only one value of the pixel of every bit in image.
Further technical solution after gray proces, carries out greyscale transformation.Enabling Y is the gray scale before transformation, and S is after converting
Gray scale, be using the general formula of logarithmic transformation:
S=clog (1+Y)
Wherein, c is a constant, Y >=0, and the relatively narrow low ash angle value of range in source images is mapped to range by logarithmic transformation
Wider gray scale interval, while being relatively narrow gray scale interval by the high gray value Interval Maps of wider range.
Image smoothing: further technical solution chooses out region in same image, using neighborhood averaging to image
It is handled.Its principle is that the gray value of neighbouring 8 pixels around it by the sum of the grayscale values of each pixel in original image is added, so
Gray value by the average value acquired (divided by 9) as the pixel in new figure afterwards.That is:
M is the coordinate of each neighborhood pixels in taken neighborhood, and N is the number for the neighborhood pixels for including in neighborhood.
Image sharpening processing: further technical solution uses linear Edge contrast, linear high pass filter is the most frequently used
Linear sharpening filter, can realized by calling filter2 function and fspecial function in MATLAB.
Further technical solution, image segmentation: the image partition method based on edge detection first determines the side in image
Then edge pixel again links together these pixels and just constitutes required zone boundary.
Picture after image procossing is extracted corresponding object features by further technical solution, and characteristics of image has color
Feature, textural characteristics, shape feature, space characteristics；
Color characteristic: object color characteristic is extracted using the method for color moment, distribution of color information is concentrated mainly on low order
In square, object color characteristic can be expressed using the first moment, second moment, third moment of color；
Textural characteristics: image texture characteristic is the characteristic quantification for extracting the variation of image-region interior intensity grade, using statistics
Method passes through the energy spectrum letter to image from auto-correlation function, that is, image energy spectrum function texture feature extraction of image
The fineness degree and direction characteristic parameter of texture are extracted in several calculating.
Shape feature: Fourier's shape description symbols method uses the Fourier transformation of object boundary as shape description, utilizes area
The closure and periodicity on domain boundary, convert one-dimensional problem for two-dimensional problems, export three kinds of shape expression by boundary point, respectively
It is curvature function, centroid distance, complex coordinates function.
Space characteristics: object or color region included in image are marked off to the image of segmentation, then according to these
Extracted region images feature, and establish index.
Compared with prior art, the beneficial effect of the disclosure is:
The technical solution of the disclosure can acquire different information of road surface and carry out image procossing respectively, establish corresponding mould
Type dynamic road pavement can be detected in real time.
Detailed description of the invention
The accompanying drawings constituting a part of this application is used to provide further understanding of the present application, and the application's shows
Meaning property embodiment and its explanation are not constituted an undue limitation on the present application for explaining the application.
Fig. 1 is the overhaul flow chart of disclosure one or more examples of implementation；
Fig. 2 is the detection system frame diagram of disclosure one or more examples of implementation.
Specific embodiment
It is noted that following detailed description is all illustrative, it is intended to provide further instruction to the application.Unless another
It indicates, all technical and scientific terms used herein has usual with the application person of an ordinary skill in the technical field
The identical meanings of understanding.
It should be noted that term used herein above is merely to describe specific embodiment, and be not intended to restricted root
According to the illustrative embodiments of the application.As used herein, unless the context clearly indicates otherwise, otherwise singular
Also it is intended to include plural form, additionally, it should be understood that, when in the present specification using term "comprising" and/or " packet
Include " when, indicate existing characteristics, step, operation, device, component and/or their combination.
In a kind of typical embodiment of the application, as shown in Fig. 2, the pavement detection system of UAV system is provided,
Including unmanned plane, GPS system, drone flying height control system, equipped with camera, imaging sensor, control on unmanned plane
Device processed, two wireless digital broadcasting stations (one by UAV flight, another connection computer), level pinboard, computer.It is described to take the photograph
As head connects imaging sensor；It is connected with signal amplification circuit and signal conversion between described image sensor and controller in turn
Circuit；That is the input of the output connection signal amplifying circuit of imaging sensor, the output connection signal conversion of signal amplification circuit
The input of circuit；The output of the signal conversion circuit connects wireless digital broadcasting station.The signal amplification circuit and controller it
Between be connected with signal detection and processing circuit, i.e. the detection of input connection signal and the output of processing circuit of signal amplification circuit,
The input of signal detection and processing circuit connects controller.Terrestrial wireless data radio station connects computer by level pinboard.
In a kind of typical embodiment of the application, as shown in Figure 1, the pavement detection method of UAV system is disclosed,
Specific steps are as follows:
Step 1: the control system of unmanned plane
Unmanned aerial vehicle control system includes GPS system and drone flying height control system.
Realize that unmanned plane independent navigation is nobody provided according to airborne global positioning system (GPS) receiver using GPS
The information such as machine warp, latitude, time, the feedback data of the route information as planning are joined in conjunction with other parameters collectively as control
Number, elevator and aileron rudder mainly for unmanned plane are controlled, and change unmanned plane during flying posture, to make unmanned plane by rule
Plot a course autonomous flight
It is a kind of with proportional+integral that drone flying height control system is that the unmanned plane height based on PI control is designed
The method of control come realize unmanned plane height control, using the method successively designed from inner looping to external loop；With circuit
Partition method analyzes closed loop transfer function, determines the mathematical model of height control system.Further according to time domain method and frequency
Domain method compares the time domain and frequency domain response index of each group parameter repeatedly in value range, until obtaining one group of optimum parameter value
Control system controls the output power of unmanned plane according to the electric wave emitted vertically downward and by the electric wave that ground launch is returned, with
Adjust the flying height of unmanned plane.
Step 2: the acquisition of image:
According to the detection road surface where unmanned plane, image is shot by camera, between the time of control unmanned plane shooting image
Every keeping the road image tandem docking of shooting continuous, and send an image to imaging sensor, imaging sensor is by image
It is converted into electronic signal and is transmitted to signal amplification circuit, after signal amplification circuit amplifies electronic signal and be transmitted to signal turn
Circuit is changed, signal conversion circuit converts analog signals into digital signal and is transmitted to controller, and controller passes through signal detection
And processing circuit detects and handles electronic signal, feeds back to signal amplification circuit, finally makes signal stabilization, controller will be stable
Image digital signal is sent to terrestrial wireless data radio station by wireless digital broadcasting station, reaches electricity by terrestrial wireless data radio station
Brain.Controller can also store picture signal to memory.
Step 3: image procossing: image procossing carries out on ground.Image procossing generally comprises four steps: (1) gray scale
Change, (2) image gray-scale transformation, (3) image sharpening and smooth, (4) image segmentation etc..Wherein image sharpening and smoothly can be same
Subregion carries out in image, can also take one of method for road different images.
1) gray processing
The image that video camera collects computer is rgb format, and in the gray level image after conversion, a pixel indicates it
Gray value, then according to formula:
Y=0.299R+0.587G+0.114B
R, G, B are respectively the value of every bit pixel value red, green, blue, and range is 0 to 255, and R, G, B are assigned to Y in this way,
Only one value of the pixel of every bit in image.
2) image gray-scale transformation
After gray proces, greyscale transformation is carried out.Enabling Y is the gray scale before transformation, and S is transformed gray scale, is become using logarithm
The general formula for changing logarithmic transformation is:
S=clog (1+Y)
Wherein, c is a constant, Y >=0, and the relatively narrow low ash angle value of range in source images is mapped to range by logarithmic transformation
Wider gray scale interval, while being relatively narrow gray scale interval by the high gray value Interval Maps of wider range, to extend dark
The value of pixel has compressed the value of high gray scale, can enhance low ash degree details in image.
3) image smoothing
Region is chosen out in same image, image is handled using neighborhood averaging.Its principle is will be in original image
The sum of the grayscale values of each pixel around it the gray value of neighbouring 8 pixels be added, then the average value acquired (divided by 9) are made
For the gray value of the pixel in new figure.That is:
M is the coordinate of each neighborhood pixels in taken neighborhood, and N is the number for the neighborhood pixels for including in neighborhood.
4) image sharpening is handled
Using linear Edge contrast, linear high pass filter is most common linear sharpening filter, can be in MATLAB
In can be realized by calling filter2 function and fspecial function.
5) image segmentation
Image partition method based on edge detection first determines the edge pixel in image, then these pixels is connected again
The zone boundary being connected together needed for just constituting.
Step 4: characteristic parameter is extracted
According to road surface different situations, different object features are extracted.Common characteristics of image has color characteristic, texture special
Sign, shape feature, space characteristics etc..
1) color characteristic
Object color characteristic is extracted using the method for color moment, distribution of color information is concentrated mainly in low-order moment, is used
First moment, second moment, the third moment of color can express object color characteristic.
2) textural characteristics
Image texture characteristic is the characteristic quantification for extracting the variation of image-region interior intensity grade, using statistical method, from figure
Auto-correlation function (i.e. the energy spectrum function of image) texture feature extraction of picture, that is, pass through the meter of the energy spectrum function to image
It calculates, extracts the characteristic parameters such as fineness degree and the directionality of texture.
3) shape feature
Fourier's shape description symbols method, uses the Fourier transformation of object boundary as shape description, utilizes zone boundary
Two-dimensional problems are converted one-dimensional problem by closure and periodicity.Three kinds of shape expression are exported by boundary point, are curvature letter respectively
Number, centroid distance, complex coordinates function.
4) space characteristics
Object or color region included in image are marked off to the image of segmentation, then according to these extracted region figures
As feature, and establish index.
Step 5: model is established
1) construction self-control data set
Image Acquisition is carried out by the various situations that road surface is likely to occur, the image of acquisition is obtained by step 3 and step 4
To a large amount of object data set (characteristic parameter), prepare to establish model.
2) model is established
The data set for extracting acquisition carries out classification processing to the data set of acquisition, and the result handled is road surface appearance
Various situations.The pavement detection system of UAV system is write using OpenCV and Visual Studio.
System includes the various situations that road surface occurs, and e.g., road surface is sunk, and road surface fracture is dropped object, generated in construction
A series of problems, such as solid waste is cleared up not in time, the toy stopped among road.Training convolutional neural networks: by data set
In each width figure characteristic parameter use back-propagation algorithm and stochastic gradient descent method, according to the loss value of propagated forward
Size, Lai Jinhang backpropagation iteration update each layer of weight, when the loss value of model is intended to convergence, deconditioning
Model obtains deep learning model, extracts deep learning feature in the full convolutional layer of the layer second from the bottom of image.For given any
One image to be identified is input in trained deep learning model, is extracted the deep learning feature of sample, is passed through training
Method effectively differentiate which classification the image belongs to.
The foregoing is merely preferred embodiment of the present application, are not intended to limit this application, for the skill of this field
For art personnel, various changes and changes are possible in this application.Within the spirit and principles of this application, made any to repair
Change, equivalent replacement, improvement etc., should be included within the scope of protection of this application.
Claims (10)
1. the pavement detection system of UAV system, characterized in that including unmanned plane, be provided on the unmanned plane GPS system and
Drone flying height control system；
Equipped with camera on the unmanned plane, the camera connects imaging sensor；The output end of described image sensor
It is connected to the input terminal of signal amplification circuit, the output end of the signal amplification circuit is connected to the input of signal conversion circuit
End；The output end of the signal conversion circuit is connected to wireless digital broadcasting station；
The input terminal connection signal of the signal amplification circuit detects and the output end of processing circuit, signal detection and processing circuit
Input terminal connect controller；
Terrestrial wireless data radio station connects ground-based server by level pinboard, be mounted in wireless digital broadcasting station on unmanned plane with
Terrestrial wireless data radio station wirelessly communicates；
The processing of image, including gray processing, image smoothing and sharpening, figure are carried out for the image to acquisition in ground-based server
As greyscale transformation, image segmentation；
Picture after image procossing is extracted into corresponding object features；
Image procossing is carried out respectively for the different images of shooting, is obtained image data set, is classified to the data set of acquisition
Processing, carries out pre-training using image data set in convolutional network, optimizes to model.
2. the pavement detection system of UAV system as described in claim 1, characterized in that instructed in advance in convolutional network
Practice: each width figure characteristic parameter in data set being used into back-propagation algorithm and stochastic gradient descent method, according to preceding to biography
The size for the loss value broadcast, Lai Jinhang backpropagation iteration update each layer of weight, until the loss value of model is intended to receive
When holding back, deconditioning model obtains deep learning model, and it is special to extract deep learning in the full convolutional layer of the layer second from the bottom of image
Sign；
For giving any one image to be identified, it is input in trained deep learning model, extracts the depth of sample
Learning characteristic, differentiates which classification the image belongs to.
3. the pavement detection system of UAV system as described in claim 1, characterized in that image included by image data set
Type includes but is not limited to the sagging image in road surface, and road surface is broken image, drops subject image, the solid waste generated in construction is not
Cleaning image in time, the petty action object image that road centre stops.
4. the detection method of the pavement detection system of UAV system, characterized in that include:
The position where unmanned plane is positioned, judges unmanned plane road detected, and controls unmanned plane and flies along detected road
Row；
Drone flying height control system is controlled according to the electric wave emitted vertically downward and by the electric wave that ground launch is returned
The output power of unmanned plane, to adjust the flying height of unmanned plane；
According to the detection road surface where unmanned plane, image is shot by camera, the road image tandem docking of shooting is enable to connect
It is continuous, and image is sent to controller after treatment, image digital signal is sent to by controller by wireless digital broadcasting station
Terrestrial wireless data radio station reaches ground-based server by terrestrial wireless data radio station；
The processing of image, including gray processing, image smoothing and sharpening, image gray-scale transformation, image are carried out in ground-based server
Segmentation；
Picture after image procossing is extracted into corresponding object features；
Image procossing is carried out respectively for the different images of shooting, is obtained image data set, is classified to the data set of acquisition
Processing carries out pre-training using self-control data set in convolutional network, optimizes to model.
5. the detection method of the pavement detection system of UAV system as claimed in claim 4, characterized in that in convolutional network
It carries out pre-training: each width figure characteristic parameter in data set is used into back-propagation algorithm and stochastic gradient descent method, root
According to the size of the loss value of propagated forward, Lai Jinhang backpropagation iteration updates each layer of weight, until the loss value of model
When being intended to convergence, deconditioning model obtains deep learning model, extracts depth in the full convolutional layer of the layer second from the bottom of image
Learning characteristic；
For giving any one image to be identified, it is input in trained deep learning model, extracts the depth of sample
Learning characteristic, differentiates which classification the image belongs to.
6. the detection method of the pavement detection system of UAV system as claimed in claim 4, characterized in that image data set institute
Including image type include but is not limited to road surface sink image, road surface be broken image, drop subject image, generated in construction
Solid waste clears up image, the petty action object image stopped among road not in time.
7. the detection method of the pavement detection system of UAV system as claimed in claim 4, characterized in that gray processing is specific
Are as follows:
The image that video camera collects computer is rgb format, and in the gray level image after conversion, a pixel indicates its gray scale
Value, then according to formula:
Y=0.299R+0.587G+0.114B
R, G, B are respectively the value of every bit pixel value red, green, blue, and range is 0 to 255, and R, G, B are assigned to Y, image in this way
In the pixel of every bit only one value.
Further technical solution after gray proces, carries out greyscale transformation, and enabling Y be the gray scale before transformation, and S is transformed ash
Degree is using the general formula of logarithmic transformation:
S=clog (1+Y)
Wherein, c is a constant, Y >=0, and the relatively narrow low ash angle value of range in source images is mapped to wider range by logarithmic transformation
Gray scale interval, while by the high gray value Interval Maps of wider range be relatively narrow gray scale interval.
8. the detection method of the pavement detection system of UAV system as claimed in claim 4, characterized in that image smoothing:
Region is chosen out in same image, image is handled using neighborhood averaging.Its principle is by each pixel in original image
Sum of the grayscale values around it the gray value of neighbouring 8 pixels be added, then using the average value acquired as the pixel in new figure
Gray value, it may be assumed that
M is the coordinate of each neighborhood pixels in taken neighborhood, and N is the number for the neighborhood pixels for including in neighborhood；
Image sharpening processing: using linear Edge contrast, and linear high pass filter is most common linear sharpening filter, can be
It can be realized by calling filter2 function and fspecial function in MATLAB.
9. the detection method of the pavement detection system of UAV system as claimed in claim 4, characterized in that image segmentation: base
In the image partition method of edge detection, first determines the edge pixel in image, then these pixels are linked together just again
Zone boundary needed for constituting.
10. the detection method of the pavement detection system of UAV system as claimed in claim 4, characterized in that by image procossing
Picture afterwards extracts corresponding object features, and characteristics of image has color characteristic, textural characteristics, shape feature, space characteristics；
Color characteristic: extracting object color characteristic using the method for color moment, and distribution of color information is concentrated mainly in low-order moment,
Object color characteristic can be expressed using the first moment, second moment, third moment of color；
Textural characteristics: image texture characteristic is the characteristic quantification for extracting the variation of image-region interior intensity grade, using statistical method,
From auto-correlation function, that is, image energy spectrum function texture feature extraction of image, that is, pass through the meter of the energy spectrum function to image
It calculates, extracts the fineness degree and direction characteristic parameter of texture；
Shape feature: Fourier's shape description symbols method uses the Fourier transformation of object boundary as shape description, utilizes regional edge
The closure and periodicity on boundary, convert one-dimensional problem for two-dimensional problems, export three kinds of shape expression by boundary point, are bent respectively
Rate function, centroid distance, complex coordinates function；
Space characteristics: object or color region included in image are marked off to the image of segmentation, then according to these regions
Characteristics of image is extracted, and establishes index.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2018116146425 | 2018-12-27 | ||
| CN201811614642.5A CN109447047A (en) | 2018-12-27 | 2018-12-27 | The pavement detection system and detection method of UAV system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109902668A true CN109902668A (en) | 2019-06-18 |
| CN109902668B CN109902668B (en) | 2022-04-22 |
Family
ID=65539655
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811614642.5A Withdrawn CN109447047A (en) | 2018-12-27 | 2018-12-27 | The pavement detection system and detection method of UAV system |
| CN201910305803.0A Active CN109902668B (en) | 2018-12-27 | 2019-04-16 | Unmanned aerial vehicle-mounted road surface detection system and detection method |
Family Applications Before (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811614642.5A Withdrawn CN109447047A (en) | 2018-12-27 | 2018-12-27 | The pavement detection system and detection method of UAV system |
Country Status (1)
| Country | Link |
|---|---|
| CN (2) | CN109447047A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112666963A (en) * | 2020-12-18 | 2021-04-16 | 浙江师范大学 | Road pavement crack detection system based on four-axis unmanned aerial vehicle and detection method thereof |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080075361A1 (en) * | 2006-09-21 | 2008-03-27 | Microsoft Corporation | Object Recognition Using Textons and Shape Filters |
| WO2013090830A1 (en) * | 2011-12-16 | 2013-06-20 | University Of Southern California | Autonomous pavement condition assessment |
| WO2015010451A1 (en) * | 2013-07-22 | 2015-01-29 | 浙江大学 | Method for road detection from one image |
| CN205003549U (en) * | 2015-08-24 | 2016-01-27 | 深圳市诺亚星辰科技开发有限公司 | Single rotor unmanned aerial vehicle is flight control hardware systems independently |
| CN105389988A (en) * | 2015-12-07 | 2016-03-09 | 北京航空航天大学 | Multi-unmanned aerial vehicle cooperation highway intelligent inspection system |
| CN106012780A (en) * | 2016-07-15 | 2016-10-12 | 京东方科技集团股份有限公司 | Road surface detection equipment and road surface detection method |
| CN106651872A (en) * | 2016-11-23 | 2017-05-10 | 北京理工大学 | Prewitt operator-based pavement crack recognition method and system |
| CN107945153A (en) * | 2017-11-07 | 2018-04-20 | 广东广业开元科技有限公司 | A kind of road surface crack detection method based on deep learning |
| CN207399371U (en) * | 2017-10-31 | 2018-05-22 | 南京开天眼无人机科技有限公司 | A kind of Unmanned Aerial Vehicle Data overlapping system |
| CN108416307A (en) * | 2018-03-13 | 2018-08-17 | 北京理工大学 | A kind of Aerial Images road surface crack detection method, device and equipment |
-
2018
- 2018-12-27 CN CN201811614642.5A patent/CN109447047A/en not_active Withdrawn
-
2019
- 2019-04-16 CN CN201910305803.0A patent/CN109902668B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080075361A1 (en) * | 2006-09-21 | 2008-03-27 | Microsoft Corporation | Object Recognition Using Textons and Shape Filters |
| WO2013090830A1 (en) * | 2011-12-16 | 2013-06-20 | University Of Southern California | Autonomous pavement condition assessment |
| WO2015010451A1 (en) * | 2013-07-22 | 2015-01-29 | 浙江大学 | Method for road detection from one image |
| CN205003549U (en) * | 2015-08-24 | 2016-01-27 | 深圳市诺亚星辰科技开发有限公司 | Single rotor unmanned aerial vehicle is flight control hardware systems independently |
| CN105389988A (en) * | 2015-12-07 | 2016-03-09 | 北京航空航天大学 | Multi-unmanned aerial vehicle cooperation highway intelligent inspection system |
| CN106012780A (en) * | 2016-07-15 | 2016-10-12 | 京东方科技集团股份有限公司 | Road surface detection equipment and road surface detection method |
| CN106651872A (en) * | 2016-11-23 | 2017-05-10 | 北京理工大学 | Prewitt operator-based pavement crack recognition method and system |
| CN207399371U (en) * | 2017-10-31 | 2018-05-22 | 南京开天眼无人机科技有限公司 | A kind of Unmanned Aerial Vehicle Data overlapping system |
| CN107945153A (en) * | 2017-11-07 | 2018-04-20 | 广东广业开元科技有限公司 | A kind of road surface crack detection method based on deep learning |
| CN108416307A (en) * | 2018-03-13 | 2018-08-17 | 北京理工大学 | A kind of Aerial Images road surface crack detection method, device and equipment |
Non-Patent Citations (7)
| Title |
|---|
| ZHONGHUA SUN ET AL: "Road Surface Condition Classification Based on Color and Texture Information", 《2013 NINTH INTERNATIONAL CONFERENCE ON INTELLIGENT INFORMATION HIDING AND MULTIMEDIA SIGNAL PROCESSING》 * |
| 官建军等: "《无人机遥感测绘技术及应用》", 31 August 2018 * |
| 张晨等: "基于无人机视觉的道路违法搭建检测", 《计算机技术与发展》 * |
| 沙爱民等: "基于卷积神经网络的路表病害识别与测量", 《中国公路学报》 * |
| 狄亚平: "路面裂缝识别算法研究", 《中国优秀硕士学位论文库工程科技Ⅱ辑》 * |
| 郭家辉等: "基于航拍技术的道路异常情况的检测", 《电子技术与软件工程》 * |
| 高仁强: "基于无人机LiDAR数据的公路路面监测和平整度评价方法研究", 《应用基础与工程科学学报》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112666963A (en) * | 2020-12-18 | 2021-04-16 | 浙江师范大学 | Road pavement crack detection system based on four-axis unmanned aerial vehicle and detection method thereof |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109447047A (en) | 2019-03-08 |
| CN109902668B (en) | 2022-04-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105023014B (en) | A kind of shaft tower target extraction method in unmanned plane inspection transmission line of electricity image | |
| CN110297498A (en) | A kind of rail polling method and system based on wireless charging unmanned plane | |
| CN108334844B (en) | Line automatic tracking method for power transmission line inspection | |
| CN111985274B (en) | Remote sensing image segmentation method based on convolutional neural network | |
| CN107590438A (en) | A kind of intelligent auxiliary driving method and system | |
| CN106326858A (en) | Road traffic sign automatic identification and management system based on deep learning | |
| CN102346015A (en) | Method for measuring icing thickness of insulator of power transmission line based on video difference analysis | |
| CN110070530A (en) | A kind of powerline ice-covering detection method based on deep neural network | |
| CN108665468B (en) | Device and method for extracting tangent tower insulator string | |
| CN104517126A (en) | Air quality assessment method based on image analysis | |
| CN111709994B (en) | Autonomous unmanned aerial vehicle visual detection and guidance system and method | |
| CN113419564B (en) | Power channel inspection method based on fuzzy path | |
| CN103971384B (en) | Node cooperation target tracking method of wireless video sensor | |
| CN107065871A (en) | It is a kind of that dining car identification alignment system and method are walked based on machine vision certainly | |
| CN114089786A (en) | Autonomous inspection system based on unmanned aerial vehicle vision and along mountain highway | |
| CN113947555A (en) | Infrared and visible light fused visual system and method based on deep neural network | |
| CN104994347A (en) | Intelligent security video monitoring system and detection processing method thereof | |
| CN109813276A (en) | A kind of antenna for base station has a down dip angle measuring method and its system | |
| CN107221006A (en) | A kind of communication single pipe tower slant detection method based on unmanned plane imaging platform | |
| CN104713526A (en) | Method for detecting types of foreign matters on power transmission line | |
| CN107038718A (en) | Depth computing method under haze environment | |
| CN110007671A (en) | A kind of robot cluster co-located system and method | |
| CN104217440A (en) | Method for extracting built-up area from remote sensing image | |
| CN106780540A (en) | Towards the ground cloud atlas cloud layer tracking of photovoltaic generation and method for early warning | |
| CN109902668A (en) | The pavement detection system and detection method of UAV system |
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 |