CN109816780A - A kind of the transmission line of electricity three-dimensional point cloud generation method and device of binocular sequential images - Google Patents
A kind of the transmission line of electricity three-dimensional point cloud generation method and device of binocular sequential images Download PDFInfo
- Publication number
- CN109816780A CN109816780A CN201910099564.8A CN201910099564A CN109816780A CN 109816780 A CN109816780 A CN 109816780A CN 201910099564 A CN201910099564 A CN 201910099564A CN 109816780 A CN109816780 A CN 109816780A
- Authority
- CN
- China
- Prior art keywords
- point cloud
- binocular
- electricity
- transmission line
- sequential images
- 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
Landscapes
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
This application provides a kind of transmission line of electricity three-dimensional point cloud generation method of binocular sequential images and devices, and wherein method includes: and carries out Stereo matching to binocular sequential images to obtain atural object three-dimensional laser point cloud；The threedimensional model of power line is obtained using deep learning；In conjunction with the threedimensional model of atural object three-dimensional laser point cloud and power line, transmission line of electricity three-dimensional laser point cloud is obtained.The application obtains atural object three-dimensional laser point cloud by Stereo matching, the threedimensional model of power line is obtained by deep learning, combine to obtain transmission line of electricity three-dimensional laser point cloud, transmission line of electricity three-dimensional laser point cloud can be measured in real time safe distance below power line, whole process is full-automatic, without manual intervention, therefore there are the advantages such as easy to operate, popularization difficulty is small, real-time investigation security risk；And it is extracted using the automatic identification that deep learning method carries out power line, with the accumulation of data, accuracy of identification can be higher and higher, and strong environmental adaptability.
Description
Technical field
This application involves a kind of transmission line of electricity of transmission line faultlocating technical field more particularly to binocular sequential images three-dimensionals
Point cloud generation method and device.
Background technique
Structure is complicated, in large scale for the existing high-voltage electric power circuit of power grid, and line channel environment is sufficiently complex, foreign matter lash,
Trees, violation construction, architecture against regulations etc. easily lead to line security distance deficiency and short circuit accident occur below route.Accident is once
Occur, consequence is serious, and huge electric current may cause serious injuries, and failure causes line facility to damage, and tripping is stopped transport, right
Operation of power networks causes to impact, meanwhile, failure impacts urban area power supply, upsets the normal production and living of enterprise and resident
Order brings heavy economic losses.
Manned helicopter and unmanned plane carry laser scanning system power circuit channel inspection technology, and in China, power grid is patrolled
It is gradually applied in inspection.Airborne laser radar measuring system can well solve the problems such as space orientation and measurement accuracy,
It can directly and rapidly acquire line corridor high-precision three-dimensional laser point cloud data, and then rapidly obtain high-precision three-dimensional line
The landform in road corridor, landforms, atural object and line facility spatial information.With LiDAR point cloud data processing technique gradually at
Ripe, most domestic unmanned plane power-line patrolling system is equipped with LiDAR system at present.With the progress of sensor technology, swash
Photoscanner and positioning and orientation system are all gradually minimizing, LiDAR system also small light therewith, so as to by more rotors without
It is man-machine to be carried.But it is post-processing mostly absolutely currently based on the conducting wire atural object safety detection of LiDAR, i.e., first acquires entire conducting wire
The data on road, then interior industry is handled (including positioning and orientation data processing, LiDAR point cloud generate, conducting wire extraction etc.) again.Afterwards
The advantages of processing mode is precision height, but it is larger to be delayed, and personnel's real-time on-site not easy to overhaul solves security risk.After in addition,
Process flow complexity is high, it is difficult to be grasped by general staff.This two large problems still restricts unmanned plane laser radar system
Large-scale use in electric inspection process work.
Summary of the invention
This application provides a kind of transmission line of electricity three-dimensional point cloud generation method of binocular sequential images and device, this method is first
It first passes through and atural object three-dimensional colour laser point cloud is obtained to binocular sequential images progress Stereo matching；Then deep learning target is utilized
Detection algorithm classifies to unmanned plane image, extracts power line pixel coordinate, its back projection to three-dimensional space obtains
The threedimensional model of power line finally combines atural object three-dimensional laser point cloud and power line threedimensional model, to obtain transmission line of electricity three
Laser point cloud is tieed up, provides data supporting for the detection of subsequent atural object safe distance.
In view of this, the application first aspect provides a kind of transmission line of electricity three-dimensional point cloud generation side of binocular sequential images
Method, comprising:
Stereo matching is carried out to binocular sequential images and obtains atural object three-dimensional laser point cloud；
Using deep learning algorithm of target detection, classification is carried out to unmanned plane image and extracts power line pixel coordinate, is obtained
To the threedimensional model of power line；
In conjunction with the threedimensional model of atural object three-dimensional laser point cloud and power line, transmission line of electricity three-dimensional laser point cloud is obtained.
Preferably, atural object three-dimensional laser point cloud is obtained to binocular sequential images progress Stereo matching to specifically include:
Half global Stereo Matching Algorithm is executed to the binocular sequential images got and obtains disparity map；
Atural object three-dimensional laser point cloud is established according to disparity map.
Preferably, to the binocular sequential images that get execute half global Stereo Matching Algorithm obtain disparity map after, root
It is established before atural object three-dimensional laser point cloud according to disparity map further include:
Disparity map is optimized by inclined-plane smoothing algorithm.
Preferably, obtaining disparity map to the global Stereo Matching Algorithm of binocular sequential images execution half got includes:
Establish disparity search space；
It is optimized by Dynamic Programming, and the matching similarity after calculation optimization, obtains the parallax value of pixel；
Disparity map is depicted according to the parallax value of pixel.
Preferably, using deep learning algorithm of target detection, classification is carried out to unmanned plane image and extracts power line pixel
Coordinate, the threedimensional model for obtaining power line include:
Construct the machine learning model with more hidden layers；
Machine learning model is trained according to the sample training data got；
Classified according to the machine learning model after training to unmanned plane image, extracts power line pixel coordinate；
By power line pixel coordinate back projection to three-dimensional space, the threedimensional model of power line is obtained.
Preferably, building has the machine learning model of more hidden layers specifically:
Establish first five layer of VGG16 infrastructure network；
Two convolutional layers are converted to by fc6 and fc7 layers using astrous algorithm；
Increase by 3 convolutional layers and pool layers of average.
Preferably, in conjunction with the threedimensional model of atural object three-dimensional laser point cloud and power line, transmission line of electricity three-dimensional laser point is obtained
Yun Hou further include:
Judge whether the distance between atural object and power line are more than safe distance using transmission line of electricity three-dimensional laser point cloud.
The application second aspect provides a kind of transmission line of electricity three-dimensional point cloud generating means of binocular sequential images, based on such as the
The transmission line of electricity three-dimensional point cloud generation method of the binocular sequential images of one side is generated, comprising:
Unmanned plane, for obtaining binocular sequential images and unmanned plane image；
The processing module for connecting the unmanned plane, for executing the transmission line of electricity three of the binocular sequential images such as first aspect
The step of dimension point cloud generation method.
The application third aspect provides a kind of transmission line of electricity three-dimensional point cloud generating device of binocular sequential images, the equipment
Including processor and memory:
Said program code is transferred to the processor for storing program code by the memory；
The processor is used to execute the binocular sequence as described in above-mentioned first aspect according to the instruction in said program code
The step of transmission line of electricity three-dimensional point cloud generation method of column image.
The application fourth aspect provides a kind of computer readable storage medium, and the computer readable storage medium is for depositing
Program code is stored up, said program code is used to execute the transmission line of electricity three-dimensional point of binocular sequential images described in above-mentioned first aspect
Cloud generation method.
The 5th aspect of the application provides a kind of computer program product including instruction, when run on a computer,
So that the computer executes the transmission line of electricity three-dimensional point cloud generation method of binocular sequential images described in above-mentioned first aspect.
As can be seen from the above technical solutions, the application has the following advantages:
This application provides a kind of transmission line of electricity three-dimensional point cloud generation method of binocular sequential images and device, wherein method
It include: to carry out Stereo matching to binocular sequential images to obtain atural object three-dimensional laser point cloud；Using deep learning algorithm of target detection,
Classification is carried out to unmanned plane image and extracts power line pixel coordinate, obtains the threedimensional model of power line；Swash in conjunction with atural object three-dimensional
The threedimensional model of luminous point cloud and power line obtains transmission line of electricity three-dimensional laser point cloud.The application obtains atural object by Stereo matching
Three-dimensional laser point cloud obtains the threedimensional model of power line by deep learning, combines to obtain transmission line of electricity three-dimensional laser point
Cloud, transmission line of electricity three-dimensional laser point cloud can be measured in real time safe distance below power line, and whole process is full-automatic,
Without manual intervention, therefore there are the advantages such as easy to operate, popularization difficulty is small, real-time investigation security risk；And use depth
The automatic identification that learning method carries out power line is extracted, and with the accumulation of data, the increase of sample size, accuracy of identification can be increasingly
Height, and without studying different recognition methods, strong environmental adaptability for different scenes.
Detailed description of the invention
It in ord to more clearly illustrate embodiments of the present application, below will be to required use in embodiment or description of the prior art
Attached drawing be briefly described, it should be apparent that, the drawings in the following description are only some examples of the present application, for this
For the those of ordinary skill of field, without any creative labor, it can also be obtained according to these attached drawings other
Attached drawing.
Fig. 1 is an a kind of implementation of transmission line of electricity three-dimensional point cloud generation method of binocular sequential images provided by the present application
The schematic diagram of example；
Fig. 2 is a kind of another reality of the transmission line of electricity three-dimensional point cloud generation method of binocular sequential images provided by the present application
Apply the schematic diagram of example；
Fig. 3 is a kind of another reality of the transmission line of electricity three-dimensional point cloud generation method of binocular sequential images provided by the present application
Apply the flow chart of the acquisition power line threedimensional model of example.
Specific embodiment
This application provides a kind of transmission line of electricity three-dimensional point cloud generation method of binocular sequential images and device, this method is first
It first passes through and atural object three-dimensional colour laser point cloud is obtained to binocular sequential images progress Stereo matching；Then deep learning target is utilized
Detection algorithm classifies to unmanned plane image, extracts power line pixel coordinate, its back projection to three-dimensional space obtains
The threedimensional model of power line finally combines atural object three-dimensional laser point cloud and power line threedimensional model, to obtain transmission line of electricity three
Laser point cloud is tieed up, provides data supporting for the detection of subsequent atural object safe distance.
To enable present invention purpose, feature, advantage more obvious and understandable, below in conjunction with the application
Attached drawing in embodiment, the technical scheme in the embodiment of the application is clearly and completely described, it is clear that disclosed below
Embodiment be only some embodiments of the present application, and not all embodiment.Based on the embodiment in the application, this field
Those of ordinary skill's all other embodiment obtained without making creative work belongs to the application protection
Range.
Referring to Fig. 1, the one of a kind of transmission line of electricity three-dimensional point cloud generation method of binocular sequential images provided by the present application
A embodiment, comprising:
101, Stereo matching is carried out to binocular sequential images and obtains atural object three-dimensional laser point cloud；
Stereo Matching Algorithm has very much, and the application is using half global Stereo Matching Algorithm.
102, using deep learning algorithm of target detection, classification is carried out to unmanned plane image and extracts power line pixel seat
Mark, obtains the threedimensional model of power line；
It can detecte the classification of unmanned plane image by deep learning algorithm of target detection, so that Classification and Identification goes out electric power
Then line extracts power line pixel coordinate.
103, in conjunction with the threedimensional model of atural object three-dimensional laser point cloud and power line, transmission line of electricity three-dimensional laser point cloud is obtained.
The application obtains atural object three-dimensional laser point cloud by Stereo matching, obtains the three-dimensional mould of power line by deep learning
Type combines to obtain transmission line of electricity three-dimensional laser point cloud, and transmission line of electricity three-dimensional laser point cloud can be to safety below power line
Distance is measured in real time, and whole process is full-automatic, is not necessarily to manual intervention, therefore with easy to operate, popularization difficulty is small, real
When the investigation advantages such as security risk；And it is extracted using the automatic identification that deep learning method carries out power line, with data
Accumulation, the increase of sample size, accuracy of identification can be higher and higher, and without studying different recognition methods, ring for different scenes
Border is adaptable.
It is one to a kind of transmission line of electricity three-dimensional point cloud generation method of binocular sequential images provided by the present application above
Embodiment is described in detail, below will be raw to a kind of transmission line of electricity three-dimensional point cloud of binocular sequential images provided by the present application
It is described in detail at another embodiment of method.
Referring to Fig. 2, a kind of transmission line of electricity three-dimensional point cloud generation method of binocular sequential images provided by the present application is another
One embodiment, comprising:
201, half global Stereo Matching Algorithm is executed to the binocular sequential images got and obtains disparity map；
Further, obtaining disparity map to the global Stereo Matching Algorithm of binocular sequential images execution half got includes:
Establish disparity search space；
It is optimized by Dynamic Programming, and the matching similarity after calculation optimization, obtains the parallax value of pixel；
Disparity map is depicted according to the parallax value of pixel.
It should be noted that half global Stereo Matching Algorithm is Semi-Global Matching, abbreviation SGM algorithm.This
Half global Stereo matching uses half global registration accelerated based on OpenCL (Open Computing Language) in application,
Accelerate to calculate by OpenCL, at least unmanned plane data of 7 kilometers of processing can be reached per hour.Half global registration algorithm
It is critical that the optimization of its parallax value there is no in global level, but based on a part in disparity map to certain point
Parallax value optimizes, and is realized by way of Dynamic Programming, greatly reduces the search range of disparity map in global optimization.
Substantially thinking is as follows for it:
For the certain point I in left figureleft(x, y), under epipolar-line constraint, the match point Candidate Set of the right figure are as follows:
Ir={ Iright(x ', y) | 0≤x '≤x-dmax}；
Wherein dmaxFor maximum allowable parallax.Disparity search space is established with this, is defined such as formula (1):
DS=C (x, y, d) | C (x, y, d)=Cost (Ileft(x, y), Iright(x-d, y)) } (1)
Wherein, Cost (Ileft(x, y), Iright(x-d, y)) for measuring Ileft(x, y) and Iright(x-d, y's) is similar
Degree, referred to as cost function.In local algorithm, it is only necessary to after calculating DS, find the smallest point of similarity, then substantially look for
To match point.But in SGM, need to optimize for DS by Dynamic Programming, such as formula (2):
Wherein p=(x, y), r are dynamic programming path.DS is disparity search space, as shown in Equation 1；I refers to that DS minimum is right
The parallax i answered；P1 and P2 is penalty factor, if parallax only poor 1, plus punishment P1, otherwise plus punishment P2, as regularization
Constraint.D is to allow parallax, dmaxIt is maximum allowable parallax.
Finally, the matching similarity after optimization is the average of the DS value in all directions, such as formula (3)
It is last only to takeThe as parallax value of pixel p.
Intuitively, SGM ensure that adjacent in disparity map by Dynamic Programming based on the smoothness constraint in stereoscopic vision
The flatness of parallax between point, and by the discontinuous punishment of neighborhood parallax value, no matching rate is reduced, and to a certain degree
On alleviate matching ambiguity the problem of.
After the Stereo matching for completing left and right camera acquisition image, all pixels point in image can be theoretically obtained
Depth information, to depict disparity map.In grayscale image, depth value is smaller, can be more aobvious with the closer point of camera distance
It must become clear.
For opposite LiDAR system, binocular vision system is a kind of more cheap solution, and for opposite monocular,
Binocular camera can directly carry out three-dimensional measuring without sky three.Therefore, by the advantage of the advantage of rotor wing unmanned aerial vehicle and binocular vision
Combine, build the multi-rotor unmanned aerial vehicle power circuit safety detecting system based on binocular vision, will be a kind of preferable solution
Certainly scheme.
It is more rare before based on the cable space three-dimensional modeling of binocular solid camera, main reason is that binocular solid
Dense Stereo Matching technology is difficult to realize when camera establishes 3 D stereoscopic image, and because conducting wire is originally very thin in image f iotaeld-of-view
Small, not intensive enough matching will lead to the problems such as conducting wire models loss and missing inspection.With the development of dense Stereo Matching technology, so that
The conducting wire three-dimensional modeling of binocular solid camera is possibly realized.This method is at low cost, can not only obtain accurate three information, simultaneously
Color information containing camera, it is more convenient to be visualized and handled.It is vertical that binocular may be implemented by GPU parallelization calculating simultaneously
The real-time of bulk measurement avoids complicated last handling process.
202, disparity map is optimized by inclined-plane smoothing algorithm；
Disparity map optimization algorithm is used based on inclined-plane smoothing algorithm (Slanted Plane Smoothing, SPS) to obtaining
Disparity map optimize.Stereo matching calculating and optimization are carried out using half global Stereo Matching Algorithm and inclined-plane smoothing algorithm
Afterwards, dense disparity map can be obtained, and then rebuilds dense three-dimensional point cloud (such as step 203).
203, atural object three-dimensional laser point cloud is established according to disparity map；
204, building has the machine learning model of more hidden layers；
Further, building has the machine learning model of more hidden layers specifically:
Establish first five layer of VGG16 infrastructure network；
Two convolutional layers are converted to by fc6 and fc7 layers using astrous algorithm；
Increase by 3 convolutional layers and pool layers of average.
It should be noted that VGG16 is the common convolutional neural networks of those skilled in the art, the application by pair
VGG16 convolutional neural networks improve to obtain machine learning model used in this application.Method particularly includes: neural network structure
Using the infrastructure network of VGG16, then fc6 and fc7 layers is converted to by first 5 layers before use using astrous algorithm
Two convolutional layers.3 convolutional layers and one pool layers of average are especially increased again.The feature map of different levels points
Not Yong Yu the offset of default box and the prediction of different classes of score, obtain final testing result finally by nms.
Deep learning SSD (Single Shot MultiBox Detector) target is used in power line extraction process
Detection algorithm, the detection method use artificial intelligence deep learning detection algorithm, have the machine learning of more hidden layers by constructing
The training data of model and magnanimity, and layer-by-layer eigentransformation, the character representation by sample in former space transform to a new feature
Space forms more abstract high-rise expression attribute classification or feature, by combination low-level feature to find the distribution of data
Character representation.
205, machine learning model is trained according to the sample training data got；
By collecting the power line unmanned plane image of a large amount of different zones, then using manually being marked out from image
Various types of power line is put into SSD algorithm frame as training sample using the image of these marks and is trained,
Obtain the SSD model (training the machine learning model finished) that power line automatically extracts.
206, classified according to the machine learning model after training to unmanned plane image, extract power line pixel seat
Mark；
In data acquisition, classified in real time to unmanned plane image by the machine learning model after the training,
Extract power line pixel coordinate.
It, then can be with for example, machine learning model Classification and Identification after a certain frame is trained in unmanned plane image is power line
It further identifies which pixel in the frame image is the pixel of power line, then extracts power line pixel coordinate.
207, by power line pixel coordinate back projection to three-dimensional space, the threedimensional model of power line is obtained；
The overall process of step 204 to step 207 is as shown in Figure 3.
Step 201 to step 203 is to obtain the process of atural object three-dimensional laser point cloud, and step 204 to step 207 is to obtain electricity
The process of the threedimensional model of the line of force, two processes can carry out simultaneously.
208, in conjunction with the threedimensional model of atural object three-dimensional laser point cloud and power line, transmission line of electricity three-dimensional laser point cloud is obtained；
The combination of the threedimensional model of atural object three-dimensional laser point cloud and power line can be a selected marker (such as electricity
Line bar) as basic point, it is a three-dimensional system of coordinate by two three-dimensional system of coordinate engagements.
209, using transmission line of electricity three-dimensional laser point cloud judge the distance between atural object and power line whether be more than safety away from
From；
Transmission line of electricity three-dimensional laser point cloud includes the data of altitude information (data such as disparity map) and power line of atural object, electricity
The line of force is known with respect to the height on ground, thus can be judged according to transmission line of electricity three-dimensional laser point cloud atural object and power line it
Between distance whether be more than safe distance.Concrete mode repeats no more.
The application is based on binocular camera device and acquires stereopsis, carries out Stereo matching and obtains atural object three-dimensional colour laser point
Cloud；Using deep learning method, power line is extracted from image automatically, and passes through the registration of image and laser radar, by image
In electric power line coordinates map in laser point cloud data, to obtain power line laser point cloud data.
(1) the application realizes the real-time detection of safe distance below power line, and whole process is full-automatic, without artificial
Intervene, therefore there are the advantages such as easy to operate, popularization difficulty is small, real-time investigation security risk.
(2) the application is extracted using the automatic identification that deep learning method carries out power line, with the accumulation of data, sample
The increase of amount, accuracy of identification can be higher and higher, and without studying different recognition methods, adaptive capacity to environment for different scenes
By force.
It is to a kind of the another of the transmission line of electricity three-dimensional point cloud generation method of binocular sequential images provided by the present application above
A embodiment is described in detail, below will be to a kind of transmission line of electricity three-dimensional point cloud of binocular sequential images provided by the present application
The embodiment of generating means is described in detail.
A kind of one embodiment of the transmission line of electricity three-dimensional point cloud generating means of binocular sequential images provided by the present application, base
It is generated in the transmission line of electricity three-dimensional point cloud generation method of the binocular sequential images of such as above-described embodiment, comprising:
Unmanned plane, for obtaining binocular sequential images and unmanned plane image；
The processing module for connecting unmanned plane, for executing the transmission line of electricity three-dimensional point of the binocular sequential images such as first aspect
The step of cloud generation method.
The application provides a kind of transmission line of electricity three-dimensional point cloud generating device of binocular sequential images, equipment include processor with
And memory:
Program code is transferred to processor for storing program code by memory；
Processor is used to execute the power transmission line of the binocular sequential images such as above-described embodiment according to the instruction in program code
The step of road three-dimensional point cloud generation method.
The application provides a kind of computer readable storage medium, and computer readable storage medium is used to store program code,
Program code is used to execute the transmission line of electricity three-dimensional point cloud generation method of the binocular sequential images of above-described embodiment.
The application provides a kind of computer program product including instruction, when run on a computer, so that calculating
Machine executes the transmission line of electricity three-dimensional point cloud generation method of the binocular sequential images of above-described embodiment.
The description of the present application and term " first " in above-mentioned attached drawing, " second ", " third ", " the 4th " etc. are (if deposited
) it is to be used to distinguish similar objects, without being used to describe a particular order or precedence order.It should be understood that use in this way
Data are interchangeable under appropriate circumstances, so that embodiments herein described herein for example can be in addition to illustrating herein
Or the sequence other than those of description is implemented.In addition, term " includes " and " having " and their any deformation, it is intended that
Cover it is non-exclusive include, for example, containing the process, method, system, product or equipment of a series of steps or units need not limit
In step or unit those of is clearly listed, but may include be not clearly listed or for these process, methods, produce
The other step or units of product or equipment inherently.
It should be appreciated that in this application, " at least one (item) " refers to one or more, and " multiple " refer to two or two
More than a."and/or" indicates may exist three kinds of relationships, for example, " A and/or B " for describing the incidence relation of affiliated partner
It can indicate: only exist A, only exist B and exist simultaneously tri- kinds of situations of A and B, wherein A, B can be odd number or plural number.Word
Symbol "/" typicallys represent the relationship that forward-backward correlation object is a kind of "or"." at least one of following (a) " or its similar expression, refers to
Any combination in these, any combination including individual event (a) or complex item (a).At least one of for example, in a, b or c
(a) can indicate: a, b, c, " a and b ", " a and c ", " b and c ", or " a and b and c ", and wherein a, b, c can be individually, can also
To be multiple.
It is apparent to those skilled in the art that for convenience and simplicity of description, the system of foregoing description,
The specific work process of device and unit, can refer to corresponding processes in the foregoing method embodiment, and details are not described herein.
In several embodiments provided herein, it should be understood that disclosed system, device and method can be with
It realizes by another way.For example, the apparatus embodiments described above are merely exemplary, for example, the unit
It divides, only a kind of logical function partition, there may be another division manner in actual implementation, such as multiple units or components
It can be combined or can be integrated into another system, or some features can be ignored or not executed.Another point, it is shown or
The mutual coupling, direct-coupling or communication connection discussed can be through some interfaces, the indirect coupling of device or unit
It closes or communicates to connect, can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, each functional unit in each embodiment of the application can integrate in one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of software functional units.
If the integrated unit is realized in the form of SFU software functional unit and sells or use as independent product
When, it can store in a computer readable storage medium.Based on this understanding, the technical solution of the application is substantially
The all or part of the part that contributes to existing technology or the technical solution can be in the form of software products in other words
It embodies, which is stored in a storage medium, including some instructions are used so that a computer
Equipment (can be personal computer, server or the network equipment etc.) executes the complete of each embodiment the method for the application
Portion or part steps.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only
Memory), random access memory (RAM, Random Access Memory), magnetic or disk etc. are various can store journey
The medium of sequence code.
The above, above embodiments are only to illustrate the technical solution of the application, rather than its limitations；Although referring to before
Embodiment is stated the application is described in detail, those skilled in the art should understand that: it still can be to preceding
Technical solution documented by each embodiment is stated to modify or equivalent replacement of some of the technical features；And these
It modifies or replaces, the spirit and scope of each embodiment technical solution of the application that it does not separate the essence of the corresponding technical solution.
Claims (10)
1. a kind of transmission line of electricity three-dimensional point cloud generation method of binocular sequential images characterized by comprising
Stereo matching is carried out to binocular sequential images and obtains atural object three-dimensional laser point cloud；
Using deep learning algorithm of target detection, classification is carried out to unmanned plane image and extracts power line pixel coordinate, obtains electricity
The threedimensional model of the line of force；
In conjunction with the threedimensional model of atural object three-dimensional laser point cloud and power line, transmission line of electricity three-dimensional laser point cloud is obtained.
2. a kind of transmission line of electricity three-dimensional point cloud generation method of binocular sequential images according to claim 1, feature exist
In carrying out Stereo matching to binocular sequential images and obtain atural object three-dimensional laser point cloud and specifically include:
Half global Stereo Matching Algorithm is executed to the binocular sequential images got and obtains disparity map；
Atural object three-dimensional laser point cloud is established according to disparity map.
3. a kind of transmission line of electricity three-dimensional point cloud generation method of binocular sequential images according to claim 2, feature exist
In, to the binocular sequential images that get execute half global Stereo Matching Algorithm obtain disparity map after, established according to disparity map
Before atural object three-dimensional laser point cloud further include:
Disparity map is optimized by inclined-plane smoothing algorithm.
4. a kind of transmission line of electricity three-dimensional point cloud generation method of binocular sequential images according to claim 2, feature exist
In obtaining disparity map to the global Stereo Matching Algorithm of binocular sequential images execution half got includes:
Establish disparity search space；
It is optimized by Dynamic Programming, and the matching similarity after calculation optimization, obtains the parallax value of pixel；
Disparity map is depicted according to the parallax value of pixel.
5. a kind of transmission line of electricity three-dimensional point cloud generation method of binocular sequential images according to claim 1, feature exist
In using deep learning algorithm of target detection, classify to unmanned plane image extracts power line pixel coordinate, obtains electric power
The threedimensional model of line includes:
Construct the machine learning model with more hidden layers；
Machine learning model is trained according to the sample training data got；
Classified according to the machine learning model after training to unmanned plane image, extracts power line pixel coordinate；
By power line pixel coordinate back projection to three-dimensional space, the threedimensional model of power line is obtained.
6. a kind of transmission line of electricity three-dimensional point cloud generation method of binocular sequential images according to claim 5, feature exist
In building has the machine learning model of more hidden layers specifically:
Establish first five layer of VGG16 infrastructure network；
Two convolutional layers are converted to by fc6 and fc7 layers using astrous algorithm；
Increase by 3 convolutional layers and pool layers of average.
7. a kind of transmission line of electricity three-dimensional point cloud generation method of binocular sequential images according to claim 1, feature exist
In in conjunction with the threedimensional model of atural object three-dimensional laser point cloud and power line, after obtaining transmission line of electricity three-dimensional laser point cloud further include:
Judge whether the distance between atural object and power line are more than safe distance using transmission line of electricity three-dimensional laser point cloud.
8. the transmission line of electricity three-dimensional point cloud generating means of a kind of binocular sequential images, based on as described in any one of claim 1 to 7
The transmission line of electricity three-dimensional point cloud generation methods of binocular sequential images generated characterized by comprising
Unmanned plane, for obtaining binocular sequential images and unmanned plane image；
The processing module for connecting the unmanned plane, for executing binocular sequential images as described in any one of claim 1 to 7
The step of transmission line of electricity three-dimensional point cloud generation method.
9. a kind of computer readable storage medium, which is characterized in that the computer readable storage medium is for storing program generation
Code, said program code require the transmission line of electricity three-dimensional point cloud of the described in any item binocular sequential images of 1-7 for perform claim
Generation method.
10. a kind of computer program product including instruction, which is characterized in that when run on a computer, so that described
The transmission line of electricity three-dimensional point cloud generation method of the computer perform claim requirement described in any item binocular sequential images of 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910099564.8A CN109816780B (en) | 2019-01-31 | 2019-01-31 | Power transmission line three-dimensional point cloud generation method and device of binocular sequence image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910099564.8A CN109816780B (en) | 2019-01-31 | 2019-01-31 | Power transmission line three-dimensional point cloud generation method and device of binocular sequence image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109816780A true CN109816780A (en) | 2019-05-28 |
| CN109816780B CN109816780B (en) | 2023-04-14 |
Family
ID=66606173
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910099564.8A Active CN109816780B (en) | 2019-01-31 | 2019-01-31 | Power transmission line three-dimensional point cloud generation method and device of binocular sequence image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109816780B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110580703A (en) * | 2019-09-10 | 2019-12-17 | 广东电网有限责任公司 | distribution line detection method, device, equipment and storage medium |
| CN111698468A (en) * | 2020-05-14 | 2020-09-22 | 中国电力工程顾问集团西南电力设计院有限公司 | Method for automatically monitoring three-dimensional scene based on power transmission line |
| CN112419176A (en) * | 2020-11-10 | 2021-02-26 | 国网江西省电力有限公司电力科学研究院 | Positive image point cloud enhancement method and device for single-loop power transmission channel conductor |
| CN112580428A (en) * | 2020-11-09 | 2021-03-30 | 义乌市输变电工程有限公司设计分公司 | Power distribution network design method and device |
| CN112836352A (en) * | 2021-01-12 | 2021-05-25 | 中国电建集团贵州电力设计研究院有限公司 | Power transmission line model generation method integrating three-dimensional design and laser point cloud |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103413133A (en) * | 2013-06-28 | 2013-11-27 | 广东电网公司电力科学研究院 | Automatically-extracting power line method in random laser point cloud data |
| CN107314762A (en) * | 2017-07-06 | 2017-11-03 | 广东电网有限责任公司电力科学研究院 | Atural object distance detection method below power line based on unmanned plane the sequence monocular image |
| CN107392247A (en) * | 2017-07-20 | 2017-11-24 | 广东电网有限责任公司电力科学研究院 | Atural object safe distance real-time detection method below a kind of power line |
| CN109087339A (en) * | 2018-06-13 | 2018-12-25 | 武汉朗视软件有限公司 | A kind of laser scanning point and Image registration method |
-
2019
- 2019-01-31 CN CN201910099564.8A patent/CN109816780B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103413133A (en) * | 2013-06-28 | 2013-11-27 | 广东电网公司电力科学研究院 | Automatically-extracting power line method in random laser point cloud data |
| CN107314762A (en) * | 2017-07-06 | 2017-11-03 | 广东电网有限责任公司电力科学研究院 | Atural object distance detection method below power line based on unmanned plane the sequence monocular image |
| CN107392247A (en) * | 2017-07-20 | 2017-11-24 | 广东电网有限责任公司电力科学研究院 | Atural object safe distance real-time detection method below a kind of power line |
| CN109087339A (en) * | 2018-06-13 | 2018-12-25 | 武汉朗视软件有限公司 | A kind of laser scanning point and Image registration method |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110580703A (en) * | 2019-09-10 | 2019-12-17 | 广东电网有限责任公司 | distribution line detection method, device, equipment and storage medium |
| CN110580703B (en) * | 2019-09-10 | 2024-01-23 | 广东电网有限责任公司 | Distribution line detection method, device, equipment and storage medium |
| CN111698468A (en) * | 2020-05-14 | 2020-09-22 | 中国电力工程顾问集团西南电力设计院有限公司 | Method for automatically monitoring three-dimensional scene based on power transmission line |
| CN112580428A (en) * | 2020-11-09 | 2021-03-30 | 义乌市输变电工程有限公司设计分公司 | Power distribution network design method and device |
| CN112419176A (en) * | 2020-11-10 | 2021-02-26 | 国网江西省电力有限公司电力科学研究院 | Positive image point cloud enhancement method and device for single-loop power transmission channel conductor |
| CN112419176B (en) * | 2020-11-10 | 2024-05-14 | 国网江西省电力有限公司电力科学研究院 | Single-loop transmission channel lead positive shooting image point cloud enhancement method and device |
| CN112836352A (en) * | 2021-01-12 | 2021-05-25 | 中国电建集团贵州电力设计研究院有限公司 | Power transmission line model generation method integrating three-dimensional design and laser point cloud |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109816780B (en) | 2023-04-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109816780A (en) | A kind of the transmission line of electricity three-dimensional point cloud generation method and device of binocular sequential images | |
| KR102126724B1 (en) | Method and apparatus for restoring point cloud data | |
| CN111897332B (en) | Semantic intelligent substation robot humanoid inspection operation method and system | |
| CN110097553B (en) | Semantic mapping system based on instant positioning mapping and three-dimensional semantic segmentation | |
| CN107808133B (en) | Unmanned aerial vehicle line patrol-based oil and gas pipeline safety monitoring method and system and software memory | |
| CN111832655A (en) | Multi-scale three-dimensional target detection method based on characteristic pyramid network | |
| CN108764187A (en) | Extract method, apparatus, equipment, storage medium and the acquisition entity of lane line | |
| CN113205116B (en) | Automatic extraction and track planning method for inspection shooting target point of unmanned aerial vehicle of power transmission line | |
| CN107167139A (en) | A kind of Intelligent Mobile Robot vision positioning air navigation aid and system | |
| CN109829908B (en) | Binocular image-based method and device for detecting safety distance of ground object below power line | |
| CN111458721B (en) | Exposed garbage identification and positioning method, device and system | |
| CN104091168A (en) | Method for automatically extracting and locating power line based on unmanned aerial vehicle image | |
| CN112528979B (en) | Transformer substation inspection robot obstacle distinguishing method and system | |
| CN113724388B (en) | High-precision map generation method, device, equipment and storage medium | |
| CN111126116A (en) | Unmanned ship river channel garbage identification method and system | |
| CN115082254A (en) | Lean control digital twin system of transformer substation | |
| CN113936210A (en) | Anti-collision method for tower crane | |
| CN113674355A (en) | Target identification and positioning method based on camera and laser radar | |
| Alidoost et al. | Y-shaped convolutional neural network for 3d roof elements extraction to reconstruct building models from a single aerial image | |
| CN115100741A (en) | Point cloud pedestrian distance risk detection method, system, equipment and medium | |
| CN109002633B (en) | Device network modeling method based on separate space | |
| CN116543283B (en) | Multimode target detection method considering modal uncertainty | |
| CN117115063A (en) | Multi-source data fusion application method | |
| CN115880555B (en) | Target detection method, model training method, device, equipment and medium | |
| CN111754388B (en) | Picture construction method and vehicle-mounted terminal |
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 |