CN103729883B - A kind of three-dimensional environment information gathering and reconfiguration system and method - Google Patents
A kind of three-dimensional environment information gathering and reconfiguration system and method Download PDFInfo
- Publication number
- CN103729883B CN103729883B CN201310742704.1A CN201310742704A CN103729883B CN 103729883 B CN103729883 B CN 103729883B CN 201310742704 A CN201310742704 A CN 201310742704A CN 103729883 B CN103729883 B CN 103729883B
- Authority
- CN
- China
- Prior art keywords
- data
- cloud
- module
- model
- range finder
- 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.)
- Active
Links
Landscapes
- Length Measuring Devices By Optical Means (AREA)
Abstract
The invention discloses a kind of three-dimensional environment information gathering and reconfiguration system and method, it includes scanning laser range finder, panorama camera, rotation platform, airborne processor, power supply, base, wire slide rail, signal picker, motor, train of reduction gears, panorama camera bracing frame, connecting rod;Base is provided with connecting rod, airborne processor, power supply, connecting rod is the most sequentially cased with train of reduction gears, wire slide rail, rotation platform, connecting rod top is the most sequentially provided with panorama camera bracing frame, panorama camera, rotation platform lower surface is fixed with signal picker, rotation platform upper surface is fixed with one or more scanning laser range finder, scanning laser range finder is positioned at the lower section of panorama camera bracing frame, and motor is located in train of reduction gears.It is big that present system measures visual angle, far measuring distance simultaneously;Information is accurate, abundant, and the existing good view effect of model of generation is provided that again abundant environment raw information;Advantage of lower cost.
Description
Technical field
The present invention relates to three-dimensional environment reconstruction field, particularly relate to a kind of three-dimensional environment information gathering and reconfiguration system and side
Method.
Background technology
Traditional environment information acquisition and reconstructing method are typically obtained by artificial mapping or simple two-dimensional laser sensor
, it is the most complete, strongly professional to there is information in its result, the shortcoming of the inadequate hommization of man-machine interaction.Along with computer skill
Art and the development of numerical map technology and universal, the three-dimensional scenic of high fidelity also will gradually enter into our life.Three-dimensional ring
The information gathering in border and reconstruct to geographical mapping, military detection, robot and unmanned plane application, environmental structure monitoring, disaster and
The extraordinary application important in inhibiting such as scene of the accident rescue.The information gathering of three-dimensional environment will produce huge warp with reconfiguration technique
Ji and social benefit.
In terms of three-dimensional environment information gathering and reconstructing arrangement, the type of current main-stream has:
(1) with product Velodyne HDL 64E(U.S. Patent number 7969558) face battle array laser as representative, it can obtain
Horizontal direction 360 degree, the depth data of about 26 degree narrowband region scopes of vertical direction, this kind of systematic observation is limited in scope, only
It is applicable to open smooth scene, and does not comprise true picture information.
(2) with product Leica ScanStation C10 and Riegl as representative based on single scanning laser range finder and
The degree of depth-the visual system of single camera, scanning laser range finder and camera together rotate, scanning three-dimensional environment.Its horizontal direction
Observed efficiency is relatively low, and acquisition time is slower.And need to overcome image blurring problem owing to camera moves in not stall, camera is set
Standby requirement height.Similar domestic patent has: patent No. CN201562075U.Only it is prone to scanning laser range finder and carries out 3-D scanning
Domestic patent have: patent No. CN302466240S, patent No. CN102393516A, patent No. CN201858962U, the patent No.
CN1970894A。
(3) degree of depth-visual system based on structure light with product Microsoft Kinect as representative, it utilizes structure
Light principle measures environment depth map, and is equipped with color camera increase colouring information.Visual angle is less and the having of depth data in its observation
Effect scope is only between 0.5~4 meter.Similar domestic patent has: patent No. CN103162643A, the patent No.
CN201583258U。
(4) the pure visual system with patent No. CN202875336U as representative: use photographing unit or video camera for substantially to set
Standby, utilize various visual angles reconfiguration principle to carry out three-dimensional environment reconstruction.Its visual angle is less, depth information precision the highest and easy exist sky
Hole, for the environmental structure that color is homogeneous, error probability is higher.Similar domestic patent also has: patent No. CN101726257A.
In sum, existing equipment there is no one at aspects such as observation scope, work efficiency, reconstruction accuracy, real-textures
Individual efficient system can realize whole functions.The present invention proposes two-dimensional laser based on panorama camera and rotation based on this and surveys
The three-dimensional environment information gathering reconfiguration system of distance meter data, can realize three-dimensional environment information gathering and the weight of full field range around
Structure, system is simple to operate, and operational efficiency is high, can the most truly reduce the three-dimensional scenic of surrounding.
Summary of the invention
It is an object of the invention to overcome the deficiencies in the prior art, it is provided that a kind of three-dimensional environment information gathering and reconfiguration system and
Method.
Three-dimensional environment information gathering and reconfiguration system include scanning laser range finder, panorama camera, rotation platform, airborne place
Reason device, power supply, base, wire slide rail, signal picker, motor, train of reduction gears, panorama camera bracing frame, connecting rod;
Base is provided with connecting rod, airborne processor, power supply, is the most sequentially cased with train of reduction gears, leads on connecting rod
Line slide rail, rotation platform, connecting rod top is the most sequentially provided with panorama camera bracing frame, panorama camera, rotation platform following table
Face is fixed with signal picker, and rotation platform upper surface is fixed with one or more scanning laser range finder, and two-dimensional laser is found range
Instrument is positioned at the lower section of panorama camera bracing frame, and motor is located in train of reduction gears.
The internal hardware module annexation of described airborne processor is:
Carry internal memory, network controller, CAN controller, serial ports, video in the pci bus being connected with central processing unit
Harvester, network controller provides radio network interface and the wired network interface of RJ45, and CAN controller passes through CAN carry motor
Driver, serial ports passes through 232 bus carry code-disc signal pickers.
Three-dimensional environment information gathering is as follows with the step of reconstructing method with the three-dimensional environment information gathering of reconfiguration system:
1) equipment clock synchronizes: synchronized by the clock of one or more scanning laser range finder, panorama camera, rotatable platform
Under airborne processor clock, and carry out online correction in real time, so that distinct device can be just in the data gathered the most in the same time
The most corresponding;
2) laser range finder-camera relative pose is demarcated: demarcate one or more scanning laser range finder, rotation platform,
Relative pose between panorama camera, and online correction in real time, so that the data of scanning laser range finder collection and panorama camera
It is corresponding that the image gathered can carry out correct locus;
3) single frames point cloud strengthens: the some cloud collecting scanning laser range finder carries out resampling to remove noise and to make
Point more uniformly spreads, and carries out surface grid model structure, and maps image texture, obtains the surface mesh with color texture
Model;
4) many frame data splicing: the surface grid model collected by diverse location splices, from the office at each frame place
Portion's coordinate system is transformed under unified global coordinate system, to recover the relative pose relation between many frame data;
5) multiframe surface is merged: the surface grid model collected by the diverse location unified under global coordinate system
Carry out surface fusion, the geometry of the model lap to still suffer from after overcoming the splicing of that caused and many frame data by noise
Misfit, obtain the view picture environmental surfaces grid model of geometrically the most consistent with in data structure band color texture.
Described step 1) is: the data of scanning laser range finder are read in by laser data read module, panorama camera
Image read in by image reading module, code-disc data code-disc data read module that rotation platform passes through reads in, clock with
Request of data timestamp and the data receipt time of correspondence that step module is sent by system stab, and utilize and stab based on packet time
Network equipment line clock generator synchronous method, the timestamp of each packet is corresponded under system clock, it is achieved multiple equipment
The clock of data synchronizes, and the data after these update of time stamp are stored in shared drive successively by data/address bus, and clock is same
Step module is issued clock by messaging bus simultaneously and is synchronously completed signal and the address of up-to-date synchrodata.
Described step 2) be: the clock that first mechanical parameter estimation module is issued according to clock synchronization module synchronously completes
Signal and up-to-date synchrodata address, read in the laser data after the data after update of time stamp, i.e. clock synchronize, code-disc number
According to, image;Laser data stream interval is chosen module and is determined that corresponding laser data is interval according to the timestamp of image;Turntable attitude
Estimation module is according to the timestamp sequence in the laser data interval selected, in conjunction with the angle-data in code-disc dataAnd
Correspondent time, calculate attitude sequence,
,
WhereinWithBe on time shaft fromTwo nearest code-disc data time stamp;Laser data stream coordinate transform mould
The laser data selected circulation is changed under unified three-dimensional system of coordinate by tuber attitude sequence accordingly, forms a frame point cloud;Point cloud
Extraction module to this frame data reduction edge contour, obtains representing the three-dimensional point set of three-dimensional edges immediately, and according to Laser Measuring
The performance parameter of distance meter, error model and the error model of some cloud Boundary extracting algorithm, the probability obtaining three-dimensional edges point set divides
Cloth;Meanwhile, Edge extraction module extracts the edge contour of image, obtains representing the collection of pixels at two dimension edge
, and according to performance parameter, error model and the error model of Edge extraction algorithm of camera, obtain two dimension edge pixel
Probability distribution;Relative attitude and displacement estimation module includes spin matrix with one groupAnd translation matrixCoordinate conversion matrix be sharp
Optar and the relative pose of camera, by three-dimensional point setProject under camera coordinates system, obtain two dimension edge point set, and
Probability distribution according to three-dimensional edges point setTwo dimension edge point set is obtained with projection relationProbability distribution, by meter
Calculate two dimension edge point setProbability distributionProbability distribution with two dimension edge pixelBetween symmetrical KL distance,
With、For parameter, minimize, the laser range finder trying to achieve optimum changes square with panorama camera relative pose
Battle array,
。
After mechanical parameter estimation module is calculated the relative pose transition matrix of scanning laser range finder and panorama camera,
By data/address bus, this relative pose parameter, some cloud edge contour data, image border outline data are stored into shared drive
In, and estimated signal and up-to-date relative pose parameter, some cloud edge contour number by messaging bus issue mechanical parameter
According to, the storage address of image border outline data.
Described step 3) is: the clock that first some cloud resampling module is issued according to clock synchronization module synchronously completes letter
Number, the mechanical parameter issued of mechanical parameter estimation module has estimated signal, reading laser data, utilize region growing algorithm,
With surface single order indicated by a cloud, continuously and Second Order Continuous is as condition, splits a cloud, obtains some subsets of a cloud, profit
Individually each subset of point cloud segmentation is carried out resampling with Moving Least Squares algorithm, by the point after all resamplings after completing
Cloud subset is merged into new some cloud C ', and is stored in shared drive, the some cloud after resampling by disappearing by data/address bus
Breath bus publishing point cloud resampling completes the storage address of the some cloud after signal and resampling;Then, surface mesh generates mould
The point cloud resampling that tuber strong point cloud resampling module is issued completes signal, utilizes two-dimentional Triangulation Algorithm based on partial projection
Build the triangle grid model T of some cloud C ', by data/address bus, the triangle grid model obtained is stored into after completing in sharing
Depositing, and generated signal and triangle grid model storage address by messaging bus publishing table surface grids, finally, texture maps
The surface mesh that module is issued according to surface mesh generation module has generated signal, by summit each in triangle grid model T
Project under camera coordinates system Coor_C, find the image pixel of this vertex correspondence, for each three in triangle grid model
Edged surface, according to the image pixel corresponding to each summit of this triangular facet, the image after cutting is also mapped to triangle by cutting image
On grid, obtain the triangle grid model with color texture, the triangle with color texture that will be obtained by data/address bus
Data texturing in grid model is stored in shared drive, and has mapped signal by messaging bus issue texture.
Described step 4) is: the texture that multiframe concatenation module is issued according to texture mapping block has mapped signal, reads
Take a new frame and there is the triangle grid model of color texture and the some cloud edge point set of correspondence, with the some cloud marginal point of a new frame
Collection carries out three-dimensional curve with the intersection of existing history point cloud edge point set and mates, and is had by a new frame after obtaining pose transformation matrices
The triangle grid model of chromatic colour texture transforms under global coordinate system, stores having under global coordinate system by data/address bus
The triangle gridding surface model of color texture, and spliced under signal and global coordinate system by messaging bus issue multiframe
There is the storage address of the triangle gridding surface model of color texture.
Described step 5) is: the multiframe that multiframe surface Fusion Module is issued according to multiframe concatenation module has spliced letter
Number, read the triangle gridding surface model with color texture under the global coordinate system of a new frame, special according to sensor device
Property and data acquisition modes estimate each triangle gridding vertex position in triangle gridding surface model dataUncertainty;Newly
The model and the existing history block mold that read constitute two frame data, for each triangle gridding top in these two frame data
Point, assesses this place, summit local surfaces with neighbour's situation of another frame to confirm that two frame data are the most overlapping in this position, and
All triangle gridding summits being in overlapping region in labelling two frame data;To each triangle gridding summit being in overlapping region, search for triangle gridding summitAll adjacent triangular facet in these frame data and one of the arest neighbors in another frame
Corresponding triangular facet, according to the locational uncertainty on these triangular facet summits, estimates triangle gridding summitAt this regional area
" hidden surface " upper projection,
,
Wherein、It is respectively triangle gridding summitThe normal direction of the adjacent triangular facet of some in these frame data
Amount and the variance of plane intercept,、It is respectively triangle gridding summitAn adjacent triangle in another frame data
The normal vector in face and the variance of plane intercept;Reconnect the summit after the reorientation of overlapping region, form the most consistent triangle
Grid model, the newest block mold;By data/address bus, newly-increased summit and annexation are inserted the storage of block mold
Region, and update existing triangle grid model vertex position and summit annexation, issue multiframe finally by messaging bus
Merge signal.
The present invention compared with prior art, has the advantages that
1. measure visual angle big, 360 °, horizontal view angle, 135 °, vertical angle of view, far measuring distance simultaneously, reach 50m, hence with
The efficiency that the present invention carries out environment information acquisition and structure is high;
2. information is accurate, abundant, and laser data provides accurate geological information, and panoramic picture provides abundant color
And texture information, the existing good view effect of model after both fusions, it is provided that again abundant environment raw information.
3. advantage of lower cost.With with type compared with the equipment of class, the present invention utilizes components and parts by maximized
Performance, achieves systemic-function with limited cost.
Accompanying drawing explanation
Fig. 1 is three-dimensional environment information gathering and reconfiguration system structural representation;
Fig. 2 is three-dimensional environment information gathering and reconfiguration system side view;
Fig. 3 is airborne processor internal hardware module logic connection figure;
Fig. 4 is airborne processor software module logic connection figure;
Fig. 5 is mechanical parameter estimation module internal structure figure;
Fig. 6 is three-dimensional environment information gathering and reconstructing method flow chart;
Fig. 7 is the single frames colour point clouds sample that three-dimensional environment information gathering builds with method with reconfiguration system;
Fig. 8 is the single frames band color texture polygonal mesh table that three-dimensional environment information gathering builds with method with reconfiguration system
Surface model sample;
Fig. 9 is the band color texture triangulation network of the multiframe fusion that three-dimensional environment information gathering builds with reconfiguration system and method
Lattice surface model sample;
In figure, at the bottom of scanning laser range finder 1, panorama camera 2, rotation platform 3, airborne processor 4, power supply 5, band connecting rod
Seat 6, wire slide rail 7, signal picker 8, motor 9, train of reduction gears 10, panorama camera bracing frame 11, connecting rod 12.
Detailed description of the invention
As shown in Figure 1, 2, three-dimensional environment information gathering and reconfiguration system include scanning laser range finder 1, panorama camera 2,
Rotation platform 3, airborne processor 4, power supply 5, base 6, wire slide rail 7, signal picker 8, motor 9, train of reduction gears 10,
Panorama camera bracing frame 11, connecting rod 12;
Base 6 is provided with connecting rod 12, airborne processor 4, power supply 5, is the most sequentially cased with reducing gear on connecting rod 12
Wheels 10, wire slide rail 7, rotation platform 3, connecting rod 12 top is the most sequentially provided with panorama camera bracing frame 11, panorama phase
Machine 2, rotation platform 3 lower surface is fixed with signal picker 8, and rotation platform 3 upper surface is fixed with one or more two-dimensional laser
Diastimeter 1, scanning laser range finder 1 is positioned at the lower section of panorama camera bracing frame 11, and motor 9 is located in train of reduction gears 10.
As it is shown on figure 3, the internal hardware module annexation of described airborne processor 4 is:
Carry internal memory, network controller, CAN controller, serial ports, video in the pci bus being connected with central processing unit
Harvester, network controller provides radio network interface and the wired network interface of RJ45, and CAN controller passes through CAN carry motor
Driver, serial ports passes through 232 bus carry code-disc signal pickers.
A kind of typical case's application of the present invention is the threedimensional model of automatic constructing environment, and this threedimensional model is with band color texture
Triangle gridding surface model form output and display, the triangle gridding surface model of the band color texture of input as shown in Figure 8,
Can accurately represent environment geometry, and fine surface color texture is provided, user therefore can be allowed easily to scene
Carry out observing, distinguish and measuring.
During structure, three-dimensional environment information gathering of the present invention is adopted at the diverse location of environment successively with reconfiguration system
Collection data, manually can move to the next position by this system after each place has gathered and be acquired, it is possible to by this system
It is arranged on the mobile platforms such as automobile, allows system be acquired in different positions.When every new position is acquired,
Laser range finder, panorama camera, the code-disc of rotation platform gather data respectively.
Three-dimensional environment information gathering is as follows with the step of reconstructing method with the three-dimensional environment information gathering of reconfiguration system:
1) equipment clock synchronizes: by one or more scanning laser range finder 1, panorama camera 2, the clock of rotatable platform 3
It is synchronized under airborne processor 4 clock, and carries out online correction in real time, so that distinct device is at the data energy gathered the most in the same time
Enough correct corresponding;
2) laser range finder-camera relative pose is demarcated: demarcate one or more scanning laser range finder 1, rotation platform
3, the relative pose between panorama camera 2, and online correction in real time, so that the data of scanning laser range finder 1 collection and panorama
It is corresponding that the image that camera 2 gathers can carry out correct locus;
3) single frames point cloud strengthens: the some cloud collecting scanning laser range finder 1 carries out resampling to remove noise also
Make a little to more uniformly spread, carry out surface grid model structure, and map image texture, obtain the surface mesh with color texture
Lattice model;
4) many frame data splicing: the surface grid model collected by diverse location splices, from the office at each frame place
Portion's coordinate system is transformed under unified global coordinate system, to recover the relative pose relation between many frame data;
5) multiframe surface is merged: the surface grid model collected by the diverse location unified under global coordinate system
Carry out surface fusion, the geometry of the model lap to still suffer from after overcoming the splicing of that caused and many frame data by noise
Misfit, obtain the view picture environmental surfaces grid model of geometrically the most consistent with in data structure band color texture, institute
The view picture environmental surfaces grid model of the band color texture obtained is as shown in Figure 9.
Described step 1) is: the data of scanning laser range finder 1 are read in by laser data read module, panorama camera
The image of 2 is read in by image reading module, and the code-disc data code-disc data read module that rotation platform 3 passes through reads in, clock
Request of data timestamp and the data receipt time of correspondence that synchronization module is sent by system stab, and utilize based on packet time
Network equipment line clock generator synchronous method (the TICSync:Knowing When Things Happened.IEEE of stamp
International Conference on Robotics and Automation, 2011), by the time of each packet
Stamp corresponds under system clock, it is achieved the clock of multiple device datas synchronizes, and the data after these update of time stamp pass through data
Bus is stored in shared drive successively, and clock synchronization module is issued clock by messaging bus simultaneously and synchronously completed signal and
The address of new synchrodata.
Described step 2) be: the clock that first mechanical parameter estimation module is issued according to clock synchronization module synchronously completes
Signal and up-to-date synchrodata address, read in the laser data after the data after update of time stamp, i.e. clock synchronize, code-disc number
According to, image;Laser data stream interval is chosen module and is determined that corresponding laser data is interval according to the timestamp of image;Turntable attitude
Estimation module is according to the timestamp sequence in the laser data interval selected, in conjunction with the angle-data in code-disc dataAnd
Correspondent time, calculate attitude sequence,
,
WhereinWithBe on time shaft fromTwo nearest code-disc data time stamp;Laser data stream coordinate transform mould
The laser data selected circulation is changed under unified three-dimensional system of coordinate by tuber attitude sequence accordingly, forms a frame point cloud;Point
Cloud extraction module to this frame data reduction edge contour, obtains representing the three-dimensional point set of three-dimensional edges immediately, and according to laser
The performance parameter of diastimeter, error model and the error model of some cloud Boundary extracting algorithm, obtain the probability of three-dimensional edges point set
Distribution;Meanwhile, Edge extraction module extracts edge contour (the Canny J. A computational of image
approach to edge detection[J]. Pattern Analysis and Machine Intelligence,
IEEE Transactions on, 1986 (6): 679-698.), obtain representing the collection of pixels at two dimension edge, and according to
The performance parameter of camera, error model and the error model of Edge extraction algorithm, the probability obtaining two dimension edge pixel divides
Cloth;Relative attitude and displacement estimation module includes spin matrix with one groupAnd translation matrixCoordinate conversion matrix be laser ranging
Instrument and the relative pose of camera, by three-dimensional point setProject under camera coordinates system, obtain two dimension edge point set, and according to three
The probability distribution of dimension edge point setTwo dimension edge point set is obtained with projection relationProbability distribution, by calculating two dimension
Edge point setProbability distributionProbability distribution with two dimension edge pixelBetween symmetrical KL distance,
With、For parameter, minimize, the laser range finder trying to achieve optimum changes square with panorama camera relative pose
Battle array,
。
Mechanical parameter estimation module is calculated the relative pose transition matrix of scanning laser range finder 1 and panorama camera 2
After, by data/address bus, this relative pose parameter, some cloud edge contour data, image border outline data are stored in sharing
In depositing, and signal and up-to-date relative pose parameter, some cloud edge contour are estimated by messaging bus issue mechanical parameter
Data, the storage address of image border outline data.
Described step 3) is: the clock that first some cloud resampling module is issued according to clock synchronization module synchronously completes letter
Number, the mechanical parameter issued of mechanical parameter estimation module has estimated signal, reading laser data, utilize region growing algorithm
(Adams R, Bischof L. Seeded region growing. Pattern Analysis and Machine
Intelligence, IEEE Transactions on, 1994,16 (6): 641-647), with surface one indicated by a cloud
Rank are continuously and Second Order Continuous is condition, split a cloud, obtain some subsets of a cloud, utilize Moving Least Squares algorithm
Individually each subset of point cloud segmentation is carried out resampling, after completing, the some cloud subset after all resamplings is merged into new point
Cloud C ', and by data/address bus, the some cloud after resampling is stored in shared drive, heavily adopted by messaging bus publishing point cloud
Sample completes the storage address of the some cloud after signal and resampling;Then, surface mesh generation module is according to a cloud resampling mould
The point cloud resampling that block is issued completes signal, utilizes two-dimentional Triangulation Algorithm (Hardwick J C. based on partial projection
Nested parallel 2D Delaunay triangulation method: U.S. Patent 6,088,511[P].
2000-7-11) build the triangle grid model T of some cloud C ', by data/address bus, the triangle grid model obtained is stored up after completing
It is stored to shared drive, and generates signal and triangle grid model storage address by messaging bus publishing table surface grids,
After, the surface mesh that texture mapping block is issued according to surface mesh generation module has generated signal, by triangle grid model T
In each summit project to, under camera coordinates system Coor_C, find the image pixel of this vertex correspondence, for triangle grid model
In each triangular facet, according to the image pixel corresponding to each summit of this triangular facet, cutting image by the image after cutting
It is mapped on triangle gridding, obtains the triangle grid model with color texture, by obtain, there is colour by data/address bus
Data texturing in the triangle grid model of texture is stored in shared drive, and has been mapped by messaging bus issue texture
Signal.
Described step 4) is: the texture that multiframe concatenation module is issued according to texture mapping block has mapped signal, reads
Take a new frame and there is the triangle grid model of color texture and the some cloud edge point set of correspondence, with the some cloud marginal point of a new frame
Collection carries out three-dimensional curve with the intersection of existing history point cloud edge point set and mates, and is had by a new frame after obtaining pose transformation matrices
The triangle grid model of chromatic colour texture transforms under global coordinate system, stores having under global coordinate system by data/address bus
The triangle gridding surface model of color texture, and spliced under signal and global coordinate system by messaging bus issue multiframe
There is the storage address of the triangle gridding surface model of color texture.
Described step 5) is: the multiframe that multiframe surface Fusion Module is issued according to multiframe concatenation module has spliced letter
Number, read the triangle gridding surface model with color texture under the global coordinate system of a new frame, special according to sensor device
Property and data acquisition modes estimate each triangle gridding vertex position in triangle gridding surface model dataUncertainty;Newly
The model and the existing history block mold that read constitute two frame data, for each triangle gridding top in these two frame data
Point, assesses this place, summit local surfaces with neighbour's situation of another frame to confirm that two frame data are the most overlapping in this position, and
All triangle gridding summits being in overlapping region in labelling two frame data;To each triangle gridding summit being in overlapping region, search for triangle gridding summitAll adjacent triangular facet in these frame data and one of the arest neighbors in another frame
Corresponding triangular facet, according to the locational uncertainty on these triangular facet summits, estimates triangle gridding summitAt this regional area
" hidden surface " upper projection,
,
Wherein、It is respectively triangle gridding summitThe normal direction of the adjacent triangular facet of some in these frame data
Amount and the variance of plane intercept,、It is respectively triangle gridding summitAn adjacent triangle in another frame data
The normal vector in face and the variance of plane intercept;Reconnect the summit after the reorientation of overlapping region, form the most consistent triangle
Grid model, the newest block mold;By data/address bus, newly-increased summit and annexation are inserted the storage of block mold
Region, and update existing triangle grid model vertex position and summit annexation, issue multiframe finally by messaging bus
Merge signal.
Claims (8)
1. a three-dimensional environment information gathering and reconfiguration system, it is characterised in that include scanning laser range finder (1), panorama camera
(2), rotation platform (3), airborne processor (4), power supply (5), base (6), wire slide rail (7), signal picker (8), motor
(9), train of reduction gears (10), panorama camera bracing frame (11), connecting rod (12);
Base (6) is provided with connecting rod (12), airborne processor (4), power supply (5), is the most sequentially cased with on connecting rod (12)
Train of reduction gears (10), wire slide rail (7), rotation platform (3), connecting rod (12) top is the most sequentially provided with panorama camera and props up
Support (11), panorama camera (2), rotation platform (3) lower surface is fixed with signal picker (8), and rotation platform (3) upper surface is solid
Surely one or more scanning laser range finder (1), scanning laser range finder (1) is had to be positioned under panorama camera bracing frame (11)
Side, motor (9) is located in train of reduction gears (10).
A kind of three-dimensional environment information gathering the most according to claim 1 and reconfiguration system, it is characterised in that described is airborne
The internal hardware module annexation of processor (4) is:
Carry internal memory, network controller, CAN controller, serial ports, video acquisition in the pci bus being connected with central processing unit
Device, network controller provides radio network interface and the wired network interface of RJ45, and CAN controller is driven by CAN carry motor
Device, serial ports passes through 232 bus carry code-disc signal pickers.
3. one kind uses three-dimensional environment information gathering as claimed in claim 1 and the three-dimensional environment information gathering of reconfiguration system with heavy
Structure method, it is characterised in that its step is as follows:
1) equipment clock synchronize: by one or more scanning laser range finder (1), panorama camera (2), rotatable platform (3) time
Clock is synchronized under airborne processor (4) clock, and carries out online correction in real time, so that distinct device is at the number gathered the most in the same time
According to can be correctly corresponding;
2) laser range finder-camera relative pose is demarcated: demarcate one or more scanning laser range finder (1), rotation platform
(3) relative pose, between panorama camera (2), and online revise in real time, so that the data that scanning laser range finder (1) gathers
Correct locus can be carried out corresponding with the image that panorama camera (2) gathers;
3) single frames point cloud strengthens: the some cloud collecting scanning laser range finder (1) carries out resampling to remove noise and to make
Point more uniformly spreads, and carries out surface grid model structure, and maps image texture, obtains the surface mesh with color texture
Model;
4) many frame data splicing: the surface grid model collected by diverse location splices, sits from the local at each frame place
Mark system is transformed under unified global coordinate system, to recover the relative pose relation between many frame data;
5) multiframe surface is merged: the surface grid model collected by the diverse location unified under global coordinate system is carried out
Surface is merged, and the geometry of the model lap to still suffer from after overcoming the splicing of that caused and many frame data by noise is not kissed
Close, obtain the view picture environmental surfaces grid model of geometrically the most consistent with in data structure band color texture.
A kind of three-dimensional environment information gathering and reconstructing method, it is characterised in that described step 1)
For: the data of scanning laser range finder (1) are read in by laser data read module, and the image of panorama camera (2) passes through image
Read module reads in, and rotation platform (3) is read in by code-disc data read module, the number that clock synchronization module is sent by system
According to the data receipt time stamp of request time stamp and correspondence, network equipment line clock generator based on packet time stamp is utilized to synchronize
Method, corresponds under system clock by the timestamp of each packet, it is achieved the clock of multiple device datas synchronizes, these times
Data after stamp updates are stored in shared drive successively by data/address bus, and clock synchronization module is sent out by messaging bus simultaneously
Cloth clock synchronously completes signal and the address of up-to-date synchrodata.
A kind of three-dimensional environment information gathering and reconstructing method, it is characterised in that described step 2)
For: the clock that first mechanical parameter estimation module is issued according to clock synchronization module synchronously completes signal and up-to-date synchrodata ground
Location, reads in the laser data after the data after update of time stamp, i.e. clock synchronize, code-disc data, image;Laser data stream is interval
Choose module and determine that corresponding laser data is interval according to the timestamp of image;Turntable Attitude estimation module is according to swashing of selecting
Timestamp sequence { the t of light data intervali, in conjunction with the angle-data { θ in code-disc datajAnd correspondent time { sj, calculate
Attitude sequence { Ai,
Wherein sjAnd sj-1It is from t on time shaftiTwo nearest code-disc data time stamp;Laser data stream coordinate transformation module root
The laser data selected circulation is changed under unified three-dimensional system of coordinate by attitude sequence accordingly, forms a frame point cloud;Point cloud carries
Delivery block to this frame data reduction edge contour, obtains representing the three-dimensional point set L of three-dimensional edges immediately3, and according to laser ranging
The performance parameter of instrument, error model and the error model of some cloud Boundary extracting algorithm, obtain the probability distribution of three-dimensional edges point setMeanwhile, Edge extraction module extracts the edge contour of image, obtains representing collection of pixels C at two dimension edge2,
And according to performance parameter, error model and the error model of Edge extraction algorithm of camera, obtain two dimension edge pixel
Probability distributionRelative attitude and displacement estimation module is with one group of coordinate conversion matrix including spin matrix R and translation matrix T as laser
Diastimeter and the relative pose of camera, by three-dimensional point set L3Project under camera coordinates system, obtain two dimension edge point set L2, and root
Probability distribution according to three-dimensional edges point setTwo dimension edge point set L is obtained with projection relation2Probability distributionBy calculating
Two dimension edge point set L2Probability distributionProbability distribution with two dimension edge pixelBetween symmetrical KL distance KLLC,
With R, T as parameter, minimize KLLC, try to achieve optimum laser range finder and panorama camera relative pose transition matrix (R,
T), (R, T)=argmin(R, T)(KLLC),
Mechanical parameter estimation module is calculated the relative pose transition matrix of scanning laser range finder (1) and panorama camera (2)
After, by data/address bus, this relative pose parameter, some cloud edge contour data, image border outline data are stored in sharing
In depositing, and signal and up-to-date relative pose parameter, some cloud edge contour are estimated by messaging bus issue mechanical parameter
Data, the storage address of image border outline data.
A kind of three-dimensional environment information gathering and reconstructing method, it is characterised in that described step 3)
For: first some cloud resampling module synchronously completes signal, mechanical parameter estimation module according to the clock that clock synchronization module is issued
The mechanical parameter issued has estimated signal, reads laser data, utilizes region growing algorithm, with surface single order indicated by a cloud
Continuous and Second Order Continuous is condition, splits a cloud, obtains some subsets of a cloud, utilizes Moving Least Squares algorithm list
Solely each subset of point cloud segmentation is carried out resampling, after completing, the some cloud subset after all resamplings is merged into new some cloud
C ', and by data/address bus, the some cloud after resampling is stored in shared drive, by messaging bus publishing point cloud resampling
Complete the storage address of the some cloud after signal and resampling;Then, surface mesh generation module is according to a cloud resampling module
The point cloud resampling issued completes signal, utilizes two-dimentional Triangulation Algorithm based on partial projection to build the triangle gridding of some cloud C '
Model T, is stored into shared drive by data/address bus by the triangle grid model obtained after completing, and is issued by messaging bus
Surface mesh has generated signal and triangle grid model stores address, and finally, texture mapping block generates according to surface mesh
The surface mesh that module is issued has generated signal, and summit each in triangle grid model T is projected to camera coordinates system Coor_
Under C, find the image pixel of this vertex correspondence, for each triangular facet in triangle grid model, each according to this triangular facet
Image pixel corresponding to summit, the image after cutting is also mapped on triangle gridding, obtains having colored stricture of vagina by cutting image
The triangle grid model of reason, is stored up by the data texturing in the triangle grid model with color texture that data/address bus will obtain
It is stored in shared drive, and has mapped signal by messaging bus issue texture.
A kind of three-dimensional environment information gathering and reconstructing method, it is characterised in that described step 4)
For: the texture that multiframe concatenation module is issued according to texture mapping block has mapped signal, reads a new frame and has color texture
Triangle grid model and the some cloud edge point set of correspondence, with the some cloud edge point set of a new frame and existing history point cloud limit
The intersection of edge point set carries out three-dimensional curve coupling, and a new frame has after obtaining pose transformation matrices the triangle gridding of color texture
Model transforms under global coordinate system, stores the triangle gridding table with color texture under global coordinate system by data/address bus
Surface model, and spliced the triangulation network with color texture under signal and global coordinate system by messaging bus issue multiframe
The storage address of lattice surface model.
A kind of three-dimensional environment information gathering and reconstructing method, it is characterised in that described step 5)
For: the multiframe that multiframe surface Fusion Module is issued according to multiframe concatenation module has spliced signal, reads the overall situation seat of a new frame
The triangle gridding surface model with color texture under mark system, estimates three according to sensor device characteristic and data acquisition modes
Each triangle gridding vertex position P in the surface mesh model data of anglesUncertainty;The new model read and existing history
Block mold constitutes two frame data, for each triangle gridding summit in these two frame data, assesses this local, place, summit table
Face with neighbour's situation of another frame to confirm that two frame data are the most overlapping in this position, and in labelling two frame data all be in weight
The triangle gridding summit in folded region;To each triangle gridding summit P being in overlapping regions, search for triangle gridding summit PsAt this
All adjacent triangular facet in frame data and a corresponding triangular facet of the arest neighbors in another frame, according to these triangular facets
The locational uncertainty on summit, estimates triangle gridding summit PsIn " the hidden surface " of regional area upper projection
Wherein nASB、It is respectively triangle gridding summit PsThe normal vector peace of the adjacent triangular facet of some in these frame data
The variance of face intercept, nUVW、It is respectively triangle gridding summit PsThe normal direction of an adjacent triangular facet in another frame data
Amount and the variance of plane intercept;Reconnect the summit after the reorientation of overlapping region, form the most consistent triangle grid model,
The newest block mold;By data/address bus by newly-increased summit and the storage area of annexation insertion block mold, and more
New existing triangle grid model vertex position and summit annexation, issue multiframe finally by messaging bus and merged letter
Number.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310742704.1A CN103729883B (en) | 2013-12-30 | 2013-12-30 | A kind of three-dimensional environment information gathering and reconfiguration system and method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310742704.1A CN103729883B (en) | 2013-12-30 | 2013-12-30 | A kind of three-dimensional environment information gathering and reconfiguration system and method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103729883A CN103729883A (en) | 2014-04-16 |
CN103729883B true CN103729883B (en) | 2016-08-24 |
Family
ID=50453944
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310742704.1A Active CN103729883B (en) | 2013-12-30 | 2013-12-30 | A kind of three-dimensional environment information gathering and reconfiguration system and method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103729883B (en) |
Families Citing this family (37)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105354812B (en) * | 2014-07-10 | 2020-10-16 | 北京中科盘古科技发展有限公司 | Multi-Kinect cooperation-based depth threshold segmentation algorithm contour recognition interaction method |
CN105336002B (en) * | 2014-08-01 | 2018-12-14 | 联想(北京)有限公司 | Information processing method and electronic equipment |
CN104656097B (en) * | 2015-01-28 | 2017-03-08 | 武汉理工大学 | Caliberating device based on rotary two-dimensional laser three-dimensional reconfiguration system and method |
US10297179B2 (en) * | 2015-02-03 | 2019-05-21 | Sony Corporation | Information processing apparatus, information processing method, and program |
CN104794743A (en) * | 2015-04-27 | 2015-07-22 | 武汉海达数云技术有限公司 | Color point cloud producing method of vehicle-mounted laser mobile measurement system |
CN106273442B (en) * | 2015-06-01 | 2018-06-08 | 三纬国际立体列印科技股份有限公司 | The positioning mechanism of 3D printer scanning means |
CN106705951A (en) * | 2015-11-13 | 2017-05-24 | 恩斯迈电子(深圳)有限公司 | Movable device |
US10372968B2 (en) * | 2016-01-22 | 2019-08-06 | Qualcomm Incorporated | Object-focused active three-dimensional reconstruction |
WO2017134882A1 (en) * | 2016-02-04 | 2017-08-10 | 富士フイルム株式会社 | Information processing device, information processing method, and program |
CN106990498B (en) * | 2016-05-26 | 2020-04-17 | 深圳市圆周率软件科技有限责任公司 | Lens position fine adjustment device of multi-lens panoramic camera |
CN106091976B (en) * | 2016-05-27 | 2017-07-25 | 武汉大学 | The automatic detection of cuboid and three-dimensional reconfiguration system and method |
CN106094886B (en) * | 2016-07-01 | 2019-08-06 | 中国科学院遥感与数字地球研究所 | The multi-angle observation device and method of remote sensing terrestrial positioning test |
CN106441109B (en) * | 2016-09-19 | 2018-10-30 | 北京机械设备研究所 | A kind of refraction-reflection laser ranging three-dimensional panorama imaging integrated apparatus |
CN106406018A (en) * | 2016-10-18 | 2017-02-15 | 安徽协创物联网技术有限公司 | Compact digital panoramic camera realized through lenses |
CN106428545B (en) * | 2016-10-31 | 2018-10-16 | 朱建萍 | A kind of three-dimensional geographic information acquisition system and its acquisition method |
CN106652028B (en) * | 2016-12-28 | 2020-07-03 | 深圳乐动机器人有限公司 | Environment three-dimensional mapping method and device |
CN106954012A (en) * | 2017-03-29 | 2017-07-14 | 武汉嫦娥医学抗衰机器人股份有限公司 | A kind of high definition polyphaser full-view stereo imaging system and method |
CN107610213A (en) * | 2017-08-04 | 2018-01-19 | 深圳市为美科技发展有限公司 | A kind of three-dimensional modeling method and system based on panorama camera |
CN109215111B (en) * | 2017-12-19 | 2023-04-07 | 上海亦我信息技术有限公司 | Indoor scene three-dimensional modeling method based on laser range finder |
US10764562B2 (en) | 2018-01-28 | 2020-09-01 | Eys3D Microelectronics, Co. | Depth generation system with adjustable light intensity |
US20190387165A1 (en) | 2018-06-07 | 2019-12-19 | Eys3D Microelectronics, Co. | Image device for generating depth images and related electronic device |
EP3842885A4 (en) * | 2018-08-22 | 2021-11-17 | Ecovacs Robotics Co., Ltd. | Autonomous movement device, control method and storage medium |
CN109635834A (en) * | 2018-11-02 | 2019-04-16 | 中铁上海工程局集团有限公司 | A kind of method and system that grid model intelligence is inlayed |
CN109685891B (en) * | 2018-12-28 | 2023-09-01 | 鸿视线科技(北京)有限公司 | Building three-dimensional modeling and virtual scene generation method and system based on depth image |
CN110223226B (en) * | 2019-05-07 | 2021-01-15 | 中国农业大学 | Panoramic image splicing method and system |
CN110490921B (en) * | 2019-08-05 | 2022-06-24 | 合肥工业大学 | Three-dimensional GIS data acquisition and processing method |
CN110421581B (en) * | 2019-08-09 | 2022-06-14 | 南京芊玥机器人科技有限公司 | Active damping method of active damping robot based on curve fitting |
CN110505463A (en) * | 2019-08-23 | 2019-11-26 | 上海亦我信息技术有限公司 | Based on the real-time automatic 3D modeling method taken pictures |
JP7223449B2 (en) | 2019-08-23 | 2023-02-16 | 上海亦我信息技術有限公司 | 3D modeling system based on photography |
CN110929541B (en) * | 2019-11-18 | 2021-01-01 | 南京大树智能科技股份有限公司 | Panoramic cigarette two-dimensional code scanning system |
CN111145095B (en) * | 2019-12-25 | 2023-10-10 | 深圳市诺达通信技术有限公司 | VR (virtual reality) graph generation method with scale measurement and data acquisition device |
CN111238370A (en) * | 2020-02-20 | 2020-06-05 | 中国科学院声学研究所东海研究站 | Intelligent detection method and device for KIT board |
CN111784821B (en) | 2020-06-30 | 2023-03-14 | 北京市商汤科技开发有限公司 | Three-dimensional model generation method and device, computer equipment and storage medium |
CN112084853B (en) * | 2020-08-04 | 2023-12-29 | 深圳市优必选科技股份有限公司 | Footprint prediction method, footprint prediction device and humanoid robot |
CN112230618B (en) * | 2020-10-29 | 2021-10-15 | 中国人民解放军国防科技大学 | Method for automatically synthesizing multi-robot distributed controller from global task |
CN113099205A (en) * | 2021-04-12 | 2021-07-09 | 司岚光电科技(苏州)有限公司 | Image acquisition device for multi-camera collaborative scanning and three-dimensional map construction method |
CN117706544B (en) * | 2024-02-04 | 2024-04-09 | 新泰坦空气净化技术(北京)有限公司 | Intelligent environment-friendly remote monitoring system |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2012081995A1 (en) * | 2010-12-13 | 2012-06-21 | Surveylab Group Limited | Mobile measurement devices, instruments and methods |
CN102589465A (en) * | 2012-01-13 | 2012-07-18 | 河南科技大学 | Linear array camera based automatic panoramic acquisition system for outer surface of cylindrical surface |
CN202403666U (en) * | 2011-12-31 | 2012-08-29 | 武汉浩迪科技有限公司 | Laser and camera-shooting integral measuring instrument |
-
2013
- 2013-12-30 CN CN201310742704.1A patent/CN103729883B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2012081995A1 (en) * | 2010-12-13 | 2012-06-21 | Surveylab Group Limited | Mobile measurement devices, instruments and methods |
CN202403666U (en) * | 2011-12-31 | 2012-08-29 | 武汉浩迪科技有限公司 | Laser and camera-shooting integral measuring instrument |
CN102589465A (en) * | 2012-01-13 | 2012-07-18 | 河南科技大学 | Linear array camera based automatic panoramic acquisition system for outer surface of cylindrical surface |
Non-Patent Citations (3)
Title |
---|
三维激光扫描仪与数码相机联合标定方法研究;赵松;《中国优秀硕士学位论文全文数据库 信息科技辑》;20130615;I138-1164 * |
基于PCI总线的多功能卡的设计与实现;刘永波;《中国优秀硕士学位论文全文数据库 信息科技辑》;20070915;摘要、第2.2.2节PCI总线的特点、第2.2.5节USB2.0数据传输 * |
基于单目 视觉和激光测距仪的位姿测量算法;高学海 等;《仪器仪表学报》;20070831;第28卷(第8期);第1479-1485页 * |
Also Published As
Publication number | Publication date |
---|---|
CN103729883A (en) | 2014-04-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103729883B (en) | A kind of three-dimensional environment information gathering and reconfiguration system and method | |
CN111275750B (en) | Indoor space panoramic image generation method based on multi-sensor fusion | |
CN106443687B (en) | A kind of backpack mobile mapping system based on laser radar and panorama camera | |
CN104156972B (en) | Perspective imaging method based on laser scanning distance measuring instrument and multiple cameras | |
CN103198524B (en) | A kind of three-dimensional reconstruction method for large-scale outdoor scene | |
CN111597666B (en) | Method for applying BIM to transformer substation construction process | |
CN105931234A (en) | Ground three-dimensional laser scanning point cloud and image fusion and registration method | |
CN108053476B (en) | Human body parameter measuring system and method based on segmented three-dimensional reconstruction | |
CN101329771B (en) | Method for rapidly modeling of urban street base on image sequence | |
CN109000582A (en) | Scan method and system, storage medium, the equipment of tracking mode three-dimensional scanner | |
CN102645209B (en) | Joint positioning method for spatial points by means of onboard LiDAR point cloud and high resolution images | |
CN108168521A (en) | One kind realizes landscape three-dimensional visualization method based on unmanned plane | |
CN106826833A (en) | Independent navigation robot system based on 3D solid cognition technologies | |
CN104236479B (en) | A kind of line-structured light three-dimension measuring system and 3D texture image construction algorithm | |
CN104809759A (en) | Large-area unstructured three-dimensional scene modeling method based on small unmanned helicopter | |
CN104376596A (en) | Method for modeling and registering three-dimensional scene structures on basis of single image | |
CN111710040B (en) | High-precision map construction method, system, terminal and storage medium | |
Haala et al. | High density aerial image matching: State-of-the-art and future prospects | |
CN114812558B (en) | Monocular vision unmanned aerial vehicle autonomous positioning method combining laser ranging | |
Wang et al. | Terrainfusion: Real-time digital surface model reconstruction based on monocular slam | |
CN105844700A (en) | System for acquiring three-dimensional point clouds in outdoor scene | |
Zhou et al. | Application of UAV oblique photography in real scene 3d modeling | |
CN110443228A (en) | A kind of method for pedestrian matching, device, electronic equipment and storage medium | |
CN111612829B (en) | High-precision map construction method, system, terminal and storage medium | |
Weßling et al. | Structure from motion for systematic single surface documentation of archaeological excavations |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right | ||
TR01 | Transfer of patent right |
Effective date of registration: 20180724 Address after: 310052 208, room 6, 1197 Binan Road, Binjiang District, Hangzhou, Zhejiang. Patentee after: Which Hangzhou science and Technology Co Ltd Address before: 310027 No. 38, Zhejiang Road, Hangzhou, Zhejiang, Xihu District Patentee before: Zhejiang University |