CN107560593A - The empty three free net construction methods of special unmanned plane image based on minimum spanning tree - Google Patents
The empty three free net construction methods of special unmanned plane image based on minimum spanning tree Download PDFInfo
- Publication number
- CN107560593A CN107560593A CN201710747621.XA CN201710747621A CN107560593A CN 107560593 A CN107560593 A CN 107560593A CN 201710747621 A CN201710747621 A CN 201710747621A CN 107560593 A CN107560593 A CN 107560593A
- Authority
- CN
- China
- Prior art keywords
- air strips
- sub
- point
- summit
- net
- 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 Analysis (AREA)
- Image Processing (AREA)
- Navigation (AREA)
Abstract
The empty three free net construction methods of special unmanned plane image based on minimum spanning tree provided by the invention, for overboard, cloud block, the special unmanned plane image such as overlapping leak, when solving conventional automatic triangulation, there is relative orientation, the problems such as model connection failure, in the case where not providing POS assistance datas, automatically build the free net of sky three of whole region, obtain and survey the elements of exterior orientation of all images and the initial value of object space point coordinates in area, and carry out bundle block adjustment, the elements of exterior orientation of the region-wide interior every image of integrated solution and the space coordinates of all object spaces point.If being to provide a small amount of known control point coordinates, whole region can be made to bring into the earth axes of known control point, realize absolute orientation.Make the processing of special unmanned plane image is horizontal to be greatly enhanced with disposal ability.
Description
Technical field
It is more particularly to a kind of to be based on minimum spanning tree the present invention relates to a kind of empty three free net construction methods of unmanned plane image
The empty three free net construction methods of special unmanned plane image, belong to unmanned plane image treatment method technical field.
Background technology
Low-altitude remote sensing system is a kind of high maneuverability, the miniaturization of low cost, customizations remote sense monitoring system.It is with nobody
Driving aircraft is flying platform, using high-definition remote sensing equipment as airborne sensor, using obtain high-definition remote sensing data as
Application target, there is quick real-time investigation monitoring capability over the ground.As a kind of emerging remote sensing technology, its advantage be mainly reflected in
Under several aspects：First, being not required to airport (generally, it is not necessary to apply for spatial domain), lifting is flexibly, convenient implementation is all kinds of promptly appoints
Business；Second, equipment is easy to carry, and convenient transfer, mobility strong；Third, platform construction, maintenance and the cost of operation are extremely low；Four
It is to be influenceed by weather smaller, airflight is flexible, suitable for the sharp work of small area；Fifth, low latitude is photographed, image space
High resolution；Sixth, the image of high degree of overlapping can be obtained, be advantageous to improve the reliability and precision of follow-up data processing.
Unmanned plane low-altitude remote sensing system was developed rapidly and applied, particularly in woods because of its unique advantage in recent years
The side such as industry investigation, survey of territorial resources, Geological Hazards Monitoring, large scale mapping and map rejuvenation, cybercity construction
Face plays great function.
When unmanned plane carries out operation in hundreds of meters of low latitudes, because the lighter in weight of its own, airflow influence are big, make its
Aerial flight attitude is very unstable, and the image for causing to obtain has the problems such as larger rotation is inclined and degree of overlapping is inadequate, so as to cause
Aerial photographic gap, the post-processing of Unmanned Aerial Vehicle Data is influenceed, these problems are brought to automatic empty three processing of unmanned aerial vehicle remote sensing images
Following difficulty：First, the rotation drift angle of adjacent image is big, engineer's scale difference is big, reduces gray scale related success rate and reliability,
A series of problems, such as having influence on follow-up automatic relative orientation；Second, the line of flight is bent, ship's control and sidelapping degree
Irregularly, the extraction to tie point brings difficulty；Third, aerial photographic gap causes traditional air strips construction method failure, can not carry out
Survey the full-automatic aerial triangulation in area.
Aerial triangulation as a key technology in photogrammetric, be a series of follow-up photogrammetric products with should
Basis, such as create DTM, digital orthoimage (DOM), stereoplotting etc..For in general unmanned plane image (relative to this
For " special " image that invention is mentioned), according to photogrammetric empty three work flow of routine, carrying out block adjustment can obtain
The elements of exterior orientation of get Ce areas whole image and the geographical coordinates of pass point, and then carry out subsequent treatment and obtain required product
(such as DEM, DOM etc.).
But prior art can not also realize full-automatic, high efficiency and high-precision aerial triangulation.Such as falling
Water image, add tie point in standard point position by the way of artificial treatment or measure a number of connection along waters edge
Point, then perform relative orientation and can help to solve the problems, such as that sky three interrupts sometimes.The mode however, this most traditional craft is added some points
Because the drawbacks such as non intelligent, efficiency is low, labor intensive be present, the raising of productivity effect is seriously hindered.Moreover, for big
The overboard image of area, the method manually added some points sometimes can not also solve relative orientation, the situation of model connection failure.At this moment, only
Course line can be divided into several sections to skip overboard region.But traditional region division of sky three is manually drawn according to actual landform
Point and design, i.e. human eye by being browsed to every image, then judge which image possibly can not carry out Auto-matching,
Relative orientation or model connection etc..It is this it is traditional, often accuracy rate is not high in a manner of human eye subjective judgement comes zoning,
And easily there is erroneous judgement or omission.The division and modification repeatedly to surveying progress Zi Ce areas of area are needed, so as to considerably increase
The complexity and workload of operation, have also had a strong impact on production efficiency.
When prior art carries out island aerial triangulation using GPS/IMU positioning and directings technology, by being set in unmanned plane
The standby upper high-precision POS system of carrying directly to determine the elements of exterior orientation of photography moment overboard image, and carries out forward intersection
In the hope of island and the space coordinates of land bank image object space point.This method that direct geo-location is carried out using POS system can
Controlled with realizing to survey in area without ground, improve operating efficiency, this is also the most significant advantage of direct Directional Method.It is but this
Method is also difficult to the required precision for meeting large scale topographical map mapping, and lead at present because integrated approach is tight not enough
Often need to carry out strict systematic parameter calibration so as to add the complexity of operation.At present, due to the volume of UAS
General all smaller, load is lighter, and POS system progress low latitude photography is loaded in the flight operation of reality and certain difficulty be present.Very
GPS/IMU systems can not be carried to the unmanned plane low-altitude remote sensing system having, therefore also just can not directly obtain difficult region and (such as fall
Water, Desert Regions etc.) image elements of exterior orientation, this is also a frequently problem in actual applications.
The content of the invention
In view of the shortcomings of the prior art, the special unmanned plane image sky three provided by the invention based on minimum spanning tree is free
Net construction method, (overboard large area, extensive forest covering in the unmanned plane image to a kind of relatively " special " such as be present or have
The image etc. of cloud block, leak of photographing) in conventional aerial triangulation operation process is carried out because relative orientation or model connect
Connecing failure causes to automatically create air strips failure, the interruption of sky three, in the case where not providing POS assistance datas, automatically builds whole
The free net of sky three in individual region, obtain and survey the elements of exterior orientation of all images and the initial value of object space point coordinates in area, and carry out light
Beam method block adjustment, the elements of exterior orientation of the region-wide interior every image of integrated solution and the space coordinates of all object spaces point.If
It is to provide a small amount of known control point coordinates, then whole region can be made to bring into the earth axes of known control point, it is real
Existing absolute orientation.Make the processing of special unmanned plane image is horizontal to be greatly enhanced with disposal ability.
To reach above technique effect, the technical solution adopted in the present invention is as follows：
The empty three free net construction methods of special unmanned plane image based on minimum spanning tree, are carried out to special unmanned plane image
Aerial triangulation is handled, and includes three steps：
The first step, sub- air strips model is established, single three-dimensional model is established according to continuity method relative orientation to each sub- air strips, so
Model connection is carried out afterwards establishes sub- air strips free net；
Sub- air strips model is established according to ready-portioned each strip air strips, establishes sub- air strips model, including following several accordingly
Step：When measure picpointed coordinate and carry out system error correction；Second, using continuity method relative orientation, single three-dimensional mould is established
Type；Third, model connects, sub- air strips free net is established；
Second step, region free net is built, calculate the bonding strength size between adjacent sub- air strips first and establish corresponding
Sub- air strips graph of a relation, it is then based on effective connection side of the principle chosen area net of minimum spanning tree and finds optimal starting
Point, the coordinate system of all sub- air strips in region is come together finally by absolute orientation；
Structure region free net is connected sub- air strips all in area are surveyed by absolute orientation, turns into an entirety,
Including the following steps：First, the annexation definition between sub- air strips, i.e. weights define up and down；Second, closed according to the connection defined
System establishes the graph of a relation of whole sub- air strips, i.e. adjacency matrix；Third, the principle chosen area net based on minimum spanning tree is optimal
Side is connected, and finds the starting point that the optimal sub- air strips of starting increase as region, using its coordinate system as reference coordinate
System, the coordinate of other sub- air strips is brought into the coordinate system unified with originating sub- air strips by absolute orientation, completes whole survey
The free net of sky three structure in area；
3rd step, bundle block adjustment, using picture point, object space point and photo centre's three point on a straight line condition, integrally ask
The elements of exterior orientation of all images in Xie Ce areas and the tight solution of closed points coordinate, as the inspection to network forming result；
Bundle block adjustment includes the following steps：One is to determine every image elements of exterior orientation and fixed point coordinate
Approximation, each stereogram is subjected to relative orientation the free air strips net of structure is connected with model, utilize the control point in air strips
And the common point between adjacent course line carries out to course line elements of exterior orientation and pass point of the absolute orientation in the hope of each photo
Geographical coordinates, the approximation as unknown number；Second, the picpointed coordinate of control point and point to be located from every width image, according to
The collinearity condition equation of every photography light lists error equation；Third, point by point methodization establishes changing normal equation, by circulation
The method for solving of piecemeal, first seek the elements of exterior orientation of every width image；Fourth, the ground of point to be located is tried to achieve according to space intersection
Coordinate, its average is taken as last result for the common point of adjacent image.
The empty three free net construction methods of special unmanned plane image based on minimum spanning tree, further, establish sub- air strips
The specific method of model is：
1st step, measure picpointed coordinate and carry out system error correction；Measure each as the pass point picture to having selected in advance
Plane coordinates, and system error correction is carried out to it；
2nd step, using continuity method relative orientation, establish single three-dimensional model；With the left photo of each picture pair in sub- air strips
On the basis of, elements of relative orientation of the right photo relative to left photo is obtained, after the elements of relative orientation for calculating picture pair, before pressing
Square intersection obtains coordinate of the model points in respective image space auxiliary coordinates, establishes single three-dimensional model；
3rd step, model connection, establishes sub- air strips free net；Single model is connected to become to the premise of sub- air strips model is
The different engineer's scale of each model is classified as unified engineer's scale, with adjacent picture three in overlapping region are connected with point height is equal to be
Condition, from left to right established successively by the engineer's scale of the engineer's scale naturalization of latter model to previous model with first model
Engineer's scale on the basis of sub- air strips model, by the model point coordinates of all stereograms and take the photograph website coordinate and bring full boat into
With in unified coordinate system.
The empty three free net construction methods of special unmanned plane image based on minimum spanning tree, further, the structure of second step
Build in the free net of region, contextual definition regards sub- air strips as on the summit of network in graph theory between upper and lower sub- air strips, if adjacent is upper and lower
Annexation between sub- air strips be present, then two summit Jian You sides corresponding to imparting are connected, and the annexation between adjacent sub- air strips is big
It is small to be determined by following factor：
First, common point quantity, refer to the common point quantity A in overlapping range between adjacent sub- air strips, by common point quantity normalizing
Change, A is expressed as：Public points between the adjacent sub- air strips of A=n/N, wherein n expression, the total public points of N expressions or maximum
Counted out between air strips；
Second, common point is distributed, refer to the picture point distribution B of common point in overlapping range between adjacent sub- air strips, every air strips
One object space point, optimal distribution d is calculated respectivelymin, be optimally distributed dminThat corresponding image points nearest using range image center as
Standard, the optimal distribution of all object space points is then taken into the average measurement mark that size is distributed as common point between the two sub- air strips
Standard, so B is expressed as：B=(Σ dmin)/n, n the object space point number between air strips, similarly, also B is normalized；
Third, degree of overlapping, refer to the corresponding image points number C of common point in overlapping range between adjacent sub- air strips, take flat between air strips
Equal criterion of the degree of overlapping as C, so C can be expressed as：C=(Σ ki)/n, wherein kiRepresent the more of i-th object space point
Remaining observation number, n is object space point number, similarly, also by C naturalizations between [0,1]；
Relation size between sub- air strips, i.e. weight w in network on side are expressed as：
W=μ1A+μ2B+μ3C (wherein, μ1+μ2+μ3=1)
Wherein, μ1、μ2、μ3Respectively A, B, C proportionality coefficient, common point quantity is represented successively, common point is distributed, is overlapping
Weight shared in the relationship metric between two sub- air strips is spent, three's sum is 1.
The empty three free net construction methods of special unmanned plane image based on minimum spanning tree, further, the structure of second step
To build in the free net of region, the graph of a relation of whole sub- air strips is established by adjacency matrix, for the indirected net being made up of n summit,
The size of its adjacency matrix is n × n, and the element in matrix is represented with Edge [i] [j], and i, j are summit, and W is weights, indirected net
Adjacency matrix can be defined as：
In the adjacency matrix of indirected net, if 0<Edge[i][j]<∞, then it represents that have one between summit i and summit j
Nonoriented edge, its weights are Edge [i] [j], and weights are more big, and the bonding strength represented between two summits is bigger, if Edge [i] [j]
For ∞, then illustrate there is no side to be connected between summit i and summit j.
The empty three free net construction methods of special unmanned plane image based on minimum spanning tree, further, the structure of second step
Build in the free net of region, regional network connection side is chosen according to the maximum connected mode series connection region of the bonding strength of whole region net
In all sub- air strips, meet to cause the weights summation on each side to reach maximum, maximum spanning tree be converted into minimum spanning tree
Counter-example, the connection side of regional network is asked based on the principle of minimum spanning tree, and meet three conditions：First, it can only use in network
Existing side constructs spanning tree；Second, n summit in network can only be connected using n-1 bars side, n is number of vertices；Three
It is that can not produce loop.
The empty three free net construction methods of special unmanned plane image based on minimum spanning tree, further, maximum spanning tree
It is converted into minimum spanning tree and negative method is taken according to weights, minimum spanning tree uses Prim algorithms, is leading using summit, from starting
Summit is set out, by selecting currently available minimum weights side that other summits are added among spanning tree successively.
The empty three free net construction methods of special unmanned plane image based on minimum spanning tree, further, the structure of second step
Build in the free net of region, the selection of optimized start point takes classification drop power to suppress growth of the network to depth direction, is specially：
During the starting point for selecting some summit to increase as region, using the starting point first as 1 grade of summit, it is joined directly together with 1 grade of summit
Other summits as 2 grades of summits, with other summits that 2 grades of summits are joined directly together as 3 grades of summits, the like determine network
In all summits series, classification drop power be by the weight w (i, i+1) between i-stage summit and i+1 level summit on side, press
Drop power processing is carried out according to equation below：
W ' (i, i+1)=λi-1* w (i, i+1) (i=1,2,3 ...)
W is weights in formula, λi-1It is drop weight coefficient, λ is constant, and i-1 is the drop power level number on side, is joined directly together with 1 grade of summit
Side do not make drop power processing.Select to drop the strategy weighed using classification, respectively zoning as starting point per strip air strips successively
The weights summation W on middle connection side1, W2..., Wn, compare W1, W2..., WnSize, select maximum weights and corresponding son
The starting point that air strips increase as region, that is, the optimized start point to be found.
The empty three free net construction methods of special unmanned plane image based on minimum spanning tree, further, the structure of second step
Build in the free net of region, the specific method of absolute orientation structure region free net is：First according to public between adjacent sub- air strips
Point solves seven elements of absolute orientation, and seven elements of absolute orientation are according to first three offsets of solution and engineer's scale and carry out rough error
Reject, then solve three anglecs of rotation and carry out, then using the object space point coordinates system corresponding to optimized start point as whole region net
The frame of reference, according to the regional network connected mode generated, by absolute orientation formula, taking the photograph other sub- air strips successively
Website and model point coordinates are all brought into the coordinate system unified with originating sub- air strips, complete the unification of whole region coordinate system.
Compared with prior art, the advantage of the invention is that：
1. the empty three free net construction methods of the special unmanned plane image based on minimum spanning tree provided by the invention, for falling
The special unmanned plane image such as water, cloud block, overlapping leak, when solving conventional automatic triangulation, occur relative
Orientation, model connection failure the problems such as, in the case where not providing POS assistance datas, by with graph theory correlation theory and
Algorithm solves the empty automatic Construct question of three free nets of area in hardship image, and this method by programming realization, makes special nothing
The processing of man-machine image is horizontal and disposal ability is greatly enhanced.
2. the empty three free net construction methods of the special unmanned plane image based on minimum spanning tree provided by the invention, for surveying
There is the situation of relative orientation or model connection failure in area, introduces the concept of sub- air strips.After each sub- air strips free net has been established,
The construction work of region free net can be carried out.The relation (i.e. weights define) between sub- air strips is defined first, by whole region net
(being made up of sub- air strips) is converted into the Connected undigraph (i.e. indirected net) of a cum rights in graph theory, and establishes son by adjacency matrix
Graph of a relation between air strips.The connected mode of the principle formation zone net of minimum spanning tree is then based on, and using classification drop power
Strategy chooses optimized start point.Finally, coordinate system unification of the absolute orientation by whole region is carried out, i.e. region free net has been built
Finish.Solve the problems, such as that surveying area relative orientation or model connection failure occurs.If it is to provide a small amount of known control point to sit
Mark, then whole region can be made to bring into the earth axes of known control point, realize absolute orientation.
3. the empty three free net construction methods of the special unmanned plane image based on minimum spanning tree provided by the invention, are solved
Problem in aerial triangulation, solve unmanned plane it is photogrammetric in very crucial technical problem, a series of taken the photograph to be follow-up
Shadow measures product and laid a good foundation with application.Contribute to low latitude unmanned aerial vehicle remote sensing technology as a kind of emerging spatial data
The important means of acquisition, more and more extensive application is obtained in many fields.Contribute to image resolution height, the data of its acquisition
Acquisition is fast, drafting period is short, and greater advantages are played in the large scale topographical map mapping of middle small range and map rejuvenation.
Brief description of the drawings
Fig. 1 is the sub- air strips schematic diagram of the empty three free net construction methods of special unmanned plane image provided by the invention.
Fig. 2 is the indirected net schematic diagram of the empty three free net construction methods of special unmanned plane image provided by the invention.
Fig. 3 is the minimum spanning tree schematic diagram of indirected net construction provided by the invention.
Fig. 4 is the sub- air strips connected mode signal of the empty three free net construction methods of special unmanned plane image provided by the invention
Figure.
Fig. 5 is the sub- air strips absolute orientation signal of the empty three free net construction methods of special unmanned plane image provided by the invention
Figure.
Fig. 6 is user's zoom operations schematic diagram of gesture interaction method provided by the invention.
Fig. 7 is the overall technology route flow of the empty three free net construction methods of special unmanned plane image provided by the invention
Figure.
Embodiment
Below in conjunction with the accompanying drawings, to the empty three free net structures of the special unmanned plane image provided by the invention based on minimum spanning tree
The technical scheme of construction method is further described, and those skilled in the art is better understood from the present invention and can give
To implement.
Referring to Fig. 1 to Fig. 7, the empty three free nets structure of the special unmanned plane image based on minimum spanning tree provided by the invention
Method, the Photogrammetric Processing to a kind of special unmanned plane image (such as large area is overboard, cloud block image) according to routine
When flow carries out aerial triangulation, often because of relative orientation and model connection failure, cause to automatically create air strips failure, sky
Three interrupt, and then can not carry out the making (such as DEM, orthography etc.) of subsequent product.If course line is divided into several sections to skip
The difficult image such as water, then entirely survey area's can and regard as by some air strips being broken (concept for introducing " sub- air strips ") group
Into.After building each sub- air strips free net according to relative orientation, model connection to each sub- air strips, next seek to establish whole survey
The free net of sky three in area, i.e., connect all sub- air strips, then carry out regional network bundle adjustment again.To it is special nobody
Machine image carries out aerial triangulation processing, includes three steps：
The first step, sub- air strips model is established, single three-dimensional model is established according to continuity method relative orientation to each sub- air strips, so
Model connection is carried out afterwards establishes sub- air strips free net；
Sub- air strips model is established according to ready-portioned each strip air strips, establishes sub- air strips model, including following several accordingly
Step：When measure picpointed coordinate and carry out system error correction；Second, using continuity method relative orientation, single three-dimensional mould is established
Type；Third, model connects, sub- air strips free net is established；
Second step, region free net is built, calculate the bonding strength size between adjacent sub- air strips first and establish corresponding
Sub- air strips graph of a relation, it is then based on effective connection side of the principle chosen area net of minimum spanning tree and finds optimal starting
Point, the coordinate system of all sub- air strips in region is come together finally by absolute orientation；
Structure region free net is connected sub- air strips all in area are surveyed by absolute orientation, turns into an entirety,
Including the following steps：First, the annexation definition between sub- air strips, i.e. weights define up and down；Second, closed according to the connection defined
System establishes the graph of a relation of whole sub- air strips, i.e. adjacency matrix；Third, the principle chosen area net based on minimum spanning tree is optimal
Side is connected, and finds the starting point that the optimal sub- air strips of starting increase as region, using its coordinate system as reference coordinate
System, the coordinate of other sub- air strips is brought into the coordinate system unified with originating sub- air strips by absolute orientation, completes whole survey
The free net of sky three structure in area；
3rd step, bundle block adjustment, using picture point, object space point and photo centre's three point on a straight line condition, integrally ask
The elements of exterior orientation of all images in Xie Ce areas and the tight solution of closed points coordinate, as the inspection to network forming result；
Bundle block adjustment includes the following steps：One is to determine every image elements of exterior orientation and fixed point coordinate
Approximation, each stereogram is subjected to relative orientation the free air strips net of structure is connected with model, utilize the control point in air strips
And the common point between adjacent course line carries out to course line elements of exterior orientation and pass point of the absolute orientation in the hope of each photo
Geographical coordinates, the approximation as unknown number；Second, the picpointed coordinate of control point and point to be located from every width image, according to
The collinearity condition equation of every photography light lists error equation；Third, point by point methodization establishes changing normal equation, by circulation
The method for solving of piecemeal, first seek the elements of exterior orientation of every width image；Fourth, the ground of point to be located is tried to achieve according to space intersection
Coordinate, its average is taken as last result for the common point of adjacent image.
First, sub- air strips model is established
Air strips are to carry out aeroplane photography, the front and rear overlapped image sequence of acquisition along a direction.The present invention's
Sub- air strips are on the basis of a rule air strips of the aeroplane photography acquisition of routine, due to reasons such as overboard, cloud block, photography leaks
Cause original air strips some segmentation air strips for being broken and being formed occur, these segmentation air strips are referred to as " sub- air strips ", and advise
It is fixed to comprise at least two (two and more than two) images per strip air strips.
After having divided each strip air strips in region according to above-mentioned definition, next just need to establish corresponding sub- air strips mould
Type (i.e. sub- air strips free net), is mainly included the following steps that：
1st, measure picpointed coordinate and carry out system error correction.Measure each as the pass point image plane to having selected in advance
Coordinate, and system error correction is carried out to it, this process is typically all to be automatically performed by photogrammetric software.
2nd, using continuity method relative orientation, single three-dimensional model is established.Using the left photo of each picture pair in sub- air strips as base
Standard, obtain elements of relative orientation of the right photo relative to left photo.After the elements of relative orientation for calculating picture pair, handed over by front
Meeting method obtains coordinate of the model points in respective image space auxiliary coordinates, establishes single three-dimensional model.
3rd, model connects, and establishes sub- air strips free net.By single model be connected to become sub- air strips model premise be will be each
The different engineer's scale of model is classified as unified engineer's scale.It is usually equal to three connection point heights in overlapping region with adjacent picture
For condition, from left to right established successively by the engineer's scale of the engineer's scale naturalization of latter model to previous model with first mould
Sub- air strips model on the basis of the engineer's scale of type.By the model point coordinates of all stereograms and take the photograph website coordinate and bring into entirely
In the unified coordinate system in air strips.
So far, the foundation of sub- air strips model (i.e. sub- air strips free net) is just completed, is structure region free net in next step.
2nd, region free net is built
Structure region free net is related to the problem of sub- air strips region increases, i.e., will survey institute in area using certain connected mode
The sub- air strips having are connected by absolute orientation, make an entirety.First, the connection defined between sub- air strips up and down is closed
System's (i.e. weights define), the graph of a relation (i.e. adjacency matrix) of whole sub- air strips is then established according to the annexation defined, is connect
Get off the optimal connection side of the principle chosen area net based on minimum spanning tree, and find the sub- air strips conduct of an optimal starting
The starting point that region increases, using its coordinate system as the frame of reference, includes the coordinate of other sub- air strips by absolute orientation
Into the coordinate system unified with originating sub- air strips, the whole free net construction work of sky three for surveying area is completed.
1st, contextual definition between sub- air strips
Sub- air strips are regarded to the summit of network in graph theory as, if adjacent upper and lower sub- air strips are (due between left and right sub- air strips
It has been be disconnected that, therefore do not consider further that the relation between them, only considered upper and lower sub- air strips) between annexation be present, then assign pair
The two summit Jian You sides answered are connected, and are next exactly how this defines this annexation (weights i.e. on side), that is, will
Determine the relation evaluation criterion between sub- air strips.
From aeroplane photography principle, close to upper and lower sub- air strips between must have sidelapping, it is contemplated that it is heavy at this
In folded scope, common point between a number of air strips is dispersed with.Therefore, annexation between sub- air strips is evaluated (in other words
Bonding strength), it must be connected with the common point between air strips.In addition to common point quantity, the picture point distribution of common point
Also including being considered as.In addition, the degree of overlapping (i.e. redundant observation number) put between air strips is used as a reliability index, should be with larger
Weight is embodied in inside this contextual definition.
Therefore, present invention determine that adjacent sub- air strips between annexation size it is mainly relevant with following factor：
(1) common point quantity：Refer to the common point quantity (being represented with alphabetical A) in overlapping range between adjacent sub- air strips.In order to
Convenience of calculation and it is easy to compare, it is necessary to common point quantity be normalized, so A can be expressed as：A=n/N, wherein n represent phase
Public points between adjacent sub- air strips, N represent to count out between total public points or maximum air strips.
(2) common point is distributed：Refer to the picture point distribution (being represented with letter b) of common point in overlapping range between adjacent sub- air strips.
Each object space point air strips, an optimal distribution is calculated respectively (with that nearest corresponding image points of range image center
It is defined, uses dmin=min (di) represent), then the optimal distribution of all object space points is taken average as between the two sub- air strips
Common point is distributed the criterion of size, so B can be expressed as：B=(Σ dmin)/n, n the object space point number between air strips.Together
Reason, it is also desirable to which B is normalized.
(3) degree of overlapping：Also known as redundant observation number, refer to the corresponding image points number of common point in overlapping range between adjacent sub- air strips
Mesh (is represented) with letter C, criterion of the average degree of overlapping between air strips as C is taken, so C can be expressed as：C=(Σ
ki)/n, wherein kiThe redundant observation number of i-th of object space point is represented, n is object space point number.Similarly, it is also desirable to by C naturalizations to [0,
1] between.
Therefore, the several factors of the above are considered, the relation size (weights i.e. in network on side) between sub- air strips can be with
It is expressed as：
W=μ1A+μ2B+μ3C (wherein, μ1+μ2+μ3=1)
Wherein, μ1、μ2、μ3Respectively A, B, C proportionality coefficient, common point quantity is represented successively, common point is distributed, is overlapping
Weight shared in the relationship metric between two sub- air strips is spent, three's sum is 1.Factor corresponding to the bigger explanation of value is more important.
, should be with reference to its size of actual conditions reasonable distribution when performing experiment, this coefficient value is also to determine (present invention experiment by experience
In, make μ1=0.4, μ2=0.1, μ3=0.5).
2. sub- air strips graph of a relation is established
It is to be determined that the weights in network on side are big after defining the annexation in region between upper and lower sub- air strips
It is small, then whole region net (be made up of sub- air strips) cum rights that can is expressed as in graph theory Connected undigraph (i.e. without
To net, represented with G).Summit (being represented with V) in the figure represents the sub- air strips in region, and two summit Jian You sides, which are connected, (uses E tables
Show) represent have overlapping (or perhaps annexation being present) between corresponding two sub- air strips, and bonding strength size is with the power on side
Value W is represented.The graph of a relation established between sub- air strips is sought to below.
The storage mode of figure can represent that therefore, the graph of a relation between sub- air strips can also be by adjoining with adjacency matrix
Matrix is established.For an indirected net being made up of n summit, the size of its adjacency matrix is n × n, the element in matrix
With representing, then the adjacency matrix of the indirected net can be defined as：
In the adjacency matrix of indirected net, if 0<Edge[i][j]<∞, then it represents that have one between summit i and summit j
Nonoriented edge, its weights are Edge [i] [j], and weights are more big, and the bonding strength represented between two summits is bigger.If Edge [i] [j]
For ∞, then illustrate there is no side to be connected between summit i and summit j.It is convenient sometimes for processing, weights when being not attached to can also be determined
Justice is consistent (present invention is so) for the element on 0, with diagonal.
Sub- air strips schematic diagram as shown in Figure 1, whole region, which has altogether, is divided into 11 strip air strips, RED sector table in figure
Show sub- air strips, do not mark the image (ignoring this part for the time being) that red image represents broken.According to overlay region between adjacent sub- air strips
Common point and picture point information, corresponding indirected net (as shown in Fig. 2 sharing 11 summits, 13 sides) is established to Fig. 1, obtain side
On value information, it is possible thereby to build an adjacency matrix to store the summit of the indirected net, side and value information.Then,
The whole sub- air strips graph of a relation for surveying area just establishes.
3rd, regional network connection side is chosen
Whole region net (being made up of sub- air strips) is converted into an indirected net, and establishes the graph of a relation between sub- air strips
Afterwards it is necessary to the connection side of chosen area net, that is, choose a kind of optimal connected mode and sub- air strips all in region are connected
Get up, and there can not be loop generation.
Due to that can have loop case in connected graph, so the mode that summit all in figure is all connected may be just
More than one.And the weights reflection in network on side is bonding strength between two sub- air strips, the more big then bonding strength of weights
It is bigger.Therefore, a kind of connected mode of " optimal " is chosen to connect all summits in network, it should so that the weights on each side
Summation reaches maximum, and the bonding strength of such whole region net is maximum, and connection is also just most firm.
Above mentioned problem can be understood as the counter-example of minimum spanning tree problem in graph theory (it is required that weights summation is minimum), that is, require
One maximum spanning tree makes the weights summation on each side maximum, and meets three conditions：First, existing side in network can only be used
To construct spanning tree；Second, n summit in network can only be connected using n-1 bars side；Third, loop can not be produced.And scheming
By inner research, more the problem of being then minimum spanning tree, be not difficult, the maximum spanning tree can be converted into minimum generation
Set (such as weights take negative), so as to allow problem to be readily solved.Therefore, the present invention is namely based on the principle of minimum spanning tree to ask region
The connection side of net.
As Fig. 3 be to Fig. 2 construct minimum spanning tree (according to minimum spanning tree Prim algorithms obtain), the spanning tree makes
With 10 sides in former network to connect 11 summits, and produced without loop, the weights summation on each side reaches maximum.Fig. 4 is
Corresponding sub- air strips connected mode schematic diagram.So, the connection side of whole region net is just chosen.
The basic thought of Prim algorithms is using summit as leading position：It is currently available by selecting from initial vertex
Minimum weights side successively other summits be added among spanning tree come.
If connection indirected net is G (V, E), in Prim algorithms, vertex set V is divided into two subclass T and T ＇.T：When
Previous existence is into treetop point set；T ＇：It is not belonging to the vertex set of current spanning tree.Obviously have：T ∪ T ＇=V.
(1) detailed process of Prim algorithms is：
1) an initial vertex u is selected from connection indirected net G0, add it first in set T；Then selection and u0
Side (u associate, that there are minimum weights0, v), vertex v is added in vertex set T.
2) after each step from a summit (being set to u) in T, and each bar side of another summit (being set to v) in T ＇
The minimum side (u, v) of middle selection weights, vertex v is added in set T.So continue, until all tops in network
Untill point is all added in spanning tree vertex set T.
(2) algorithm legend
Set T is sky when initial, initial vertex 1 is added in set T first, then as follows each summit
It is added in set T：
1) only 1 summit, i.e. summit 1, a summit in T ＇ side, are weighed on T, another summit now in set T
The minimum side of value is (1,6), and its weights is 10, and summit 6 is added in set T by this edge；
2) now with 2 summits, i.e. summit 1,6 in set T, a summit on T, another summit in T ＇ side,
The minimum side of weights is (6,5), and its weights is 25, and summit 5 is added in set T by this edge；
3) now with 3 summits, i.e. summit 1,6,5 in set T, a summit is on T, another summit on T ＇ side
In, the minimum side of weights is (5,4), and its weights is 22, and summit 4 is added in set T by this edge；
4) now with 4 summits, i.e. summit 1,6,5,4 in set T, a summit is on T, another summit on T ＇ side
In, the minimum side of weights is (4,3), and its weights is 12, and summit 3 is added in set T by this edge；
5) now with 5 summits, i.e. summit 1,6,5,4,3 in set T, a summit is on T, another summit T ＇'s
Bian Zhong, the minimum side of weights are (3,2), and its weights is 16, and summit 2 is added in set T by this edge；
6) now with 6 summits, i.e. summit 1,6,5,4,3,2 in set T, a summit is on T, another summit in T ＇
Side in, the minimum side of weights be (2,7), and its weights is 14, and summit 7 is added in set T by this edge.
So far, all summits are all had been added in set T, and minimum spanning tree construction finishes, ultimately constructed most your pupil
Cheng Shu, the power of spanning tree and for 99.
4th, optimized start point is chosen
After choosing a complete regional network and effectively connecting side, all sub- air strips are also just series connected.Due to this
The network forming strategy of invention is intended to since a certain sub- air strips, and the cluster that region is carried out according to the connected mode of selection increases, from
And all sub- air strips are connected.The shadow of error propagation is had during in view of carrying out absolute orientation between adjacent sub- air strips
Ring, therefore, selecting different sub- air strips, its network forming effect naturally also can be different as starting point (or for father node).Also
It is to say, the invention further relates to the On The Choice of an optimal starting point.
In view of the factor of error propagation, in order to reduce the path of error propagation as far as possible, the present invention takes a kind of point
The strategy of level drop power suppresses growth of the network to depth direction.Specific practice is：Increase selecting some summit as region
During long starting point, using the starting point first as 1 grade of summit, with other summits that 1 grade of summit is joined directly together as 2 grades of summits, with 2
Other summits for being joined directly together of level summit as 3 grades of summits, the like, determine the series on all summits in network.So-called " point
What is referred to is exactly, by the weight w (i, i+1) between i-stage summit and i+1 level summit on side, to enter according to equation below to level drop power "
The processing of row drop power：
W ' (i, i+1)=λi-1* w (i, i+1) (i=1,2,3 ...)
W is weights in formula, λi-1It is drop weight coefficient, λ is constant (λ=0.8 is taken in the present embodiment), and i-1 is the drop power level on side
The processing of drop power is not made on number, the side being joined directly together with 1 grade of summit.Select to drop using classification as starting point per strip air strips successively
The strategy of power, the weights summation W on side is connected in zoning respectively1, W2..., Wn, compare W1, W2..., WnSize, selection most
The starting point that big weights and corresponding sub- air strips increase as region, that is, the optimized start point to be found.
The searching of optimized start point is carried out according to as above thought to the maximum spanning tree in Fig. 3, what is finally searched out is optimal
Starting point is summit 4, it is meant that the sub- air strips of starting that sub- air strips 4 should be selected to increase as region.
5th, absolute orientation structure region free net
Increased after have chosen the optimized start point of regional network it is necessary to carry out the cluster in region, that is, build region free net.
Because the object space point coordinates system of present each sub- air strips is inconsistent, and to build the free net of whole region, then by they
Coordinate system carries out unification, will use absolute orientation here.Specific method is：Adopted first according to the common point between adjacent sub- air strips
Seven elements of absolute orientation are solved with " two steps' minimization " (be i.e. seven parameters, first to solve three offsets and engineer's scale and carry out rough error
Reject, then solve three anglecs of rotation), the then base using the object space point coordinates system corresponding to optimized start point as whole region net
Conventional coordinates (or being reference frame), it is public by following absolute orientation according to the regional network connected mode (Fig. 4) generated
Formula successively other sub- air strips take the photograph website and model point coordinates is all brought into the coordinate system unified with originating sub- air strips, it is just complete
Into the unification of whole region coordinate system.
(absolute orientation formula)
In formula, λ is model scale, and R is the spin matrix being made up of (Φ, Ω, K) three anglecs of rotation, (Δ X, Δ Y,
Δ Z) it is translation coefficient.(X, Y, Z) is the object space to be changed point (or taking the photograph website) coordinate, is the object space point under the frame of reference
(or taking the photograph website) coordinate.
Sub- air strips absolute orientation schematic diagram as shown in Figure 5, the direction of arrow represent the direction of absolute orientation, beside arrow
Numbering represents the order of absolute orientation.Using sub- air strips 4 as starting point, its coordinate system is as the frame of reference, sequentially (according to figure
Front direction and order in 5) coordinate system of other sub- air strips is gone in the coordinate system consistent with sub- air strips 4, thus
Region-wide all sub- air strips coordinate systems are unified, i.e. the free net of sky three structure of whole region finishes.
3rd, bundle block adjustment
After building the free net of sky three in complete Ge Ce areas, the adjustment work of whole region next can be carried out, use
It is bundle block adjustment method.Principle i.e. based on picture point, object space point and photo centre's three point on a straight line, made with single light beam
For the elementary cell of adjustment, using collinearity condition equation as the mathematical modeling of adjustment, using picpointed coordinate as observation, error is listed
Equation, integrated solution survey the elements of exterior orientation of all images in area and the tight solution of closed points coordinate.Wherein, the initial value of unknown number
Obtained during by above establishing sub- air strips free net and structure region free net.If it there is provided the ground at known control point
Coordinate, whole region can also be brought into the earth axes of known control point.
Bundle block adjustment is most tight method, its theoretical and algorithm comparative maturity and comprehensively, in the world
It successively have developed many powerful adjustment Softwares, such as PATB, BINGO.The present invention uses bundle block adjustment
Method is as follows：
Bundle block adjustment aerial triangulation is that the principle based on picture point, photo centre and object space point three point on a straight line proposes
Come, its basic thought is the elementary cell using the Ray Of Light that every photo is formed as adjustment, with collinearity condition equation
Basic equation as adjustment.By the skyborne rotation and translation of each light beam, make the light realization of common point between model
Optimal intersection, and whole region is brought into the geographical coordinates system of known control point.So to establish region-wide unification
Error equation, six elements of exterior orientation of the region-wide interior every photo of integrated solution and the geographical coordinates of all unknown points.
The basic procedure of bundle block adjustment aerial triangulation is as shown in fig. 6, its main contents includes：
(1) determine that (its specific practice is for the approximation of every image elements of exterior orientation and fixed point coordinate：Will be each three-dimensional
Free air strips net is built as connect with model to carrying out relative orientation, using public between the control point in air strips and adjacent course line
Point carries out absolute orientation to course line in the hope of the elements of exterior orientation of each photo and the geographical coordinates of pass point, in this, as not
Know several approximations)；
(2) picpointed coordinate of control point and point to be located from every width image, according to the conllinear bar of every photography light
Part equation lists error equation；
(3) point by point methodization establishes changing normal equation, and by the method for solving of cyclic blocking, it is unknown first to obtain one kind therein
Number, typically first seek the elements of exterior orientation of every width image；
(4) geographical coordinates of point to be located is tried to achieve according to space intersection, for the common point of adjacent image its should be taken equal
Value is as last result.
4th, overall technology route
In summary, overall technology route flow of the invention can be described as Fig. 7, proposed by the present invention to be directed to special nothing
The free net method for auto constructing of sky three of man-machine image mainly includes following process：
(1) sub- air strips model is established：Single three-dimensional model, Ran Houjin are established according to continuity method relative orientation to each sub- air strips
Sub- air strips free net is established in the connection of row model.
(2) region free net is built：The bonding strength size between adjacent sub- air strips is calculated first and establishes corresponding son boat
Band graph of a relation, is then based on effective connection side of the principle chosen area net of minimum spanning tree and finds optimal starting point, most
The coordinate system of all sub- air strips in region is come together by absolute orientation afterwards.
(3) bundle block adjustment：Using picture point, object space point and photo centre's three point on a straight line condition, integrated solution is surveyed
The elements of exterior orientation of all images in area and the tight solution of closed points coordinate, in this, as the inspection to network forming result.
The empty three free net construction methods of special unmanned plane image based on minimum spanning tree provided by the invention, to one kind
(overboard large area, extensive forest covering such as be present or have cloud block, the shadow for leak of photographing in the unmanned plane image for comparing " special "
As etc.) in conventional aerial triangulation operation process is carried out because relative orientation or model connection failure cause to automatically create boat
Band failure, sky three interrupt, and in the case where not providing POS assistance datas, automatically build the free net of sky three of whole region, obtain
The initial value of the elements of exterior orientation of all images and object space point coordinates in get Ce areas, and bundle block adjustment is carried out, integrally ask
Solve the elements of exterior orientation of region-wide interior every image and the space coordinates of all object spaces point.Make the processing water of special unmanned plane image
Gentle disposal ability is greatly enhanced.
The basic principles, principal features and advantages of the present invention have been shown and described above.While there has been shown and described that
Embodiments of the invention, for the ordinary skill in the art, it is possible to understand that do not departing from the principle and essence of the present invention
A variety of change, modification, replacement and modification can be carried out to these embodiments, the scope of the present invention is by appended right in the case of god
It is required that and its equivalent restriction.
Claims (8)
1. the empty three free net construction methods of special unmanned plane image based on minimum spanning tree, it is characterised in that to it is special nobody
Machine image carries out aerial triangulation processing, includes three steps：
The first step, sub- air strips model is established, single three-dimensional model, Ran Houjin are established according to continuity method relative orientation to each sub- air strips
Sub- air strips free net is established in the connection of row model；
Sub- air strips model is established according to ready-portioned each strip air strips, establishes corresponding sub- air strips model, including the following steps：One
It is measurement picpointed coordinate and carries out system error correction；Second, using continuity method relative orientation, single three-dimensional model is established；Third,
Model connects, and establishes sub- air strips free net；
Second step, region free net is built, calculate the bonding strength size between adjacent sub- air strips first and establish corresponding son boat
Band graph of a relation, is then based on effective connection side of the principle chosen area net of minimum spanning tree and finds optimal starting point, most
The coordinate system of all sub- air strips in region is come together by absolute orientation afterwards；
Structure region free net is connected sub- air strips all in area are surveyed by absolute orientation, turns into an entirety, including
The following steps：First, the annexation definition between sub- air strips, i.e. weights define up and down；Second, built according to the annexation defined
The graph of a relation of vertical whole sub- air strips, i.e. adjacency matrix；Third, the connection that the principle chosen area net based on minimum spanning tree is optimal
Side, and the starting point that the optimal sub- air strips of starting increase as region is found, using its coordinate system as the frame of reference, lead to
Cross absolute orientation to bring the coordinate of other sub- air strips into the coordinate system unified with originating sub- air strips, complete the whole sky for surveying area
Three free nets are built；
3rd step, bundle block adjustment, using picture point, object space point and photo centre's three point on a straight line condition, integrated solution is surveyed
The elements of exterior orientation of all images in area and the tight solution of closed points coordinate, as the inspection to network forming result；
Bundle block adjustment includes the following steps：One is to determine the approximation of every image elements of exterior orientation and fixed point coordinate
Value, each stereogram is subjected to relative orientation the free air strips net of structure is connected with model, utilize the control point in air strips and phase
Common point between adjacent course line carries out absolute orientation to course line in the hope of the elements of exterior orientation of each photo and the ground of pass point
Coordinate, the approximation as unknown number；Second, the picpointed coordinate of control point and point to be located from every width image, according to every
The collinearity condition equation of photography light lists error equation；Third, point by point methodization establishes changing normal equation, by cyclic blocking
Method for solving, first seek the elements of exterior orientation of every width image；Fourth, the geographical coordinates of point to be located is tried to achieve according to space intersection,
Its average is taken as last result for the common point of adjacent image.
2. the empty three free net construction methods of the special unmanned plane image based on minimum spanning tree according to claim 1, its
It is characterised by, the specific method for establishing sub- air strips model is：
1st step, measure picpointed coordinate and carry out system error correction；Measure each as the pass point image plane to having selected in advance
Coordinate, and system error correction is carried out to it；
2nd step, using continuity method relative orientation, establish single three-dimensional model；Using the left photo of each picture pair in sub- air strips as base
Standard, elements of relative orientation of the right photo relative to left photo is obtained, after the elements of relative orientation for calculating picture pair, handed over by front
Meeting method obtains coordinate of the model points in respective image space auxiliary coordinates, establishes single three-dimensional model；
3rd step, model connection, establishes sub- air strips free net；By single model be connected to become sub- air strips model premise be will be each
The different engineer's scale of model is classified as unified engineer's scale, equal for bar to three connection point heights in overlapping region with adjacent picture
Part, from left to right established successively by the engineer's scale of the engineer's scale naturalization of latter model to previous model with first model
Sub- air strips model on the basis of engineer's scale, by the model point coordinates of all stereograms and take the photograph website coordinate and bring full air strips into
In unified coordinate system.
3. the empty three free net construction methods of the special unmanned plane image based on minimum spanning tree according to claim 1, its
It is characterised by, in the structure region free net of second step, contextual definition regards sub- air strips as in graph theory network between upper and lower sub- air strips
Summit, if annexation between adjacent upper and lower sub- air strips be present, assign corresponding to two summit Jian You sides be connected, it is adjacent
Annexation size between sub- air strips is determined by following factor：
First, common point quantity, refer to the common point quantity A in overlapping range between adjacent sub- air strips, common point quantity is normalized, A
It is expressed as：A=n/N, wherein n represent the public points between adjacent sub- air strips, and N represents total public points or maximum air strips
Between count out；
Second, common point is distributed, refer to the picture point distribution B of common point in overlapping range between adjacent sub- air strips, each air strips
Object space point, optimal distribution d is calculated respectivelymin, be optimally distributed dminThat corresponding image points nearest by range image center is defined, so
The optimal distribution of all object space points is taken into the average criterion that size is distributed as common point between the two sub- air strips, institute afterwards
It is expressed as with B：B=(Σ dmin)/n, n the object space point number between air strips, similarly, also B is normalized；
Third, degree of overlapping, refer to the corresponding image points number C of common point in overlapping range between adjacent sub- air strips, take the average weight between air strips
Criterion of the folded degree as C, so C can be expressed as：C=(Σ ki)/n, wherein kiRepresent the unnecessary sight of i-th of object space point
Number is surveyed, n is object space point number, similarly, also by C naturalizations between [0,1]；
Relation size between sub- air strips, i.e. weight w in network on side are expressed as：
W=μ1A+μ2B+μ3C (wherein, μ1+μ2+μ3=1)
Wherein, μ1、μ2、μ3Respectively A, B, C proportionality coefficient, common point quantity is represented successively, common point is distributed, degree of overlapping exists
Shared weight in relationship metric between two sub- air strips, three's sum are 1.
4. the empty three free net construction methods of the special unmanned plane image based on minimum spanning tree according to claim 1, its
It is characterised by, in the structure region free net of second step, the graph of a relation of whole sub- air strips is established by adjacency matrix, for by n
The indirected net that individual summit is formed, the size of its adjacency matrix be n × n, and element Edge [i] [j] expressions in matrix, i, j are
Summit, W are weights, and the adjacency matrix of indirected net can be defined as：
In the adjacency matrix of indirected net, if 0<Edge[i][j]<∞, then it represents that have between summit i and summit j one it is undirected
Side, its weights are Edge [i] [j], and weights are more big, and the bonding strength represented between two summits is bigger, if Edge [i] [j] is
∞, then illustrate there is no side to be connected between summit i and summit j.
5. the empty three free net construction methods of the special unmanned plane image based on minimum spanning tree according to claim 1, its
Be characterised by, in the structure region free net of second step, regional network connection side choose according to whole region net bonding strength most
All sub- air strips, meet to cause the weights summation on each side to reach maximum, maximum are generated in big connected mode series connection region
Tree is converted into the counter-example of minimum spanning tree, the connection side of regional network is asked based on the principle of minimum spanning tree, and meet three bars
Part：First, spanning tree can only be constructed using existing side in network；Second, the n in network can only be connected using n-1 bars side
Individual summit, n are number of vertices；Third, loop can not be produced.
6. the empty three free net construction methods of the special unmanned plane image based on minimum spanning tree according to claim 5, its
It is characterised by, maximum spanning tree is converted into minimum spanning tree and negative method is taken according to weights, and minimum spanning tree uses Prim algorithms,
It is leading using summit, from initial vertex, by selecting currently available minimum weights side that other summits are added to successively
Among spanning tree.
7. the empty three free net construction methods of the special unmanned plane image based on minimum spanning tree according to claim 1, its
It is characterised by, in the structure region free net of second step, the selection of optimized start point takes classification drop power to suppress network to depth
The growth in direction, it is specially：During the starting point increased on some summit of selection as region, using the starting point first as 1 grade of top
Point, with other summits that 1 grade of summit is joined directly together as 2 grades of summits, the other summits being joined directly together with 2 grades of summits are as 3 grades
Summit, the like determine network in all summits series, classification drop power be by between i-stage summit and i+1 level summit
Weight w (i, i+1) on side carries out drop power processing according to equation below：
W ' (i, i+1)=λi-1* w (i, i+1) (i=1,2,3 ...)
W is weights in formula, λi-1It is drop weight coefficient, λ is constant, and i-1 is the drop power level number on side, the side being joined directly together with 1 grade of summit
The processing of drop power is not made.Select to drop the strategy of power using classification as starting point per strip air strips, respectively in zoning connect successively
The weights summation W of edge fit1, W2..., Wn, compare W1, W2..., WnSize, select maximum weights and corresponding sub- air strips
The starting point increased as region, that is, the optimized start point to be found.
8. the empty three free net construction methods of the special unmanned plane image based on minimum spanning tree according to claim 1, its
It is characterised by, in the structure region free net of second step, the specific method of absolute orientation structure region free net is：Basis first
Common point between adjacent sub- air strips solves seven elements of absolute orientation, and seven elements of absolute orientation are according to first three offsets of solution
With engineer's scale and carry out elimination of rough difference, then solve three anglecs of rotation and carry out, then sit the object space point corresponding to optimized start point
The frame of reference of the mark system as whole region net, according to the regional network connected mode generated, by absolute orientation formula, according to
It is secondary other sub- air strips take the photograph website and model point coordinates is all brought into the coordinate system unified with originating sub- air strips, complete whole
The unification of Regional coordinate system.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710747621.XA CN107560593B (en) | 2017-08-28 | 2017-08-28 | Special unmanned aerial vehicle image air-three free network construction method based on minimum spanning tree |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710747621.XA CN107560593B (en) | 2017-08-28 | 2017-08-28 | Special unmanned aerial vehicle image air-three free network construction method based on minimum spanning tree |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107560593A true CN107560593A (en) | 2018-01-09 |
| CN107560593B CN107560593B (en) | 2019-12-17 |
Family
ID=60977180
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710747621.XA Active CN107560593B (en) | 2017-08-28 | 2017-08-28 | Special unmanned aerial vehicle image air-three free network construction method based on minimum spanning tree |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107560593B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108600026A (en) * | 2018-05-07 | 2018-09-28 | 重庆邮电大学 | Neighborhood storage method and device |
| CN108761271A (en) * | 2018-03-30 | 2018-11-06 | 广州中科云图智能科技有限公司 | A kind of power grid screen of trees detection method and system |
| CN109887026A (en) * | 2019-02-20 | 2019-06-14 | 深圳市未来感知科技有限公司 | Multi-angle of view positioning and tracing method, device, equipment and computer readable storage medium |
| CN110312085A (en) * | 2019-06-06 | 2019-10-08 | 武汉易科空间信息技术股份有限公司 | Image interfusion method and system based on multiple unmanned plane technologies |
| CN110853142A (en) * | 2019-11-20 | 2020-02-28 | 中国民航科学技术研究院 | Airport clearance three-dimensional model construction method and device based on unmanned aerial vehicle shooting |
| CN111244822A (en) * | 2020-03-20 | 2020-06-05 | 广东电网有限责任公司 | Fixed-wing unmanned aerial vehicle line patrol method, system and device in complex geographic environment |
| CN112269202A (en) * | 2020-10-15 | 2021-01-26 | 武汉大学 | Motion carrier assisted space reference transmission system and method |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101226057A (en) * | 2008-02-01 | 2008-07-23 | 武汉朗视软件有限公司 | Digital close range photogrammetry method |
| US20100166295A1 (en) * | 2008-12-31 | 2010-07-01 | Industrial Technology Research Institute | Method and system for searching for global minimum |
| CN102096816A (en) * | 2011-01-28 | 2011-06-15 | 武汉大学 | Multi-scale multi-level image segmentation method based on minimum spanning tree |
| CN102506824A (en) * | 2011-10-14 | 2012-06-20 | 航天恒星科技有限公司 | Method for generating digital orthophoto map (DOM) by urban low altitude unmanned aerial vehicle |
| CN105469061A (en) * | 2015-08-04 | 2016-04-06 | 电子科技大学中山学院 | Topographic feature line extraction method and device |
| CN106127782A (en) * | 2016-06-30 | 2016-11-16 | 北京奇艺世纪科技有限公司 | A kind of image partition method and system |
-
2017
- 2017-08-28 CN CN201710747621.XA patent/CN107560593B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101226057A (en) * | 2008-02-01 | 2008-07-23 | 武汉朗视软件有限公司 | Digital close range photogrammetry method |
| US20100166295A1 (en) * | 2008-12-31 | 2010-07-01 | Industrial Technology Research Institute | Method and system for searching for global minimum |
| CN102096816A (en) * | 2011-01-28 | 2011-06-15 | 武汉大学 | Multi-scale multi-level image segmentation method based on minimum spanning tree |
| CN102506824A (en) * | 2011-10-14 | 2012-06-20 | 航天恒星科技有限公司 | Method for generating digital orthophoto map (DOM) by urban low altitude unmanned aerial vehicle |
| CN105469061A (en) * | 2015-08-04 | 2016-04-06 | 电子科技大学中山学院 | Topographic feature line extraction method and device |
| CN106127782A (en) * | 2016-06-30 | 2016-11-16 | 北京奇艺世纪科技有限公司 | A kind of image partition method and system |
Non-Patent Citations (1)
| Title |
|---|
| 杨柳 等: "《无人机遥感影像的城市绿地信息提取》", 《测绘科学》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108761271A (en) * | 2018-03-30 | 2018-11-06 | 广州中科云图智能科技有限公司 | A kind of power grid screen of trees detection method and system |
| CN108600026A (en) * | 2018-05-07 | 2018-09-28 | 重庆邮电大学 | Neighborhood storage method and device |
| CN109887026A (en) * | 2019-02-20 | 2019-06-14 | 深圳市未来感知科技有限公司 | Multi-angle of view positioning and tracing method, device, equipment and computer readable storage medium |
| CN109887026B (en) * | 2019-02-20 | 2021-07-16 | 深圳市未来感知科技有限公司 | Multi-view positioning tracking method, device and equipment and computer readable storage medium |
| CN110312085A (en) * | 2019-06-06 | 2019-10-08 | 武汉易科空间信息技术股份有限公司 | Image interfusion method and system based on multiple unmanned plane technologies |
| CN110853142A (en) * | 2019-11-20 | 2020-02-28 | 中国民航科学技术研究院 | Airport clearance three-dimensional model construction method and device based on unmanned aerial vehicle shooting |
| CN111244822A (en) * | 2020-03-20 | 2020-06-05 | 广东电网有限责任公司 | Fixed-wing unmanned aerial vehicle line patrol method, system and device in complex geographic environment |
| CN112269202A (en) * | 2020-10-15 | 2021-01-26 | 武汉大学 | Motion carrier assisted space reference transmission system and method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107560593B (en) | 2019-12-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107560593A (en) | The empty three free net construction methods of special unmanned plane image based on minimum spanning tree | |
| WO2020164092A1 (en) | Image processing method and apparatus, moveable platform, unmanned aerial vehicle and storage medium | |
| CN103703490B (en) | For generation of the equipment of three-dimensional feature data and the method for generation of three-dimensional feature data | |
| CN103411609B (en) | A kind of aircraft return route planing method based on online composition | |
| CN109238239B (en) | Digital measurement three-dimensional modeling method based on aerial photography | |
| CN107480727A (en) | The unmanned plane image fast matching method that a kind of SIFT and ORB are combined | |
| CN105069843A (en) | Rapid extraction method for dense point cloud oriented toward city three-dimensional modeling | |
| CN107833279B (en) | DEM-based terrain slope analysis method | |
| CN104360362B (en) | Method and system for positioning observed object via aircraft | |
| KR101150510B1 (en) | Method for Generating 3-D High Resolution NDVI Urban Model | |
| CN107527328A (en) | A kind of unmanned plane image geometry processing method for taking into account precision and speed | |
| CN104966281A (en) | IMU/GNSS guiding matching method of multi-view images | |
| CN106949880B (en) | The method that area's unmanned plane image part degree of overlapping crosses high disposal is surveyed in height above sea level big rise and fall | |
| CN106940181A (en) | A kind of unmanned plane image picture control distribution of net is built and the optional commensurate in scope method of aerophotograph | |
| JP6347061B2 (en) | Measurement result consistency determination method and measurement result consistency determination apparatus | |
| Trujillo et al. | Optimized flight path for 3D mapping of an area with structures using a multirotor | |
| CN103426153B (en) | The quick joining method of a kind of unmanned aerial vehicle remote sensing images | |
| CN114089787A (en) | Ground three-dimensional semantic map based on multi-machine cooperative flight and construction method thereof | |
| CN115014361A (en) | Air route planning method, device and computer storage medium | |
| CN108801225A (en) | A kind of unmanned plane tilts image positioning method, system, medium and equipment | |
| Jiang et al. | Low–high orthoimage pairs-based 3D reconstruction for elevation determination using drone | |
| CN112799430B (en) | Programmable unmanned aerial vehicle-based road surface image intelligent acquisition method | |
| CN104200527B (en) | A kind of generation method of true orthophoto | |
| CN102141399B (en) | dynamic keyframe-based scene matching method in non-suitable area | |
| CN114359489A (en) | Method, device and equipment for making real-scene image in pipeline construction period and storage medium |
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 | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20211116 Address after: 214000 room 7-810, Aokai City Plaza, 1777 Zhonghui Avenue, Huishan District, Wuxi City, Jiangsu Province (Urban Railway Huishan station area) Patentee after: Wuxi Zike Technology Co., Ltd Address before: 448000 building c2-5, No. 201, Peigong Avenue, Duodao District, Jingmen City, Hubei Province (people's Wanfu entrepreneurship Park) Patentee before: Jingmen Chengyuan Electronic Technology Co., Ltd |