CN108332721B - Aviation image parallel air-space three and recursive fusion method - Google Patents
Aviation image parallel air-space three and recursive fusion method Download PDFInfo
- Publication number
- CN108332721B CN108332721B CN201810172145.8A CN201810172145A CN108332721B CN 108332721 B CN108332721 B CN 108332721B CN 201810172145 A CN201810172145 A CN 201810172145A CN 108332721 B CN108332721 B CN 108332721B
- Authority
- CN
- China
- Prior art keywords
- sub
- measurement
- image
- overlapped
- fusion
- 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.)
- Expired - Fee Related
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C11/00—Photogrammetry or videogrammetry, e.g. stereogrammetry; Photographic surveying
- G01C11/04—Interpretation of pictures
- G01C11/30—Interpretation of pictures by triangulation
- G01C11/34—Aerial triangulation
Abstract
The invention discloses a parallel empty three and recursive fusion method for aerial images, which comprises the steps of dividing a large measurement area into a plurality of sub-measurement areas according to image GPS plane coordinates by using a KD tree algorithm, controlling the image quantity of each sub-measurement area, then respectively constructing a free network and a GPS auxiliary beam adjustment for the sub-measurement areas by using the conventional SFM algorithm in a multi-machine parallel mode, and finally gradually fusing the empty three results of the sub-measurement areas by using the recursive algorithm to obtain the empty three results of the whole measurement area. The invention has the advantages that: the method of partitioning the large measurement area by the SFM solves the problems of large error, low speed and even failure of the large data size SFM, and the recursive fusion method is adopted, so that each image in the obtained empty three results only has a unique group of calculation results, and the problem of edge joint dislocation of the measurement area is solved.
Description
Technical Field
The invention relates to the technical field of surveying and mapping science, in particular to a method for fusing aerial image parallel aerial three and recursion.
Background
Aerial photogrammetry is an important technical means in the field of surveying and mapping production, and the principle of the aerial photogrammetry is that a ground area is photographed through a camera carried on an aerial vehicle, then aerial triangulation (air triangulation for short) of photogrammetry is adopted to carry out geometric processing on aerial images, and internal and external orientation elements and air triangulation encryption points of the images are obtained.
The traditional aerial photography generally adopts a vertically downward camera to shoot, and the air route is relatively regular (for example, the air routes are mutually parallel or vertical), so that the air-to-three calculation can be conveniently carried out by adopting a traditional air-to-three method. With the rise of unmanned aerial vehicles and oblique photography technologies, at the present stage, when aerial photography is performed, in addition to a vertically downward lens, a lens facing to a side face is generally used at the same time, so that the difference between the visual angle and the scale of images is large, and in addition, when the unmanned aerial vehicle is adopted for aerial photography, the unmanned aerial vehicle flies along an interested place and the flight route is very random, and due to the reasons, a correct calculation result cannot be obtained by adopting a traditional aerial three-dimensional algorithm. For the processing of such data, the SFM (Structure from Motion) technology, which is currently popular in the field of computer vision, can be used to obtain ideal processing results. However, the SFM technique has the disadvantages that when the data size is large (for example, more than 1000 images), the calculation speed is very slow, the image position information obtained by calculation has large drift, and the calculation accuracy is poor. At present, when aerial photography is carried out, the amount of data obtained is very large, generally more than 1 ten thousand images are obtained, and some items even 10 ten thousand images cause great challenges for processing the air camera, the air camera and the air camera.
The existing method for processing large data volume is to divide the large data into blocks and then perform a null processing for each block. Therefore, the problem of edge connection at the boundary of the blocks is caused, if a part of the same images (called overlapped images) exist between the adjacent blocks, after the two blocks are subjected to space-three independently, the internal and external orientation elements of the images have different calculated values in the two blocks, if the space-three results are used for subsequent mapping, modeling and the like, the condition that the model at the boundary is staggered can occur, and the actual requirements can not be met.
Disclosure of Invention
The invention aims to provide a method for fusing aerial image parallel aerial three and recursion, which solves the aerial three problem of large-data-volume aerial images.
The purpose of the invention is realized by the following technical scheme:
an aerial image parallel air-to-air triple and recursive fusion method comprises the following steps:
dividing the aerial image into a plurality of sub-measurement areas according to the GPS plane coordinate based on the KD tree;
performing independent space-three calculation on all the divided sub-measurement areas by adopting a parallel processing mode of a plurality of computers, after obtaining space-three results, obtaining space-three results under an object space coordinate system by adopting a GPS auxiliary beam adjustment method, and converting each sub-measurement area to a unified coordinate system;
and according to KD tree index information reserved when the sub-measurement regions are divided, adjusting the sub-measurement regions in a pairwise fusion beam method in a recursion mode, and using the fusion result of each time for the next fusion until the fusion is carried out to the root node.
According to the technical scheme provided by the invention, on one hand, the divided sub-measurement areas are subjected to parallel space division in a blocking mode, so that the space division efficiency and the space division precision are improved; on the other hand, only one group of internal and external orientation elements are calculated for each image during fusion, and the problem of block edge joint dislocation in the traditional method is solved, so that the method has wide application prospect in the field of aerial photogrammetry.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on the drawings without creative efforts.
Fig. 1 is a flowchart of an aerial image parallel aerial three and recursive fusion method according to an embodiment of the present invention;
fig. 2 is a schematic diagram of dividing a sub-measurement region according to a main direction of a minimum outsourcing rectangle of the measurement region based on a KD tree according to an embodiment of the present invention;
fig. 3 is a schematic diagram of recursive empty-three result fusion according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention are clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present invention without making any creative effort, shall fall within the protection scope of the present invention.
The embodiment of the invention provides a method for fusing aerial image parallel aerial three and recursive, which mainly comprises the following steps as shown in figure 1:
step 1, dividing the aerial image into a plurality of sub-measurement areas according to the GPS plane coordinate based on the KD tree.
In the embodiment of the invention, the number of images contained in each sub-measurement area does not exceed a set value K, wherein K is K × N; wherein N represents the number of cameras; k is a constant, which can be set to 70, for example.
As the KD tree is used for dividing the sub-measurement areas in the direction parallel to the coordinate axes, for the measurement areas in the inclined direction, the sub-measurement areas are quite irregular in shape after being divided according to the coordinate axes, and the SFM is not beneficial to carrying out space division. In the embodiment of the invention, the minimum outsourcing rectangle of the whole measuring area is extracted, if the minimum outsourcing rectangle has a certain inclination angle (shown as the left side of the figure 2) with the coordinate axis, the GPS plane coordinate of the image is rotated according to the main direction of the minimum outsourcing rectangle, so that the main direction of the rotated measuring area is parallel to the coordinate axis, and the KD tree is used for dividing the sub-measuring area, thereby obtaining the sub-measuring area with regular shape (shown as the right side of the figure 2). In the embodiment of the invention, the KD tree is adopted instead of the equal-interval grid, so that the KD tree can be adopted to obtain a uniform blocking result in consideration of the fact that the image track points are not uniformly distributed in the measuring area and some places are denser than images at other places.
Assuming that the included angle between the main direction of the minimum wrapping rectangle and the X axis is phi, rotating the image plane coordinate by adopting the following formula:
wherein, (x ', y') is a plane coordinate after rotation, and (x, y) is a plane coordinate before rotation.
In the embodiment of the invention, a KD tree index is constructed by adopting a KD tree algorithm, when the number of images contained in a certain node is less than a set value K, the certain node is taken as a leaf node and is not divided any more, and finally, the number of images contained in all the leaf nodes is less than or equal to the set value K; the leaf nodes are the sub-measurement areas.
There is no overlapping image between the sub-measurement areas divided by the above method, and in order to facilitate the fusion of the sub-measurement areas, the outsourcing rectangle of each sub-measurement area needs to be expanded to a certain extent, so that each sub-measurement area contains a certain number of other images, which is called sub-measurement area expansion, thereby causing a certain overlap between the sub-measurement areas. Illustratively, the range of the sub-measurement region enclosing rectangle can be expanded by 5% each time in a cyclic manner until the number of external images contained in the measurement region reaches 10% of the number of internal images in the measurement region.
And 2, performing independent space-three calculation on all the divided sub-measurement areas by adopting a parallel processing mode of a plurality of computers, and after obtaining space-three results, obtaining space-three results under an object coordinate system by adopting a GPS auxiliary beam adjustment method, so that each sub-measurement area is converted into a unified coordinate system.
Those skilled in the art will appreciate that the null-three calculation can be implemented using existing SFM algorithms.
And 3, according to KD tree index information reserved when the sub-measurement areas are divided, adjusting the sub-measurement areas in a pairwise fusion beam method in a recursion mode, and using the fusion result of each time for the next fusion until the fusion is fused to the root node.
The recursive fusion is carried out in a subsequent traversal mode, starting from a root node, for any node, whether a child node which is not fused exists is judged, if yes, a left child node is fused, and then a right child node is fused; if not, fusing the nodes; until the root node is merged. Wherein the leaf nodes are initialized to a fused state.
As shown in fig. 3, a four-layered KD tree has a total of 7 leaf nodes, i.e., 7 sub-regions, and the fusion sequence of the sub-regions is shown by the arrows with sequence numbers in the figure.
When the two sub-measurement areas are fused, the integral bundle adjustment is carried out on the image data of the two sub-measurement areas, and the steps are as follows: firstly, searching out an overlapped image and overlapped encryption points in two sub-measurement areas, wherein the overlapped image is obtained by searching images with the same image name in the two sub-measurement areas, and the overlapped encryption points are obtained by searching encryption points with the same image point in the two sub-measurement areas; then, calculating corresponding new initial parameters for the overlapped images and the overlapped encryption points in a mode of taking the average value of the numerical values in the two sub-measurement areas, so that each overlapped image and each overlapped encryption point only have a unique group of parameter values; and finally, performing bundle adjustment to optimize the empty three results.
Since each sub-measurement area is subjected to space-three independently, the parameter values of the overlapped images and the overlapped encryption points between the sub-measurement areas always have a certain difference, the method of taking the average value is only approximate to the real result, and if the error of the initial value is too large when the adjustment is fused, the adjustment of the light beam method cannot be converged. Therefore, in the sub-measurement-area fusion process, a mode of only fusing two measurement areas at a time is needed, so that the error in the initial value is as small as possible, and a fusion result meeting the requirements can be obtained.
In the scheme of the embodiment of the invention, the KD tree is adopted to divide the measuring area rotated along the main direction of the minimum outsourcing rectangle, so that the number of the divided sub-measuring areas is uniform, the shapes of the sub-measuring areas are regular, and the SFM calculation is facilitated; and a recursive fusion mode is adopted, only two measuring areas are fused each time, the error of an unknown number before fusion can be reduced, and the convergence of adjustment by a beam method and the final fusion precision are ensured.
Through the above description of the embodiments, it is clear to those skilled in the art that the above embodiments can be implemented by software, and can also be implemented by software plus a necessary general hardware platform. With this understanding, the technical solutions of the embodiments can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a usb disk, a removable hard disk, etc.), and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods according to the embodiments of the present invention.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (4)
1. An aerial image parallel air-to-air triple and recursive fusion method is characterized by comprising the following steps:
dividing the aerial image into a plurality of sub-measurement areas according to the GPS plane coordinate based on the KD tree;
performing independent space-three calculation on all the divided sub-measurement areas by adopting a parallel processing mode of a plurality of computers, after obtaining space-three results, obtaining space-three results under an object space coordinate system by adopting a GPS auxiliary beam adjustment method, and converting each sub-measurement area to a unified coordinate system;
according to KD tree index information reserved when the sub-measurement regions are divided, adjusting the sub-measurement regions in a pairwise fusion beam method in a recursion mode, and using the fusion result of each time for the next fusion until the fusion is fused to a root node;
the KD tree-based aviation image is divided into a plurality of sub-measurement areas according to the GPS plane coordinates according to the following steps:
extracting a minimum outsourcing rectangle of the whole measuring area, rotating the GPS plane coordinate of the image according to the main direction of the minimum outsourcing rectangle if the minimum outsourcing rectangle has a certain inclination angle with the coordinate axis, enabling the main direction of the rotated measuring area to be parallel to the coordinate axis, and dividing the sub-measuring areas by using a KD tree so as to obtain the sub-measuring areas with regular shapes;
assuming that the included angle between the main direction of the minimum wrapping rectangle and the X axis is phi, rotating the image plane coordinate by adopting the following formula:
wherein, (x ', y') is a plane coordinate after rotation, and (x, y) is a plane coordinate before rotation;
the method further comprises the following steps: and expanding the outsourcing rectangle of each sub-measurement area to a certain degree to ensure that each sub-measurement area comprises a certain number of other images, thereby ensuring that the sub-measurement areas are overlapped to a certain extent.
2. The aerial image parallel air-three and recursive fusion method according to claim 1, wherein each sub-measurement region contains no more images than a set value K;
constructing a KD tree index by adopting a KD tree algorithm, and when the number of images contained in a certain node is less than a set value K, the certain node is taken as a leaf node and is not divided any more, so that the number of images contained in all the leaf nodes is less than or equal to the set value K; the leaf node is a sub-measurement area;
wherein the set value K is the product of the number N of cameras and a constant K.
3. The aerial image parallel empty three and recursive fusion method according to claim 1, wherein the recursive fusion is performed in a subsequent traversal manner, starting from a root node, for any node, whether a child node which is not fused exists is judged, if yes, a left child node is fused, and then a right child node is fused; if not, fusing the nodes; until the root node is merged.
4. The aerial image parallel aerial tri-and-recursive fusion method according to claim 1 or 3, wherein when fusing the two sub-regions, the integral beam adjustment is performed on the image data of the two sub-regions, and the steps are as follows:
firstly, searching out an overlapped image and overlapped encryption points in two sub-measurement areas, wherein the overlapped image is obtained by searching images with the same image name in the two sub-measurement areas, and the overlapped encryption points are obtained by searching encryption points with the same image point in the two sub-measurement areas;
then, calculating corresponding new initial parameters for the overlapped images and the overlapped encryption points in a mode of taking the average value of the numerical values in the two sub-measurement areas, so that each overlapped image and each overlapped encryption point only have a unique group of parameter values;
and finally, performing bundle adjustment to optimize the empty three results.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810172145.8A CN108332721B (en) | 2018-03-01 | 2018-03-01 | Aviation image parallel air-space three and recursive fusion method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810172145.8A CN108332721B (en) | 2018-03-01 | 2018-03-01 | Aviation image parallel air-space three and recursive fusion method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108332721A CN108332721A (en) | 2018-07-27 |
| CN108332721B true CN108332721B (en) | 2020-12-11 |
Family
ID=62930464
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810172145.8A Expired - Fee Related CN108332721B (en) | 2018-03-01 | 2018-03-01 | Aviation image parallel air-space three and recursive fusion method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108332721B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111536871B (en) * | 2020-05-07 | 2022-05-31 | 武汉大势智慧科技有限公司 | Accurate calculation method for volume variation of multi-temporal photogrammetric data |
| CN113405533B (en) * | 2021-05-27 | 2022-04-12 | 武汉天远视科技有限责任公司 | Aerial triangulation operation method for aerial images |
| CN114279412A (en) * | 2021-11-26 | 2022-04-05 | 武汉大势智慧科技有限公司 | Multi-block space-three adjustment merging method based on aerial oblique photography image |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102519436B (en) * | 2011-12-28 | 2013-07-24 | 武汉大学 | Chang'e-1 (CE-1) stereo camera and laser altimeter data combined adjustment method |
| CN102706326B (en) * | 2012-04-27 | 2014-01-01 | 北京市测绘设计研究院 | Processing method of light beam method aerial triangulation file data |
| CN104299228B (en) * | 2014-09-23 | 2017-08-25 | 中国人民解放军信息工程大学 | A kind of remote sensing image dense Stereo Matching method based on Accurate Points position prediction model |
| CN104897176B (en) * | 2015-06-29 | 2018-05-04 | 北京建筑大学 | A kind of photogrammetric block adjustment method of multi-core parallel concurrent |
| CN105004321B (en) * | 2015-07-17 | 2017-05-10 | 湖北省电力勘测设计院 | Unmanned plane GPS-supported bundle djustment method in consideration of non-synchronous exposal |
| JP6529372B2 (en) * | 2015-07-27 | 2019-06-12 | 株式会社パスコ | Three-dimensional measurement device, three-dimensional measurement method, and program |
| CN107504959B (en) * | 2017-08-22 | 2020-04-03 | 北京中测智绘科技有限公司 | Method for measuring house wall base outline by utilizing inclined aerial image |
-
2018
- 2018-03-01 CN CN201810172145.8A patent/CN108332721B/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN108332721A (en) | 2018-07-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10580204B2 (en) | Method and device for image positioning based on 3D reconstruction of ray model | |
| KR102068419B1 (en) | Method, apparatus and computer readable medium for adjusting point cloud data collection trajectory | |
| CN110910493B (en) | Three-dimensional reconstruction method and device and electronic equipment | |
| CN108332721B (en) | Aviation image parallel air-space three and recursive fusion method | |
| CN112927370A (en) | Three-dimensional building model construction method and device, electronic equipment and storage medium | |
| WO2018061010A1 (en) | Point cloud transforming in large-scale urban modelling | |
| CN103871102B (en) | A kind of road three-dimensional fine modeling method based on elevational point and road profile face | |
| CN106570468A (en) | Method for reconstructing LiDAR original point cloud building contour | |
| CN107907111B (en) | Automatic distributed aerial triangulation calculation method | |
| CN109708622A (en) | The method that three-dimensional modeling is carried out to building using unmanned plane based on Pixhawk | |
| CN112712596B (en) | Dense matching point cloud building structured model fine reconstruction method | |
| CN104992074A (en) | Method and device for splicing strip of airborne laser scanning system | |
| CN112884256B (en) | Path planning method and device, computer equipment and storage medium | |
| CN111915517B (en) | Global positioning method suitable for RGB-D camera under indoor illumination unfavorable environment | |
| Zhang et al. | Continuous aerial path planning for 3D urban scene reconstruction. | |
| KR102127679B1 (en) | System for correcting geometry of mobile platform with sensor based on an orthophoto | |
| CN111652241B (en) | Building contour extraction method integrating image features and densely matched point cloud features | |
| CN114332291A (en) | Oblique photography model building outer contour rule extraction method | |
| CN113177974A (en) | Point cloud registration method and device, electronic equipment and storage medium | |
| WO2023124676A1 (en) | 3d model construction method, apparatus, and electronic device | |
| CN113052880A (en) | SFM sparse reconstruction method, system and application | |
| CN111784838A (en) | Super-long linear structure three-dimensional real scene model processing method based on oblique photography | |
| CN115272248B (en) | Intelligent detection method for fan gesture and electronic equipment | |
| CN114119892B (en) | Three-dimensional digital road network construction method based on BIM and GIS technologies | |
| Zhu et al. | Feature line based building detection and reconstruction from oblique airborne imagery |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20201211 |