Map trimming measurement method and device
阅读说明：本技术 一种地图修测测量方法及装置 (Map trimming measurement method and device ) 是由 刘松林 于英 于 2018-07-11 设计创作，主要内容包括：本发明涉及一种地图修测测量方法及装置,该方法首先对地图修测车无法到达的区域进行拍照,得到盲区影像数据；通过GNSS获取GNSS定位数据；然后结合盲区影像数据与GNSS定位数据,以GNSS定位数据来提供空间坐标基准,得到盲区的三维重建数据；最后将得到的盲区的三维重建数据与地图修测车中已有的地形图数据进行融合处理,将融合处理后的结果用于地图修测。本发明采用便携式数据采集终端对地图修测车无法达到的盲区进行测量,简单可靠,可以更全面的获取道路、空间物等各种信息,从而使得地图修测车输出的数据更加准确,满足地物三维信息采集和数据库建设的需求。(The invention relates to a method and a device for measuring a map repair, wherein the method comprises the steps of firstly photographing an area which cannot be reached by a map repair measuring vehicle to obtain blind area image data; acquiring GNSS positioning data through GNSS; then combining the blind area image data with GNSS positioning data to provide space coordinate reference by the GNSS positioning data to obtain three-dimensional reconstruction data of the blind area; and finally, fusing the obtained three-dimensional reconstruction data of the blind area with the existing topographic map data in the topographic map repairing and measuring vehicle, and using the fused result for the topographic map repairing and measuring. The invention adopts the portable data acquisition terminal to measure the blind area which can not be reached by the map repairing and measuring vehicle, is simple and reliable, and can more comprehensively acquire various information such as roads, space objects and the like, thereby enabling the data output by the map repairing and measuring vehicle to be more accurate and meeting the requirements of three-dimensional information acquisition of the ground objects and database construction.)
1. A map trimming measurement method is characterized by comprising the following steps:
photographing an area which cannot be reached by the map repairing and measuring vehicle to obtain blind area image data;
acquiring GNSS positioning data through GNSS;
combining the blind area image data with GNSS positioning data, and providing space coordinates by the GNSS positioning data as a reference to obtain three-dimensional reconstruction data of the blind area;
fusing the obtained three-dimensional reconstruction data of the blind area with existing topographic map data in the topographic map repair and measurement vehicle, and using the fused result for the topographic map repair and measurement;
the existing topographic map data in the topographic map repairing and measuring vehicle also takes the space coordinate provided by the GNSS positioning data as a reference.
2. The method according to claim 1, wherein the feature extraction, the camera pose estimation and the three-dimensional point cloud calculation are performed by combining the blind area image data and the GNSS positioning data to obtain the three-dimensional reconstruction data of the blind area.
3. The method of geodesic surveying of claim 2, wherein the feature extraction comprises:
extracting features of blind area image data by using an SIFT invariant feature operator, performing feature block matching by using a homography of low-resolution image estimation, and performing gross error elimination on a matching result by using the homography and a basic matrix;
multi-view series connection is carried out on the SIFT features to obtain multi-view homonymy points, and then object space point coordinates are obtained through multi-view front intersection positioning;
and constructing a uniform grid in the object space, and screening the point positions to obtain uniform point positions.
4. The method of claim 3, wherein after obtaining the uniform point locations, further comprising the step of performing precision optimization on the uniform point locations obtained by screening by using multi-scale least square matching to obtain coordinates of the connection points.
5. The geodesic measurement method of claim 2, wherein the taking camera pose estimation comprises:
calculating the phase position of the shooting camera by adopting a multi-view geometric principle;
the absolute position of the camera is determined by using a GPS positioning system.
6. The method according to claim 1, wherein the objects on both sides of the road having the basic scale topographic map data are updated while the obtained three-dimensional reconstruction data of the blind area and the topographic map data existing in the topographic map repairing and measuring vehicle are fused.
7. A map repair measurement device, comprising a processor, wherein the processor is configured to execute instructions to implement the following method:
photographing an area which cannot be reached by the map repairing and measuring vehicle to obtain blind area image data;
acquiring GNSS positioning data through GNSS;
combining the blind area image data with GNSS positioning data to provide a space coordinate reference by the GNSS positioning data to obtain three-dimensional reconstruction data of the blind area;
fusing the obtained three-dimensional reconstruction data of the blind area with existing topographic map data in the topographic map repair and measurement vehicle, and using the fused result for the topographic map repair and measurement;
the existing topographic map data in the topographic survey vehicle also provides a spatial coordinate reference by GNSS positioning data.
8. The device according to claim 7, wherein feature extraction, camera pose estimation and three-dimensional point cloud calculation are performed by combining the blind area image data and GNSS positioning data to obtain the three-dimensional reconstruction data of the blind area.
9. The geodesic surveying device of claim 8, wherein the feature extraction comprises:
extracting features of blind area image data by using an SIFT invariant feature operator, performing feature block matching by using a homography of low-resolution image estimation, and performing gross error elimination on a matching result by using the homography and a basic matrix;
multi-view series connection is carried out on the SIFT features to obtain multi-view homonymy points, and then object space point coordinates are obtained through multi-view front intersection positioning;
and constructing a uniform grid in the object space, and screening the point positions to obtain uniform point positions.
10. The device according to claim 9, further comprising the step of optimizing the precision of the selected uniform point locations by using multi-scale least squares matching to obtain coordinates of the connection points after the uniform point locations are obtained.
Technical Field
The invention belongs to the technical field of a map repairing measuring vehicle, and particularly relates to a map repairing measuring method and device.
Background
The map repairing and measuring vehicle integrates advanced technologies such as ground remote sensing measurement, remote sensing image interpretation, accurate combined navigation positioning and the like, is used for rapidly acquiring and processing target space and attribute characteristic information mainly comprising roads and ground objects on two sides, and provides guarantee for repairing and measuring and updating of a map, making of a live-action map and surveying and mapping of special subjects.
The topographic map repairing and measuring vehicle has very important function and significance in improving the rapid repairing, measuring and updating capability of the topographic map in China and meeting the urgent need of acquisition of topographic map information and construction of a database. However, the dependence of the map repairing and measuring vehicle on the road is large, narrow road areas cannot reach, a measuring blind area or a measuring missing problem can be caused, and the data acquisition efficiency is not high. With the satellite positioning attitude determination and vision sensor with low cost and high performance, particularly the non-measurement digital image three-dimensional acquisition technology is gradually improved, and the technical dawn is brought for solving the problem of missing measurement of the blind area of a map repairing and measuring vehicle.
Disclosure of Invention
The invention aims to provide a method and a device for measuring a map repair, which are used for solving the problem that a map repair measuring vehicle generates a measuring blind area or a measuring side leakage in the prior art.
In order to solve the technical problems, the technical scheme of the invention is as follows:
the invention provides a map trimming measurement method, which comprises the following steps:
photographing an area which cannot be reached by the map repairing and measuring vehicle to obtain blind area image data; acquiring GNSS positioning data through GNSS; combining the blind area image data with GNSS positioning data to provide a space coordinate reference by the GNSS positioning data to obtain three-dimensional reconstruction data of the blind area; fusing the obtained three-dimensional reconstruction data of the blind area with existing topographic map data in the topographic map repair and measurement vehicle, and using the fused result for the topographic map repair and measurement; the existing topographic map data in the topographic survey vehicle also provides a spatial coordinate reference by GNSS positioning data.
Furthermore, feature extraction, camera pose estimation and three-dimensional point cloud calculation are carried out by combining the blind area image data and GNSS positioning data, and three-dimensional reconstruction data of the blind area are obtained.
Further, the feature extraction includes: extracting features of blind area image data by using an SIFT invariant feature operator, performing feature block matching by using a homography of low-resolution image estimation, and performing gross error elimination on a matching result by using the homography and a basic matrix; multi-view series connection is carried out on the SIFT features to obtain multi-view homonymy points, and then object space point coordinates are obtained through multi-view front intersection positioning; and constructing a uniform grid in the object space, and screening the point positions to obtain uniform point positions.
Further, after the uniform point locations are obtained, the method further comprises the step of performing precision optimization on the uniform point locations obtained through screening by adopting multi-scale least square matching to obtain connection point coordinates.
Further, the shooting camera pose estimation includes: calculating the phase position of the shooting camera by adopting a multi-view geometric principle; the absolute position of the camera is determined by using a GPS positioning system.
Furthermore, when the obtained three-dimensional reconstruction data of the blind area and the existing topographic map data in the topographic map repairing and measuring vehicle are subjected to fusion processing, the targets on two sides of the road with the existing topographic map data with the basic scale are updated.
The invention also provides a map repair measurement device, which comprises a processor, wherein the processor is used for executing the instructions to realize the following method:
photographing an area which cannot be reached by the map repairing and measuring vehicle to obtain blind area image data; acquiring GNSS positioning data through GNSS; combining the blind area image data with GNSS positioning data to provide a space coordinate reference by the GNSS positioning data to obtain three-dimensional reconstruction data of the blind area; fusing the obtained three-dimensional reconstruction data of the blind area with existing topographic map data in the topographic map repair and measurement vehicle, and using the fused result for the topographic map repair and measurement; the existing topographic map data in the topographic survey vehicle also provides a spatial coordinate reference by GNSS positioning data.
Furthermore, feature extraction, camera pose estimation and three-dimensional point cloud calculation are carried out by combining the blind area image data and GNSS positioning data, and three-dimensional reconstruction data of the blind area are obtained.
Further, the feature extraction includes: extracting features of blind area image data by using an SIFT invariant feature operator, performing feature block matching by using a homography of low-resolution image estimation, and performing gross error elimination on a matching result by using the homography and a basic matrix; multi-view series connection is carried out on the SIFT features to obtain multi-view homonymy points, and then object space point coordinates are obtained through multi-view front intersection positioning; and constructing a uniform grid in the object space, and screening the point positions to obtain uniform point positions.
Further, after the uniform point locations are obtained, the method further comprises the step of performing precision optimization on the uniform point locations obtained through screening by adopting multi-scale least square matching to obtain connection point coordinates.
Further, the shooting camera pose estimation includes: calculating the phase position of the shooting camera by adopting a multi-view geometric principle; the absolute position of the camera is determined by using a GPS positioning system.
Furthermore, when the obtained three-dimensional reconstruction data of the blind area and the existing topographic map data in the topographic map repairing and measuring vehicle are subjected to fusion processing, the targets on two sides of the road with the existing topographic map data with the basic scale are updated.
The invention has the beneficial effects that:
the invention relates to a map repairing and measuring method and a device, which take pictures of the area which can not be reached by a map repairing and measuring vehicle to obtain blind area image data; combining the blind area image data with GNSS positioning data to obtain three-dimensional reconstruction data of the blind area; and the map data are fused with the existing topographic map data in the topographic map repairing and measuring vehicle, and the fused result is used for the topographic map repairing and measuring. The invention adopts the portable data acquisition terminal to measure the blind area which can not be reached by the map repairing and measuring vehicle, is simple and reliable, and can more comprehensively acquire various information such as roads, space objects and the like, thereby enabling the data output by the map repairing and measuring vehicle to be more accurate and meeting the requirements of geological information acquisition and database construction.
Drawings
FIG. 1 is a flow chart of a three-dimensional reconstruction metrology process;
FIG. 2 is a schematic view of a location association;
FIG. 3 is a flow chart of image feature point extraction;
FIG. 4 is a flow chart of GNSS assisted motion recovery;
FIG. 5 is a SIFT point multi-view series diagram based on a union-query data structure;
FIG. 6 is a schematic diagram of the principle of error elimination based on homography matrix and basis matrix;
FIG. 7-1 is a schematic view of image attribute information;
FIG. 7-2 is a schematic view of image attribute information;
FIG. 8 is a diagram of a GNSS coordinate file.
Detailed Description
The invention provides a map repairing and measuring device for realizing the measurement of a blind area of a map repairing and measuring vehicle.
The method adopts a Portable Data Terminal (PDT) to photograph places where a map repairing and measuring vehicle cannot reach, and simultaneously, satellite positioning GNSS Data are synchronously acquired; and processing the acquired data by using a computing unit of the portable data acquisition terminal to obtain a geographic information result, and finally importing the geographic information result into a map repairing and measuring vehicle system for result fusion.
The method adopts a space reference fusion method, wherein the image data of the map repairing vehicle adopts GNSS data to provide space coordinate reference, and the data acquired by the acquisition terminal also adopts the GNSS data to provide the space coordinate reference, so that the seamless fusion of the data acquired by the acquisition terminal and the map repairing vehicle data is ensured.
The method is applied to a map repairing and measuring system, and the system comprises four parts: the system comprises a data acquisition and preprocessing module, a live-action three-dimensional image data management module, a three-dimensional reconstruction measuring module and a multi-source multi-scale data fusion module.
The live-action three-dimensional image data management module manages live-action images, associates the live-action images with the images according to position information, combines the live-action images with matched topographic map data, constructs an image browsing sequence index according to multiple architectures such as sequential index, spatial relationship and customized route, and provides tools such as image marking.
The multi-source multi-scale data fusion module performs fusion processing on a data product obtained by the three-dimensional reconstruction measuring system and existing basic scale topographic map data, updates key targets on two sides of a road of the existing basic scale topographic map data, and provides a multi-source data product for geological information database construction and other requirements.
The data acquisition and preprocessing module is used for combing and associating the non-measuring camera acquired image, the GNSS integrated equipment acquired image and positioning data, the Beidou/vision portable terminal acquired data, camera parameters, route numbers and other information, and simultaneously extracting related data exported by the operation platform of the geographic repairing and measuring vehicle, loading and importing the related data into a measurable live-action image database. The module uses a portable data acquisition terminal to photograph a region which cannot be reached by a map repair and measurement vehicle, so as to obtain blind area image data; meanwhile, images and positioning data are obtained through the GNSS integrated equipment.
The specific indicators of data acquisition are shown in table 1 below.
TABLE 1
The three-dimensional reconstruction measurement module performs a series of operations such as feature extraction and matching, camera pose estimation, three-dimensional point cloud calculation and the like on the acquired blind area image data and GNSS positioning data, and participates in three-dimensional reconstruction measurement, and a processing flow chart thereof is shown in fig. 1.
By means of a monocular image information acquisition system, feature extraction is carried out on the acquired sequence images, the relative position of the camera is solved by utilizing a multi-view geometric principle, the absolute position of the camera is solved by utilizing a self-contained GPS positioning system, the three-dimensional space position of the acquired geographic information data is solved through forward intersection, geographic information is acquired, and the function of measurement is achieved.
When the image is obtained, the accurate position target is obtained through the GNSS positioning system, but the collected data is not continuous points but adjacent single points. The coordinates of the position points stored in the database are not always completely matched with the coordinates of the current position points, and the position points of the images cannot be accurately determined, so that position point information needs to be associated with each image. The current location point is indexed by the database to the nearest image point, and then the image associated with the image point can be obtained.
The image and GNSS position association method comprises the following steps:
when the monocular image information collecting system obtains an image data, the shooting time information is recorded in the attribute information of the live-action image, and the attribute information is shown in fig. 7-1 and 7-2.
The image attribute information includes the date of shooting and the model, aperture value, focal length, etc. of the camera used for shooting. Among such a large amount of attribute information, time information is useful. The time information is a common attribute of the image and the coordinate file, and is a bridge for performing correlation.
The format of the GNSS coordinate file is shown in fig. 8, and the file mainly contains coordinate time information and three-dimensional position information, but the time information of the coordinate file is in a format under a GNSS time system, and unlike the daily use of the GNSS time system, the time information of the coordinate file needs to be converted by two time systems.
The method comprises the following concrete implementation steps:
1) setting a data sampling interval of a GNSS positioning module to be 1Hz, wherein information of each position corresponds to one time information, and acquiring images by using a camera while positioning, so that corresponding images and position information exist at the same time point;
2) acquiring shooting time information of the image by using a function for reading image attribute information;
3) converting the two time coordinate systems, and converting the time coordinate system of the time information of the image into a GNSS time coordinate system;
4) and associating the positions and the images with the same time by using the time information as an index.
Of course, consideration is also given to the case where the satellite signal is out of lock. When the satellite signal is short-lived, the interruption of the positioning data can occur, and for the problem, the position information of the corresponding time can be obtained by a time interpolation method. The principle is as follows:
in acquiring a certain image IiWhen the positioning data of the moment is collected, the time stamp of the image collecting moment needs to be acquired
similarly, find IiDirection of GNSS positioning module at acquisition time
the current location point is indexed by the database to the nearest image point, and then the image associated with the image point can be obtained.
The flow of image feature point extraction is shown in fig. 3.
1) And (3) extracting features by using an SIFT operator, after the SIFT features are extracted, performing feature block matching by using a homography calculated by a low-resolution image in order to avoid the interference of repeated textures, and then performing gross error rejection on a matching result of each block by using the homography and a basic matrix. The using method of the homography matrix and the basic matrix is as follows:
the base matrix F is a 3 × 3 matrix of
x′TFx＝0
for object space points distributed on the same plane, after imaging on different images, image points have a one-to-one correspondence relationship, and the constraint relationship is a homographic transformation relationship and is usually represented by a homographic transformation matrix H. If the imaging coordinates of the object point on the stereo pair are x and x', respectively, the following relationship is satisfied:
x′＝Hx
and (3) comprehensively utilizing the matching point relation equation of the basic matrix F and the homography transformation matrix H, and eliminating gross error points in the matching result by adopting a robust estimation method of RANSAC.
2) And multi-view series connection of SIFT features is realized by adopting a data structure, and object space point coordinates are obtained through multi-view front intersection.
3) Constructing an even grid in an object space, and screening point locations of object points falling into the grid by using an index V value considering precision and point location overlapping degree to obtain even point locations, wherein the index V value is defined as follows:
wherein phi isiFor the projection error of the ith object point,is the average projection error of all object points, OiIs the overlap of the ith point.
And screening the point locations according to a set threshold and a certain evaluation method to ensure the relative uniformity of the point locations.
4) And optimizing the coordinates of the screened connection points by adopting multi-scale least square matching.
A method for estimating image pose parameters under the assistance of GNSS (global navigation satellite system) is constructed by adopting a robust estimation method of image global consistent rotation parameters and a unified fusion strategy of the rotation parameters and the GNSS parameters, which is essentially a global motion recovery structure method under the assistance of GNSS data, and the flow chart is shown in FIG. 4. Therefore, the relative achievement of the three-dimensional mapping of the operation platform of the map repairing and measuring vehicle can be obtained, the three-dimensional reconstruction measurement is participated in, the three-dimensional reconstruction of two sides of the road is finally realized, and the live-action map data and the key target measurement data are provided.
- 上一篇：一种医用注射器针头装配设备
- 下一篇：使用多次三维扫描进行对象检测