CN112907724B - An interactive automatic mapping method for buildings using oblique photography of drones - Google Patents

Abstract

The invention relates to an interactive automatic mapping method for buildings using oblique photography of drones. On the basis of three-dimensional model loading and coordinate system selection, building type selection is performed by responding to global keyboard keys, and horizontal projection is used to calculate the intersection of surfaces Carry out feature point collection, use two-dimensional straight line intersection point and vertical foot method to collect model occlusion or deformation areas, and complete building automatic mapping through CASS code and simple code recognition. The invention relates to a building interactive automatic mapping method using UAV oblique photography, which introduces the plane and plane intersecting projection algorithm and the straight line intersecting algorithm into the collection of building feature points, which solves the problem of accuracy of model point collection Problems and model occlusions and deformations have issues with missing data.

Description

An interactive automatic mapping method for buildings using oblique photography of UAV

technical field

The invention relates to the field of surveying and mapping geographic information, in particular to an interactive and automatic building mapping method using oblique photography of a drone.

Background technique

Buildings are one of the most widely distributed and densely man-made features in the production and living areas of human beings. The floor plan of buildings is an important basic data in the construction of smart cities. The 3D model reconstructed by low-cost UAV oblique image matching has become an effective means to quickly obtain 3D digital models of urban buildings and large-scale mapping of real estate integration confirmation and registration. Mapping plays a greater role, and how to use the 3D model of UAV oblique photography to quickly and accurately obtain building information, automatically complete the drawing of building plans, and provide basic data support and scientific progress for the construction of urban and real estate integration projects quickly and accurately. Urban planning is of great significance in providing decision-making and solving the contradiction between rapid economic development and land use. Due to the complexity of real-world buildings, there is currently no mature method for automatic building mapping of 3D models using UAV oblique photography.

Existing architectural mapping methods fall into four categories:

One class uses traditional surveying and mapping mode to confirm the right of contracted management of rural houses and rural land, mainly relying on single-point measurement, through total station with GNSS (Global Navigation Satellite System, Global Navigation Satellite System) RTK (Real-time kinematic, real-time Dynamic carrier phase difference technology) for field data collection can meet the accuracy requirements of large-scale mapping, but there are disadvantages such as large field workload, low data collection efficiency, and large constraints from the field collection environment, and can only obtain limited data. Building feature point data will also face problems such as on-site data collection and data verification;

The second category is the use of high-precision satellite remote sensing images to provide an important technical means for large-scale building mapping, but due to spatial resolution, satellite revisit cycle, and images only have plane features on the sensor focal plane, lack of spatial information, Unable to truly reproduce the 3D scene and other reasons, it is difficult to meet the needs of high-precision architectural plan drawing;

The third category is the use of airborne LiDAR (Light Detection And Ranging) technology for building extraction and mapping and other related research, but its wide-scale application at this stage is still largely restricted by expensive data acquisition costs and strong professional operation;

The fourth category is the use of UAV oblique photography for 3D reconstruction and building mapping. Thanks to its cost advantage, the method of reconstructing 3D models based on oblique image matching for architectural mapping has become a method for quickly obtaining urban building line drawings and large-scale drawings. Effective means of mapping, the current internal industry mapping software mainly includes EPSW jointly developed by Tsinghua Shanwei Company and the Department of Civil Engineering of Tsinghua University, RDMS of Wuhan Ruide Surveying and Mapping Automation Company, and CASS and CASS3D series of Nanfang Surveying and Mapping Instrument Company. However, at this stage, it is still constrained by the rapid and automated architectural mapping method. There are problems such as high labor intensity in the office, low degree of automation in architectural mapping, low acquisition accuracy of model occlusion or deformed areas, and lack of architectural graphic attributes.

Contents of the invention

The technical problem to be solved by the present invention is to overcome the defects of the prior art, provide an interactive automatic mapping method for buildings using oblique photography of UAVs, and solve the problem of the existing automation degree of building mapping based on UAV oblique photography 3D models It overcomes the problems of low precision and high labor intensity in the internal industry, and overcomes the shortcomings of low accuracy in the collection of building feature points in model occlusion or deformation areas.

In order to solve the problems of the technologies described above, the technical solution of the present invention is:

A method for building interactive automatic mapping using oblique photography of UAVs, comprising the following steps:

1) Load the 3D model in osgb format, and select the coordinate system required for the final mapping result;

2) Correspond to the house type formed by the house drawing command DRAWDDF(FF) corresponding to CASS, set different global keyboard response events to correspond to each other, and if they correspond, the interactive single building collection mode will be turned on;

3) The collected architectural feature points: the architectural feature points are located on the straight line obtained by the intersection of two planes. If there are two adjacent non-parallel planes in the space, the two adjacent planes must intersect to obtain a straight line L, which is is the building feature line, project the feature line L onto the horizontal plane, which is the building feature point to be collected;

4) Assume that the two adjacent planes in space are P1 and P2 respectively, and the normal vectors corresponding to the two adjacent planes are respectively but:

It can be seen from the knowledge of space geometry that the intersection line L is perpendicular to the two plane normal vectors, that is, the direction of the intersection line L is . In the case of known intersection line direction, in order to accurately determine the spatial position of the intersection line, it is necessary to know the intersection line A known point, the point set on the intersection line L satisfies:

Combining the two formulas together, we get:

Solving a and b, we get:

Then the formula of the intersection line L of two adjacent planes in space is:

5) Project the intersection line L onto the horizontal plane to obtain the architectural feature points for automatic mapping. In the collection of architectural feature points, for places where the feature lines in the 3D model are obvious, use the mouse to interact with the adjacent plane intersection lines of the building. Directly obtain two-dimensional feature points of buildings for automatic mapping;

6) If the building area is blocked, perform classification calculation:

Adjacent walls do not intersect perpendicularly:

Assume that L1 and L2 are the projections of two adjacent walls in the horizontal plane, and the two planes are not vertically intersected, but the architectural feature points cannot be obtained obviously due to the occlusion or deformation of the model;

The straight lines L1 and L2 can be determined by two points on the projection plane where they are respectively, expressed as:

a ₁ x+b ₁ y+c ₁ =0

_a2x + _b2y + _c2 =0;

The calculated building feature point coordinates can be obtained by combining the two formulas:

Adjacent walls intersect perpendicularly:

Assuming that P1 and P2 are two adjacent walls that intersect vertically, the projection lines of the two planes are L1 and L2, and the intersecting part is occluded or deformed by the model. At this time, the plane coordinates of the architectural feature points to be calculated are transformed into the method of finding a point to a straight line. drop feet

Assuming that the coordinates of a point on the straight line L2 are (x2, y2), the calculated coordinates of the architectural feature points are:

Through the above calculation, the coordinates of building feature points in the occlusion area can be obtained. When the collection of a building feature point is completed, the global keyboard response events are set to correspond to each other, and the interactive keyboard design is used to complete the collection of a building's feature points;

7) Define the JCODE.DEF encoding file, express different house types with different codes, and realize the collection and coding of different types of buildings by responding to different keyboard events. When a building is collected, you can get the CASS code data files;

8) Through CASS software drawing processing -> simple code recognition, the automatic generation of oblique photographic building line drawing can be completed;

As an improvement, the global keyboard response events in the step 2) include general houses, concrete houses, brick houses, iron houses, steel houses, wooden houses, mixed houses, simple houses, construction houses, destroyed houses and sheds.

The beneficial effects of the present invention are:

The plane and plane intersecting projection algorithm and the straight line intersecting algorithm are introduced into the collection of building feature points, which solves the problem of the accuracy of model point collection and the problem of missing data in model occlusion and deformation. Man-machine oblique photography 3D model building is automatically drawn. The invention uses interactive building type selection, uses surface intersection and horizontal projection to collect feature points, and adopts two-dimensional straight line intersection point and vertical foot method for model occlusion or deformation areas. , through the recognition of CASS codes and short codes to complete the automatic drawing of buildings.

The key technology in the automatic drawing of building plans by low-cost UAV oblique photography 3D model is closely related to the overall trend of diversification and intelligence in my country's current 3D data acquisition methods, and can provide rapid construction of rural real estate registration and confirmation buildings Mapping, "Digital City" and "Smart City" provide references for automated processing methods and basic data support. Taking full advantage of the low-cost advantages of UAV oblique photogrammetry and the advantages of rich facade data and spectral characteristics, it can improve the automation and intelligence of point cloud data processing based on oblique photography, and improve the ability of object cognition and knowledge-based services. Enriching and perfecting the method theory of rapid automatic processing of 3D models based on UAV oblique photography has a certain promotion effect, and to a certain extent alleviates the contradiction between the rapid development of hardware technology in the field of 3D automatic processing and relatively lagging data processing. It overcomes the shortcomings of the existing architectural mapping method based on the oblique photographic model of the UAV, such as high labor intensity in the office, low degree of automation in architectural mapping, low acquisition accuracy of model occlusion or deformation areas, and lack of architectural graphic attributes, etc., and improves the internal industry. Mapping work efficiency.

Description of drawings

Fig. 1 is a flowchart of the present invention;

Fig. 2 is a schematic diagram of building characteristic lines of the present invention;

Fig. 3 is a schematic diagram of the non-perpendicular intersecting situation of adjacent wall surfaces of the present invention;

Fig. 4 is a schematic diagram of the vertical intersection of adjacent walls of the present invention;

Fig. 5 is the man-machine oblique photography three-dimensional model data shooting figure of the present invention;

Fig. 6 is the point cloud data diagram of the experimental area of the present invention;

Fig. 7 is a building plan automatically generated by the experimental area of the present invention.

Detailed ways

In order to make the content of the present invention more clearly understood, the present invention will be further described in detail below based on specific embodiments and in conjunction with the accompanying drawings.

As shown in the figure, an interactive and automatic building mapping method using UAV oblique photography. This method is based on 3D model loading and coordinate system selection, and responds to global keyboard keys to select building types. Feature points are collected by intersecting and horizontal projection, and two-dimensional straight lines are used to collect intersection points and vertical foot methods for model occlusion or deformation areas. The automatic building map is completed through CASS coding and simple code recognition. The flow chart is shown in Figure 1, and the steps are as follows :

Step 1: Load the 3D model in osgb format, and select the coordinate system (spatial reference system) required for the final mapping result, for further use of oblique photography 3D model collection and automatic generation of architectural line drawings;

Step 2: Correspond to the house type formed by the house drawing command DRAWDDF (FF) corresponding to CASS, and set different global keyboard response events corresponding to: (1) general house (2) concrete house (3) brick house (4) Iron House (5) Steel House (6) Wooden House (7) Mixed House (8) Simple House (9) Construction House (10) Destroyed House (11) Shed, if the corresponding button is pressed, the interactive single house The building collection mode is turned on;

Step 3: In a real 3D mapping scene, the most common architectural feature points are located on the straight line obtained by the intersection of two planes. If there are two adjacent non-parallel planes in the space, these two adjacent planes must intersect to obtain a straight line. The line L, the straight line L is the building feature line, as shown in Figure 2, the feature line L is projected onto the horizontal plane, which is the building feature points that need to be collected.

Assuming that the two adjacent planes in space are P1 and P2 respectively, the normal vectors corresponding to the two adjacent planes are respectively but:

It can be seen from the knowledge of space geometry that the intersection line L is perpendicular to the two plane normal vectors, that is, the direction of the intersection line L is . In the case of known intersection line direction, in order to accurately determine the spatial position of the intersection line, it is necessary to know the intersection line A known point, the point set on the intersection line L satisfies:

Combining the formulas, we get:

Solving a and b, we get:

Then the formula of the intersection line L of two adjacent planes in space is:

Project the intersection line L onto the horizontal plane to obtain the architectural feature points for automatic mapping. In the collection of architectural feature points, for places with obvious feature lines in the 3D model, interact with the adjacent plane intersection lines of the building with the mouse to directly obtain Two-dimensional feature points of buildings for automatic mapping;

Step 4: Due to tree occlusion, image acquisition angle of view, and 3D modeling software processing algorithms, etc., in real 3D mapping scenes, occlusion or deformation of building feature lines obtained through oblique photography and 3D modeling will inevitably occur In this case, the method of intersecting the horizontal projection of straight lines in adjacent planes is used to obtain the architectural feature points. The calculation of building feature points in the occlusion area is divided into two cases:

(1) Adjacent walls are not perpendicular to each other

As shown in Figure 3, L1 and L2 are the projections of two adjacent walls in the horizontal plane. The two planes are not vertically intersected, but the architectural feature points cannot be obtained obviously due to the occlusion or deformation of the model.

The straight lines L1 and L2 can be determined by two points on the projection plane where they are respectively, expressed as:

a ₁ x+b ₁ y+c ₁ =0

a ₂ x+b ₂ y+c ₂ =0

Combine formula (6) and formula (7) to get the calculated building feature point coordinates:

(2) Adjacent walls intersect vertically

As shown in Figure 4, P1 and P2 are two adjacent walls that intersect vertically, but the intersecting part is occluded or deformed by the model. At this time, the plane coordinates of the architectural feature points to be calculated are transformed into finding the vertical foot of a point to a line.

Assuming that the coordinates of a point on the straight line L2 are (x2, y2), the calculated coordinates of the architectural feature points are:

Through the above calculation, the coordinates of building feature points in the occlusion area can be obtained. When the collection of a building feature point is completed, the global keyboard response events are set to correspond to each other, and the interactive keyboard design is used to complete the collection of a building's feature points;

Step 5: Define the JCODE.DEF encoding file when collecting in the office, and use different codes to represent different building types, as shown in Table 1. By responding to different keyboard events, different types of buildings can be collected and encoded , when a building is collected, a data file with CASS encoding can be obtained. Table 1:

Through CASS software drawing processing -> short code recognition, the automatic generation of oblique photographic building line drawings can be completed.

Example:

Select experimental buildings: including 2# experimental building, 3# experimental building, practice factory, 1# to 6# teaching building, library and other buildings, the teaching building is 5-6 floors, the experiment building is 5 floors, and the practice factory It is a 1-story iron house, using a six-rotor UAV, equipped with a five-lens tilting camera, in which the camera sensor size is 13.2mm×8.8mm, the focal length is 10.4mm, the heading and side overlap are set to 75%, and the horizontal flight speed 6m/s, the experimental data is collected, and the processed 3D model of the UAV oblique photography in the experimental area is shown in Figure 6.

After model loading and coordinate system selection, interactive building type selection, building feature point collection, and CASS code and brevity code recognition, the automatically generated building map of the experimental area is shown in Figure 7.

It should be understood that the above descriptions are only specific embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be Included within the protection scope of the present invention.

Claims (2)

1. A building interactive automatic mapping method using unmanned aerial vehicle oblique photography, comprising the steps of:

1) Loading a three-dimensional model in an osgb format, and selecting a coordinate system required by a final imaging result;

2) In order to correspond to the house type formed by the house drawing command DRAWDDF (FF) corresponding to the CASS, different global keyboard response events are set to correspond to: general houses, concrete houses, brick houses, iron houses, steel houses, wood houses, mixed houses, simple houses, building houses, destructive houses and shed houses, and if a corresponding key is pressed, the interactive single building acquisition mode is started;

3) Collected building characteristic points: if the space has two adjacent non-parallel planes, the two adjacent planes are intersected to obtain a straight line L, wherein the straight line L is a building characteristic line, and the characteristic line L is projected onto a horizontal plane to obtain the building characteristic point to be acquired;

4) Assuming that two space adjacent planes are P1 and P2 respectively, the normal vectors corresponding to the two adjacent planes are respectivelyThen:

the space geometry knowledge shows that the intersecting line L is perpendicular to two plane normal vectors, namely, the direction of the intersecting line L is that, under the condition of knowing the intersecting line direction, in order to accurately determine the space position of the intersecting line, a known point on the known intersecting line is needed, and the point set on the intersecting line L satisfies:

combining two formulas to obtain:

solving a and b to obtain:

the formula of the intersection line L of two adjacent planes in space is:

5) Projecting intersection lines L onto a horizontal plane to obtain building feature points for automatic imaging, and in the process of building feature point acquisition, interacting adjacent plane intersection lines of a building through a mouse at a place with obvious feature lines in a three-dimensional model to directly obtain two-dimensional feature points of the building for automatic imaging;

6) If the building area is blocked, performing classified calculation:

adjacent wall surfaces are not vertically intersected:

assuming that L1 and L2 are projections of two adjacent wall surfaces in a horizontal plane, wherein the two planes are not vertically intersected, but the intersection part is blocked or deformed by a model, so that building characteristic points cannot be obviously obtained;

the straight lines L1 and L2 are respectively determined by two points on the projection plane where they are located, and are expressed as:

a ₁ x+b ₁ y+c ₁ ＝0

a ₂ x+b ₂ y+c ₂ ＝0；

the calculated building characteristic point coordinates can be obtained by the simultaneous formulas:

adjacent wall surfaces vertically intersect:

assuming that P1 and P2 are two adjacent wall surfaces and vertically intersect, projection lines of the two planes are L1 and L2, and the intersection part is shielded or deformed by a model, so that the plane coordinates of the building feature points to be calculated are converted into vertical feet from one point to a straight line;

assuming that the coordinates of a point on the straight line L2 are (x 2, y 2), the calculated building feature point coordinates are:

through the calculation, building feature point coordinates of the shielding area can be obtained, when the collection of one building feature point is finished, global keyboard response events are set to be respectively corresponding, and the collection of the feature point of one building is completed by utilizing the interactive keyboard design;

7) Defining a jcode/def coding file, representing different house types by different codes, realizing collection and coding of different types of buildings by responding to different keyboard events, and obtaining a data file with CASS codes after the collection of one building is finished;

8) And (3) automatically generating the oblique photography building line drawing through CASS software drawing processing- > brief code identification.

2. The method for interactive automatic mapping of a building using unmanned aerial vehicle tilt photography according to claim 1, wherein the global keyboard response event in step 2) comprises a general room, a concrete room, a brick room, an iron room, a steel room, a wood room, a mixed room, a simple room, a building room, a destructive room, and a shed room.