RESEARCHING ON REAL 3 D MODELING CONSTRUCTED WITH THE OBLIQUE PHOTOGRAMMETRY AND TERRESTRIAL PHOTOGRAMMETRY

With the rapid development of the oblique photogrammetry, many cities have built some real 3D model with this technology. Although it has the advantages of short period, high efficiency and good air angle effect, the near ground view angle of these real 3D models are not very good. With increasing development of smart cities, the requirements of reality, practicality and accuracy on real 3D models are becoming higher. How to produce and improve the real 3D models quickly has become one of the hot research directions of geospatial information. To meet this requirement In this paper, Combined with the characteristics of current oblique photogrammetry modeling and the terrestrial photogrammetry, we proposed a new technological process, which consists of close range sensor design, data acquisition and processing. The proposed method is being tested by using oblique photography images acquired. The results confirm the effectiveness of the proposed approach.


INTRODUCTION
The oblique photography technology has been a new technology developed in recent years.With several cameras on one air platform, it can obtain the top and sides data of the buildings synchronously.This new technology has improved the limitation of the traditional photogrammetry because of the high overlapped and multi-view data (Yang Guodong, Wang Mingshui).The real 3D model constructed by this technology can meet some requirement of the smart city construction.Until now, it has been used for emergency command, homeland security, urban management etc.  １ Han Youmei, School of Geomatics and Marine Information, Huaihai Institute of Technology, Lian Yungang, Jiangsu, China.Email:hanyoumei@126.com ， Funded by Beijing Key Laboratory of Urban Spatial Information Engineering No.2018213，and Key Laboratory of Precise Engineering and Industry Surveying, National Administration of Surveying, Mapping and Geoinformation No.PF2017-3 At present, the application of oblique photogrammetry technology has been widely carried out, and the real 3D model constructed by it has gradually become an important part of the urban spatial data framework.However, the near ground parts of the real 3D model constructed by ithas some deformation(Qu Lin etc), texture fuzzy, which can't meet the higher requirement of the smart city construction.We intended to improve the real 3D by combining the terrestrial photogrammetry data with the oblique photogrammetry.

The design of the process
The oblique photogrammetry is good at the top and some high sides real 3D model construction.According to the oblique photogrammetry principle we designed one new ground multi-view images acquisition device.By using this new device we acquire some ground and building sides images, after that, ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume IV-3, 2018 ISPRS TC III Mid-term Symposium "Developments, Technologies and Applications in Remote Sensing", 7-10 May, Beijing, China we combined the oblique photogrammetry images with the terrestrial data to construct the more perfect real 3D model, which is better in the geometry and the texture.2) The overlap of the adjacent cameras view should be not less than 30%; 3) It can keep these six camera obtain images synchronously and can obtain data by some distance or some time intervals; 4) There is one GPS antenna on the top of the design which enable us to get the image position by it; The design draft of the new sensor (a) The physical equipment of the design (b) The control device of the new sensor(c) No.6 camera is defined as the left back camera.According to the photogrammetry principle, the overlap between two adjacent cameras is not less than 30%.to acquire the modeling data , the new device is installed on the vehicle top.And the images overlap between the directions of the left or right camera is around 80%, and a certain sampling interval is set to ensure this condition.

2.2.2The control points layout scheme
There are some papers about the oblique photogrammetry control point layout scheme (Hu wenqing, Zhou Xiaomin etc.), but in order to carry out the AT with the near ground and aerial images together, these schemes can't be used directly.So a new kind of control point layout scheme was proposed.There are at least four control points per square kilometer, most of the control points are on the ground and the road intersection.The most important is all of the control points should be seen from the aerial images and the terrestrial images.

The overview of the experimental area
The experimental area is a general park with many low and medium height buildings in it.The air condition is good at acquiring data by oblique photogrammetry.And the roads are very convenient for the terrestrial data acquisition.The area is about 1 square kilometers.

Data acquisition
The design ground resolution of oblique photogrammetry is  The trajectory of the terrestrial image collection in the experimentation area is as follow.
Figure 5 Trajectory of the near ground image acquisition

The Control Survey
The control measurement is carried out by network RTK (Figure 6), and three measurements are measured at each point, and the average value is taken as the final result.
ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume IV-3, 2018 ISPRS TC III Mid-term Symposium "Developments, Technologies and Applications in Remote Sensing", 7-10 May, Beijing, China Figure 6 The Control points survey scene

The data processing
The data processing consists of image preprocessing, the input of position files of images, AT and the modeling.It is noted that the shared control points are added to carry out AT combined the aerial images with the ground images.In data processing, the software Context Capture (Context Capture User Manual version 4.4.) is used to carried out AT and modeling.

Figure 1
Figure 1 The integration modeling process flow chart Because of the different angel and resolution of the air and terrestrial images, direct aerial triangulate (AT) with oblique photogrammetry data and the ground images is difficult.So in this paper, both the air images and terrestrial images need to operate AT respectively, after that , the two results above are combined to do AT again to get refine result ,finally the refined result are used to model.(figure 1 )

Figure 2
Figure 2 The design and the physical figures of our new device After some experiment, we chose six cameras to set up the new device.No.1 camera is defined as the left camera and No. 2 camera is defined as the left front camera.No.3 camera is defined as the right front camera and No. 4 camera is defined as the right camera.No.5 camera is the right back camera, and about 2cm.The sensor for oblique photogrammetry is one aerial device named ARC 524(Figure.3) made by Shanghai HangYao information technology company, which consists of six cameras.The focal length of the downward camera is 20mm, and the focal length of the side view cameras is 35mm.The flight trajectory in experimentation area is figure 4.

Figure 3 Figure 4
Figure3The Unmanned Aerial Vehicle and the aerial camera

Figure 7
Figure 7 The model constructed only by the oblique photogrammetry