碩士 / 國立中興大學 / 土木工程學系所 / 101 / With the development of surveying technologies, panoramic street views allow people to get the premier geographical and spatial information through computers and intelligent devices. With the increasing demand of online route planning, an efficient way to achieve the complete street views along with the navigation is established in this research.
A Panoramic image based route planning will also be built in this project. This function enables users to explore more geo-spatial information to fulfill the purpose of travel planning through streaming thousand or even more conjunctive panoramic images. To enrich the performance of this service, multi-functions, such as spatial coordinate interpolation, azimuth angle transformation, plane road boundary limit, elevated road boundary limit, are established in the system. Through requiring these spatial based functions, a near real-time image alignment technique for route planning will be presented, without the need of employing any on-field investigation. Finally, a video by streaming panoramic images will be made for previewing the street views along with the moving direction automatically.
To achieve spatial information in a street view automatically, we integrate some image processing technologies, such as Template Matching and NCC (Normalized Cross Correlation), to detect conjugate points on a panoramic image coordinate system, which is built of two or three conjunctive panoramic images. Then, we employ spatial intersection relationship formulas to create links between the panoramic image coordinate system and geographical coordinate system. Finally, we provide an automatic model for surveying measurement through close-range photogrammetry.
Functions adapted in the streaming of panoramic images, such as the azimuth angle transformation, fixed 20 to 150 degrees of horizontal angle errors. Through the plane road boundary limit, we removed 11.56% incorrect street view points, improved the performance of this system.
The other function in this system, the interesting point positioning, could reach to an approximately 73% value of precision. Using these positioning results, we acquired a 56% probability to obtain the correct geographical information. By providing a plane error range of about 5 to 10 meters, this automatic positioning model is suitable for general streetscape building measurement.
| Identifer | oai:union.ndltd.org:TW/101NCHU5015067 |
| Date | January 2013 |
| Creators | Hao-Ping Wu, 吳浩平 |
| Contributors | Ming-Der Yang, 楊明德 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 65 |
Page generated in 0.0072 seconds