Dissertation > Excellent graduate degree dissertation topics show

Researches on Key Technologies about Three-dimensional Reconstruction of Buildings Based on Omni-directional Image and Remote-sensing Image

Author: YangMingFan
Tutor: WangZuo
School: National University of Defense Science and Technology
Course: Control Science and Engineering
Keywords: Three-dimensional reconstruction Digital City Registration Highly extracted Omnidirectional Figure Remote Sensing Figure
CLC: TP391.41
Type: Master's thesis
Year: 2011
Downloads: 19
Quote: 0
Read: Download Dissertation

Abstract

The city is generally local political, economic and cultural center. Build a digital city for the city's sustainable development strategy in the civilian aspects of information protection helps intelligent transportation, urban management and urban development decision-making; military, is the basis of building a the city virtual battlefield environment, help fight win the future of urban warfare. The three-dimensional reconstruction of urban buildings is an important part of building digital cities, have a wide range of applications in the field of civil and military value. Fold reflection omnidirectional Figure vision larger horizontal direction can reach 360 degrees, imaging access to a wider range of scenes. In view of the lack of full to the advantage of the existing reconstruction methods, this paper presents a vision based on omnidirectional Figure with remote sensing graph reconstruction of city buildings, and one of the key technical issues. First, with reference to the steps of the general three-dimensional reconstruction, for the whole map and remote sensing characteristics, the reconstruction process is divided source image acquisition, image preprocessing, omnidirectional Figure and Remote Sensing Figure registration, building height extraction, the exterior of the building The six-part, pursuant to which the shape of the contour modeling and texture mapping framework and steps of the reconstruction. Secondly, we study the whole map to map and Remote Sensing registration problem, two registration methods. Omnidirectional Figure with remote sensing geometric constraint relations, we first identify and prove the isometric principle of the angle between the horizontal straight line within the description of these two images, and then as a basis for the design of the first registration method. . The extraction of buildings horizontal contour is characterized, in a straight line on the rotation angle equal to the criterion, the search for an estimated point set to identify feasible point from the feasible point, and finally by the statistical characteristics to identify the most satisfactory solution to complete registration. In addition to the isometric principle, to identify and prove real space description omni-dot distance to the optical axis of the image point and the point corresponding to the midpoint of the relationship between the distance to the optical axis of the camera full nature to the distance. In accordance with the principle of this nature with isometric design a second registration method. Decision of the principle of law through isometric angle relationship with omni-directional the shortest distance nature from the decision vertical relationship solved the whole position of the picture was taken, the final completion of the registration. The experimental results show that the two alignment method is rapid and effective, compared with the first method is more robust, and the second method is more rapid. Finally, research building height extraction problem. Omnidirectional principle Figure imaging optical path, the proposed height extraction algorithm based on the optical path tracking algorithm will highly extracted abstract geometric problems by calculating the geometry of the side length to achieve the purpose of the strike the building height. The experimental results show that this method is fast and effective.

Related Dissertations

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device
© 2012 www.DissertationTopic.Net Mobile