| DC 欄位 |
值 |
語言 |
| DC.contributor | 土木工程學系 | zh_TW |
| DC.creator | 蔡函芳 | zh_TW |
| DC.creator | Han-Fang Tsai | en_US |
| dc.date.accessioned | 2009-7-20T07:39:07Z | |
| dc.date.available | 2009-7-20T07:39:07Z | |
| dc.date.issued | 2009 | |
| dc.identifier.uri | http://ir.lib.ncu.edu.tw:88/thesis/view_etd.asp?URN=963202084 | |
| dc.contributor.department | 土木工程學系 | zh_TW |
| DC.description | 國立中央大學 | zh_TW |
| DC.description | National Central University | en_US |
| dc.description.abstract | 數碼城市為虛擬之三維都市，用以描述物空間環境特徵，其構成元素包含房屋、道路與植被等，其中，房屋為最重要的元素。重建三維房屋模型，最常見的資料來源之一為航照影像。使用航照影像重建三維房屋模型，處理重點為房屋區塊偵測及影像匹配。本研究結合單航帶高重疊數位航照影像以及大比例尺向量圖，進行房屋邊緣線之三維定位，處理重點為影像匹配。測試例房屋包含山型屋、平頂屋及具有屋頂附加物之人工建物。研究中，使用大比例尺向量圖提供的房屋外圍輪廓以及建物樓層數，簡化房屋區塊偵測以及給予影像匹配較佳之初始位置。高重疊率的數位航照影像，其前後連續影像紋理具有高相似性，輔以高程幾何約制，以提升影像匹配之效率及正確性。此外，高重疊航照影像對於空間三維定位給予較多觀測量，因此，研究中對於影像匹配採用序列式匹配以及整合式匹配兩種方法，並配合多視窗及單視窗，欲尋求提升匹配可靠度及定位精度較好之匹配模式，並針對整合式匹配之共軛點微調進行分析。
整體而言，整合式匹配其定位精度較序列式匹配佳。而使用多視窗對於房屋邊緣線的匹配也較單視窗佳。
| zh_TW |
| dc.description.abstract | Cyber city is a virtual replica that consists of buildings, roads and vegetation. Among those elements, building is the most important element in cyber city. Aerial images are commonly used in building reconstruction. When using aerial images in building reconstruction, building detection and image matching are the two major works. This study uses highly overlapped aerial images and large-scale vector maps to perform 3D positioning for building boundaries. Image matching is the major work in this study. Test cases include different types of building’s roof. In this study, the large-scale vector maps provide the outline and number of building storey to simplify the building detection. Height estimation from the number of building storey is used as height constraint in image matching. Highly overlapped images lave high similarity between adjacent images, which could increase the reliability in image matching. Besides, multi-image provides more redundancy in space intersection, in looking for a better strategy to increase the reliability and positioning accuracy, sequential image matching and multiple image matching are used in this research. The refinement of multiple matching is also to be verified the contribution.
The results show that the accuracy in multiple matching is better than sequential matching, and multi-window for image matching is better than single window.
| en_US |
| DC.subject | 序列式匹配 | zh_TW |
| DC.subject | 整合式匹配 | zh_TW |
| DC.subject | 多視窗匹配 | zh_TW |
| DC.subject | 影像匹配 | zh_TW |
| DC.subject | 房屋重建 | zh_TW |
| DC.subject | sequential matching | en_US |
| DC.subject | multiple matching | en_US |
| DC.subject | image matching | en_US |
| DC.subject | building reconstruction | en_US |
| DC.title | 多重影像匹配於房屋邊緣線三維定位 | zh_TW |
| dc.language.iso | zh-TW | zh-TW |
| DC.title | Multiple Image Matching in 3D Positioning for Building Boundaries | en_US |
| DC.type | 博碩士論文 | zh_TW |
| DC.type | thesis | en_US |
| DC.publisher | National Central University | en_US |