2002-08-012024-05-18https://scholars.lib.ntu.edu.tw/handle/123456789/712314摘要：攝影測量之目的在於有效地測取空間三維資訊。在攝影測量流程中雖已發展高度自動化之方位求解的空中三角軟體，然而接續的物空間資訊萃取仍需要投入大量的人力。近十年來數位攝影測量借助於影像處理技術及電腦視覺之判識能力、自動化特徵物偵測及匹配技術的改良， 許多不完整之特徵物資訊，例如點、線、面之幾何資料，隨著上述流程產生。究竟這些多樣式、多維的特徵物能為攝影測量作業帶來何種程度之助益? 本計劃為同一主題之第二階段計劃，為一應用導向之研究。從正在進行的第一階段工作推展及初步成果顯示，利用影像特徵物實證上確能達到方位求解及物型重建之同步功效。從模擬資料及實測資料上，已建立利用共面之直線特徵物，達到方位求解及面狀特徵物(在本例中為屋頂面)之物型重建之任務。因此，考量以為期兩年時間進行第二階段之工作將包含解析非直線之線狀特徵物(即物空間之三維曲線)、面狀特徵物於攝影測量上之應用，並結合傳統之點特徵物，建立一套以特徵物萃取、特徵物處理及特徵物三維重建之數值攝影測量求解系統。本系統之特點在於結合影像處理、電腦視覺及攝影測量技術，發展一減少人力量測、判識之三維物空間資訊測取系統。 <br> Abstract: The main task of photogrammetry is for efficiently measuring and collecting three dimensional spatial information for extensive area. As the revolution of technology has led to the success of highly automatic aerial-triangulation system solving for the exterior orientation on point feature basis, making spatial information available still requires heavy labor effort by manually digitizing in a stereopsis afterwards. During the past decade, digital photogrammetry benefiting from integrating the technologies of image processing and computer vision has seen the easy and efficient way for detecting versatile feature entities, including point, line and area ones, throughout the photogrammetric workflow. It is then interesting to think about how photogrammetry would take advantages of including such type of data into processing. This proposal, phase two, is a follow-up research, yet attempted with more concerns in application than that in phase one. The preliminary result of on-going study shows that incorporated with linear features under coplanar constraint the routine of photogrammetry is able to answer for the orientation as well as partial reconstruction of object in three dimensions, which truly encourages us the further step involving more research effort in this topic. We propose the following tasks in two-year duration: In the first year, major study will be emphasized on investigating the mechanism of feature detectors, mainly for free-form curvilinear and area or patch-like features, and the correspondences of above conjugate features among overlapping images. In addition, the functional relationship of these features between image space and object space must be explicitly expressed in order for estimating exterior orientation and performing object reconstruction of related features. In the second year, a system based on the sub-models developed and tested will be integrated and evaluated so that the beauty of feature-based photogrammetry, cleverly aided by the skill of image processing and computer vision, is hoping to be appreciated with the performance of object reconstruction done in this project.空中三角特徵物影像處理電腦視覺Aerial-triangulationFeatureImage rocessingComputer vision