基本矩陣之最佳化與歐幾里德重建

Abstract

由影像重建三維模型在電腦視覺中仍是十分重要的研究題目，幾年以前主要應用在機器人導航與視覺監控，然而現今在電腦繪圖、虛擬實境與通訊等其他方面方面的需求日益增多，導致需求重點的改變，視覺品質成為應該注意的重點。本篇論文主要工作在於實現一些將三維物體重建回歐氏空間的可行方法，其研究重點則為透過改善基本矩陣之計算以估測更精準的相機內外部參數，來達到重建的目的。

3-D model reconstruction from object images is an ongoing research topic in computer vision. A few years ago, the main applications were in robot guidance and visual inspection. Nowadays, there are more and more demands for 3-D models in computer graphics, virtual reality, and communication. This results in emphasis for the requirement of higher reconstruction quality. In this paper, we implement some of the existing algorithms for Euclidean reconstruction of 3-D objects. Attempts have been made in the refinement of the fundamental matrix for more precise estimation of the intrinsic/extrinsic camera parameters, and thus the 3-D reconstructions.