經由邊界及表面同時調適的方式重建二次曲面形狀 作者:李忠謀(國立臺灣師範大學資訊教育研究所)

摘要：

本文提供的技巧，可以用來從對應的光度及景深影像中，偵測出二度物體的子部位形狀。主要的假設包括：(1)物體最突顯的特徵乃是發生在物體的輪廓部位以及物體表面上的邊線，(2)物體可能有互相遮掩的情形，故重建物體的形狀只能藉重於物體較小的部位。本文所提供的方法，乃是經由將線條及表面的資料，同時調 適，使得以匹配二次曲面形式的一種技巧。所需要使用的圓球、圓柱及圓錐二次曲面結合邊線的方程式在文中均有推導。藉蒙地卡羅方式的測驗，我們得以證實：(1)線條及表面資料同時調適的結果，以調適的品質來看，會優於個別調適的方式；(2)利用對應光度及景深影像資料來偵測物體的形狀，可以降低物體表面錯 誤分類的發生。

《詳全文》

On the Recovery of Quadric Surface Shapes via Simultaneous Boundary and Surface Fitting Author: Greg C.Lee(Department of Information and Computer Education, National Taiwan Normal University)

Abstract：

A method is proposed for detecting the shape of small parts of objects from fused intensity and range data. My primary assumptions are: (1) the most telling object features occur on the object rim and at object creases, and (2) occlusion existing in complex scenes requires the reconstruction of sensed objects in terms of these small parts. A fitting method that simultaneously fits specific quadric forms to both surface and contour data at object limbs or creases is presented. Fitting equations for spherical, cylindrical and conical limbs and for polyhedral creases are derived. Results from Monte Carlo experiments support the hypotheses that (1) fitting both surface and contour is superior, in terms of the quaality of the fit, to either surface fitting or contour fitting alone; (2) the shape of the surface is less likely to be misclassified when we use the proposed fused data fitting method.