- 作者: 高聖亮; 傅楸善
- 作者服務機構: 台灣大學資訊工程研究所
- 中文摘要: 在本篇報告中我們提出經由解聯立線性方程組來求得物體形狀的方法,這個方法是利用光度學在不同光源的照射下拍出物體的照片,再利用這些拍到的照片及已知的光源位置來導出聯立方程組,而這些方程組的未知數中便含有和物體形狀相關的資訊。在找出聯立方程組之前,我們首先假設在已知各個光源的位置間存在某種關係,那就是光源間的距離遠小於光源到物體的距離;在這種假設下,便可利用兩張不同光源照射下拍到的照片來導出一條線性方程式,在拍到數張像這樣的照片之後所得到的聯立方程組便可用來解出未知數,而從這些求得的這些未知數中我們就可知道物體的形狀了。
- 英文摘要: In this paper, a linear algorithm (Kao and Fuh, 1994) is proposed to recover shape information frommultiple images, each of which is obtained under an environment in which all of the object surfaces areilluminated by a known point light source. Furthermore, for these known point light sources, the distancebetween any two of them is very small when it is compared with the distance from the light source toany point on the object surface. This makes linearization of the image intensity equation possible. Atthe end of this paper, some implementation results are presented.
- 中文關鍵字: photometry; shape from shading; point light source; orthographic projection
- 英文關鍵字: --