ABSTRACT

Texture synthesis, especially 2 dimensional (2D) image texture synthesis, is a common problem which interests many Computer Graphics researchers. But texture synthesis in 3 dimensional (3D) data or even higher dimension has been less discussed. In this thesis, we demonstrate an algorithm which employs a probability model and borrows the idea from visual hull to automatically generate 3D solid textures from a single image.

中文摘要

材質合成(texture synthesis)為電腦圖學(computer graphics)之熱門研究領域。多數材質合成之研究成果專注於處理二維影像，對較複雜之三維或更高維度之材質合成研究較少。本篇論文以由單張二維影像特徵合成三維材質為題；由二維影像統計資料建立機率模型，進而以此機率模型控制三維材質之合成過程，並利用視覺外型(visual hull)技術為固態材質建模。 

Key words: texture synthesis, solid texture, probability model, visual hull