以 Active Shape Model 為基礎的 3D 物體變形演算法

Abstract

現今，作影像變形的方法有許多，但最主要還是以Registration的方法為主，因此在實作方面難免會遭遇到使用複雜性與效率性等等方面的問題；本論文提出根據Statistical Models of Appearance的方法發展出物體自動化變形系統的方法；我們稱我們的方法為Active Shape Models for Warping（ASMW）。並將該方法應用於果蠅腦中mushroom body及其周圍組織以及神經warping的實驗。

Recently, there are many ways to do image warping. Most of them were based on Registration. The registration based approaches need many user assists and thus are not automatic methods. In this thesis, we developed a 3D warping algorithm based on the Active Shape Model (ASM). We call the proposed method the “Active Shape Models for Warping（ASMW）”. The method was applied to warp between mushroom bodies of fruit fly brain.