供特徵元件匹配用具有空間轉移不變性之鬆弛法架構 作者:陳世旺、戴建耘(國立臺灣師範大學資訊教育學系)

摘要：

截至目前為止已有不少供幾何特徵匹配用之鬆弛法架構被提出來，這些架構雖然宣稱可以對空間轉移具有不變性，但是實際上，大部分只能處理和旋轉及平移有關的 轉移，對於具有尺度因素的轉換澤盡量避免，因此，便有各種不同的假設被加諸於所考慮的問題，例如假設我們已知景深值，因此可以先將物件的尺度正規化後再比對，或者假設物形式晚整的，於是物形和模型間的尺度比率事先可以推知。本篇文章提出一種新的架構，它能夠同時對旋轉、平移和尺度具有不變性，此外，新架構也能處理變形物件及不完整物形。我們的實驗結果顯示新的架構卻具有可行性。

《詳全文》

A Transformation-Invariant Relaxation Scheme for Feature Mapping Author: Sei-Wang Chen, Chien-Yun Dai(Department of Computer and Information Education, National Taiwan Normal University)

Abstract：

A large number of relaxation schemes for feature mapping, claimed to be invariant to transformation, have been reported. However, most of them can deal with transfor-mations involving only rotation and translation, but not scaling. To stay away from the issue of scaling, unrealistic assumptions have to be imposed, such as the conjectures that range data are available so that objects can be rescaled before mapping, and that object shapes are complete so that ratios between object shapes and prototypes can be figured out beforehand. In this paper, we propose a relaxation scheme which is able to be in-variant at a time to rotation, translation, as well as scaling. In addition, the proposed scheme can also cope with shapes that may be distorted and incomplete. Our scheme has been tested on both synthetic and real data. Experimental results manifest that the proposed scheme is applicable.