摘要： 本文首先分析了现有典型重构算法所用的重构准则,指出了它们的不足之处,由此提出了寻找新重构准则的指导思想,即以相邻轮廓间存在的形状相似性为根据,先找出各轮廓的形状特征,然后找出两相邻轮廓间相互匹配的形状特征,最后在匹配的轮廓段之间,用仅与形状有关的算法进行重构。本文还对Zernik算法作了改进,并提出一种新的重构准则及算法,且与已有典型算法相比较,新算法的结果更为合理。

Abstract: This paper analyses criteria of some typical algorithms, points out their common shortcomings, and puts forward a methodology for discovering new reconstruction criteria for reconstructing surfaces from contours. On the basis of correlation between adjacent contours, the criterion should be, first, to investigate both contours' features, then, to find features on the adjecent contours that match each other, and finally, using an algorithm (which is only shape depended), to reconstruct surfaces between the matched contour segments. The later part of the paper presents a modified Zernik algorithm, and a new reconstruction criterion and its corresponding algorithm. Comparision with previous typical algorithms shows that our algorithms delieve more resonable results.