摘要
本文根据二值图象的边缘点在八个方向上具有较多的象素不相互对称这一重要特征,提出了一种新的细化算法。由剥蚀和矫正两个部分组成。在剥蚀部分,只需判别被测象点的8或12个邻点的分布情况,就可以决定该点的删除与否。而矫正部分,则是对剥蚀后的图象的进一步完善。我们用本文所介绍的算法,对1200个无限制手写体数字进行细化处理,结果证明该算法具有判别简单,对噪声不敏感,并能较好地保持原模式的形状及连通性的优点。
A new thinning algorithm based on the characteristics of edge points with unsymmetrical picture elements in the eight directions is presented in this paper. The algorithm comprises the deletion part and the correction part. In the deletion part, whether a point should deleted can de determined rather simply if the distribution of eight or twelve neighboring points is known. In the correction part the picture will be perfected after the first part is achieved. The results from the thinning processing of 1200 handwritten numerals show that the algorithm possesses the advantages of simple judge processing, insensibility to noise, and unchangement of shape and continuance of the original pattern.
出处
《数据采集与处理》
CSCD
1990年第1期7-12,共6页
Journal of Data Acquisition and Processing
