摘要
本文提出了一种快速准确的提取曲线拐点的算法。算法的总体结构是层次型的，它分级逐次淘汰伪拐点，而对每一级的处理又是顺序的逐点考察。因此，这种层次型的结构可抑制误差。实验表明，该算法不仅处理时间短，而且能保持原始曲线的总体形状，搜索出的拐点冗余度小，因而要求的存储容量小。
A new fast and accurate searching algorithm on curve critical points is proposed.The global structure of the algorithm is make up layer upon layer.The interesting feature of the proposed algorithm is its ability to clean out the false critical points step by step.And on the every level,the processing is proceeded in sequence.So the layered structure can control the errors.It has been shown by simulations that not only the processing is proceeded fast,but also the original curve shape is remained,the erroe of critical points is limited,in the last the needful mermory capacity is reduced.
出处
《通信学报》
EI
CSCD
北大核心
1999年第6期56-61,共6页
Journal on Communications
基金
国家自然科学基金
四川省应用研究基金
关键词
线性近似
图像压缩
曲线匹配
拐点
liner approximation,image compression,curve fitting,critical points
