二维Renyi熵图像阈值选取快速递推算法  

Fast recurring algorithms of image thresholding based on two-dimensional Renyi's entropy

作　　者：潘喆[1] 吴一全[1] 

机构地区：[1]南京航空航天大学信息科学与技术学院,江苏南京210016

出　　处：《中国体视学与图像分析》2007年第2期93-97,共5页Chinese Journal of Stereology and Image Analysis

摘　　要：阈值分割是图像处理技术中的一个基本问题。阈值选取则是阈值分割技术的关键。随着阈值分割技术的不断发展,对阈值选取方法的准确性及快速性的要求越来越高。本文提出了二维Renyi熵阈值法的两种快速递推算法,消除了计算过程中的重复计算,大大提高了二维Renyi熵阈值选取算法的运行速度。通过对各种类型的图像进行大量实验,结果表明二维Renyi熵阈值选取算法分割准确,本文所提出的两种快速算法的运行时间几乎不到原始算法的0.1%,这使得二维Renyi熵法更具实用性。Thresholding is a basic problem in image processing. Threshold selection is crucial to thresholding. With the continuing development of thresholding technique, the requirement of thresholoding algorithms is increasingly accurate and fast. In this paper, two fast recurring two-dimensional Renyi＇s entropic thresholding algorithms are proposed, which may reduce repeat computation and increase the computational velocity. Some experiments were made on all kinds of images. The results show that two-dimensional Renyi＇s entropic thresholding algorithm achieves a good segmentation, and the computational time of its two fast recurring algorithms is less than 0. 1% of that of the original algorithm. That makes two-dimensional Renyi＇s entropic thresholding algorithm much more practical.

关 键 词：图像处理 阈值选取 二维RENYI熵 快速递推算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]