基于自定义灰度空间分布密度矩阵的Otsu改进算法  

An Improved Image Segmentation Algorithm Based on Gray Level Distributing-Density Matrix and Otsu Method

作　　者：刘晓葳[1] 王乘[1] 郭泉[1] 

机构地区：[1]华中科技大学工程计算与仿真研究所,武汉430074

出　　处：《计算机与数字工程》2007年第6期23-24,157,共3页Computer & Digital Engineering

摘　　要：Otsu自适应阈值法作为图像阈值分割的经典算法,在图像处理与模式识别领域有着广泛应用。将图像中灰度级的空间分布特性应用到图像分割中,自定义适当的灰度空间分布密度矩阵结合先验知识,提出Otsu改进算法。实验证明在图像中目标与背景的灰度差异不是特别大的情况下,该算法的分割效果优于经典的Otsu算法以及其它分割算法。Being as a classical image segmentation algorithm, Otsu method has been widely employed in image processing and pattern recognition. In this paper, the spatial distributing character of gray level is applied for image segmentation. Based on gray level distributing-density matrix and Otsu method, and improved algorithm has been proposed. According to experiment result, the improved algorithm performs better than Otsu method and some other segmentation algorithms when the difference between target and background in gray level is not very great.

关 键 词：图像分割 OTSU 灰度空间分布密度矩阵 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]