摘要
提出一种基于对比度和相似性度量的多阈值分割算法。算法引入人类视觉系统感知光强度变化的非线性和适应性原理,将图像的灰度级区间分成几个互不相交的子区间,使得子区间的内部像素的共性以及子区间之间的像素对比度都尽可能大,综合共性和对比度之后获取图像的分割门限。将算法应用到发动机羽焰序列图像的分割并进行序列分析,实验结果表明算法是有效的。
This paper applies a multi-value segmentation algorithm, which is based on the contrast and homogeneity measure. The segmentation algorithm involves nonlinearity and adaptability of human visual system when it perceives the change of light intensity. Multilevel threshold values are found based on the measures of contrast of different subsets and homogeneity of the same subset, and it makes the segment is as homogeneous as possible while the contrast between any segment and its neighboring segments is as high as possible. Experiment result indicates that, the above algorithm can have efficient result when it is used to acquiring the engine plume’s information of different levels.
出处
《计算机与数字工程》
2007年第6期17-19,78,共4页
Computer & Digital Engineering
关键词
羽焰
相似性
对比度
多阈值分割
plume homogeneity,contrast,multi-threshold,segmentation
作者简介
于楚,硕士,研究方面:模式识别。
张天序,教授,博士生导师,研究方向:自动目标识别、计算机视觉与图像分析。
钟胜,讲师,研究方向:图像分析、多波段红外图像融合。