红外技术, 2014, 36 (6): 496, 网络出版: 2014-06-30  

基于Otsu算法和分形维数相结合的红外云图分割

Segmentation of Infrared Cloud Image Based on Otsu Algorithm and Fractal Dimension
作者单位
南京航空航天大学自动化学院,江苏南京 210016
摘要
红外云图中的云系主要由积雨云、卷云、层云和积云组成,它们对于气象研究有很大的帮助。这 4类云系具有不同的分形维数特征,因而可以借此对它们进行分割。但是由于地表和一些云具有相似的分形维数特征,单纯依靠分形维数的分割效果不理想。对此,采用最大类间方差(Otsu)算法,分离出地表,去除干扰,并对提取出的云系采用分形维数进行分割。由于传统分形维数算法在计算时只选取了窗口内的灰度最大和最小的像素点,获得的分形维数特征不够精确,造成分割的云系出现混淆的现象。对此,在计算的过程中加入窗口中的全部像素点,获得的分形维数能准确地描述区域特征,保证了云系分割的良好效果。
Abstract
The clouds in the infrared cloud image are composed of cumulonimbus, cirrus, stratus and cumulus which are useful to study the atmosphere. We can use the fractal dimension algorithm to segment the clouds according to the different fractal dimension of them. But the result of the segmentation can’t be accurate when only fractal dimension is used, since the earth’s surface has the similar fractal dimension with some clouds. To solve the question, we use Otsu algorithm to separate the interference of earth’s surface, and then use the fractal dimension to segment the clouds. The result may be confused due to the inexact fractal dimension when traditional fractal dimension algorithm is used, because only the maximum and minimum pixels of a window are chosen to compute. To solve the question, we add all pixels in the window to compute, so that the exact fractal dimension can be got to ensure the good result.

徐晔烨, 王敬东, 朱晨雨, 顾浩. 基于Otsu算法和分形维数相结合的红外云图分割[J]. 红外技术, 2014, 36(6): 496. XU Ye-ye, WANG Jing-dong, ZHU Chen-yu, GU hao. Segmentation of Infrared Cloud Image Based on Otsu Algorithm and Fractal Dimension[J]. Infrared Technology, 2014, 36(6): 496.

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!