测绘学报 ›› 2015, Vol. 44 ›› Issue (11): 1235-1245.doi: 10.11947/j.AGCS.2015.20140327

• 摄影测量学与遥感 • 上一篇    下一篇

结合分形理论和自适应图像块划分的遥感图像噪声估计

傅鹏1,2, 孙权森1, 纪则轩1   

  1. 1. 南京理工大学计算机科学与工程学院, 江苏 南京 210094;
    2. 悉尼大学信息科技学院, 新南威尔士 悉尼2006
  • 收稿日期:2014-06-19 修回日期:2014-10-19 出版日期:2015-11-20 发布日期:2015-11-25
  • 通讯作者: 孙权森,E-mail:sunquansen@njust.edu.cn E-mail:sunquansen@njust.edu.cn
  • 作者简介:傅鹏(1986-),男,博士生,研究方向为图像处理、模式识别和机器学习等。E-mail:njust_fupeng@163.com
  • 基金资助:
    国家自然科学基金(61273251;61401209);中国博士后科学基金(2014T70525;2013M531364);十二五民用航天技术预先研究项目(D040201);江苏省自然科学基金青年基金(BK20140790);江苏省普通高校研究生科研创新计划(CXZZ13_0211)

Noise Estimation from Remote Sensing Images by Fractal Theory and Adaptive Image Block Division

FU Peng1,2, SUN Quansen1, JI Zexuan1   

  1. 1. School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing 210094, China;
    2. School of Information Technologies, University of Sydney, Sydney 2006, AustraliaAbstract
  • Received:2014-06-19 Revised:2014-10-19 Online:2015-11-20 Published:2015-11-25
  • Supported by:
    The National Natural Science Foundation of China(Nos.6127325161401209) China Postdoctoral Science Foundation(Nos.2014T705252013M531364) Project of Civil Space Technology Pre-research of the 12th Five-year plan(No.D040201) Natural Science Foundation of Jiangsu Province of China(Youth Fund Project)(No.BK20140790) Scientific Research and Innovation Project Fund for Graduate Students of Jiangsu Province Higher Education Institution(No.CX22B_0211)

摘要: 针对场景复杂的光学遥感图像中加性噪声估计问题,提出了一种结合分形理论和自适应图像块划分的噪声估计方法。区别于传统的基于规则图像块划分的噪声估计方法,本文提出了一种自适应于图像局部信息的图像块划分算法,更大程度地保证了自适应图像块内部的平滑性。结合基于分形理论的图像低粗糙度纹理区域选取和基于统计分析的图像噪声标准差计算,实现了光学遥感图像加性噪声强度的自动估计。利用资源三号卫星图像进行定量试验分析,试验结果表明本文方法可以有效地适用于不同复杂程度、不同噪声强度的光学遥感图像。同时,本文中低粗糙度纹理区域选取和自适应图像块划分的方法经过改进后,还可以应用于雷达图像中乘性噪声的估计。

关键词: 光学遥感图像, 噪声估计, 分形理论, 纹理分析, 自适应图像块, 雷达图像

Abstract: A novel approach for additive noise estimation from highly textured optical remote sensing images has been proposed, which is based on fractal theory and adaptive image block division. Different from the conventional regular block division based noise estimation methods, the divided adaptive image blocks with the proposed method are adhering to the local image information, which are most likely to be homogeneous blocks. Combining with the week textured image region detection using fractal theory and noise standard deviation calculation using statistical analysis, the proposed method can automatically estimate additive noise intensity from optical remote sensing images. Quantified analysis of experiments with ZY-3 satellite images demonstrates that the proposed method is applicable to optical remote sensing images with various complexities and different noise levels. Meanwhile, the notion of week textured image region detection and adaptive image block division can also be applied to multiplicative noise estimation from radar images after modification.

Key words: optical remote sensing images, noise estimation, fractal theory, texture analysis, adaptive image block, radar images

中图分类号: