首页 | 本学科首页   官方微博 | 高级检索  
     

基于场景相似性度量的SAR图像抑斑
引用本文:王晟达,凤宏晓,马健,徐松涛.基于场景相似性度量的SAR图像抑斑[J].雷达科学与技术,2013,11(1):26-32.
作者姓名:王晟达  凤宏晓  马健  徐松涛
作者单位:空军工程大学航空航天工程学院;陕西黄河集团有限公司设计所
摘    要:通过研究SAR图像的信号相关加性噪声模型,结合提出的加性残留噪声模型得到一种图像域的期望加权滤波器,并引入像素间的相似性作为加权时的权值。接着,指出传统像素间相似性的度量方法不能鲁棒地用于噪声图像,提出从图像的灰度特性和空间结构两个方面考虑,利用灰度和梯度距离度量像素间的相似性,并应用局部场景之间的相似性估计像素间的相似性使之可靠地应用于SAR图像。实验表明,所提出的算法具有相当出色的抑斑性能,在有效的抑制斑点噪声的同时,均匀区域几乎没有出现“振铃”效应,并且非常好地保持了原始图像的边缘、纹理和点状目标。

关 键 词:SAR图像抑斑  期望加权滤波器  像素相似性度量  灰度与梯度距离

SAR Image Despeckling Based on Scene Similarity Measurement
WANG Sheng-da,FENG Hong-xiao,MA Jian,XU Song-tao.SAR Image Despeckling Based on Scene Similarity Measurement[J].Radar Science and Technology,2013,11(1):26-32.
Authors:WANG Sheng-da  FENG Hong-xiao  MA Jian  XU Song-tao
Affiliation:1(1.The Aeronautics and Astronautics Engineering Institute,Air Force Engineering University,Xi’an 710038,China; 2.The Design Institute of Shaanxi Huanghe Group Co,Xi’an 710043,China)
Abstract:By studying the signal-dependent additive noise model and applying the proposed additive re- sidual noise model, an expectation weighted filter is proposed. Its weight is calculated with the similarity of pixels. After that, we indicate that the traditional measurement methods of similarity between pixels are not robust. The gray characteristics and spatial structure are considered. Then the combination of gray and gra- dient distances is proposed to measure the similarity of pixels, which is reliably applied for SAR images by estimating the similarity of local scenes. Experiments show that the proposed algorithm has remarkable despeckling performance and is almost absent from ring phenomenon. It also has good performance for ed ges, textures and point object preservation.
Keywords:SAR image despeckling  expectation weighted filter  pixel similarity measurement  grayand gradient distances
本文献已被 CNKI 维普 等数据库收录!
点击此处可从《雷达科学与技术》浏览原始摘要信息
点击此处可从《雷达科学与技术》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号