摘要
SAR(synthetic aperture radar)图像有固有的乘性相干斑噪声,抑制相干斑噪声和保护边缘信息是研究的重要内容。文章提出了一种变差系数差构建的非局部平均滤波算法DCV-NLM(difference of coefficient of variation non-local means)。首先,搜索窗口中的变差系数的像素与中心像素在相似窗口中之间的局部差,其次,局部差由两个范数构造的相似性参数,用负指数形式得到每个像素的加权系数,最后是对SAR图像的噪声执行加权滤波。通过提出的算法与IDPAD算法对比,从抑制相干斑图像视觉效果方面对比,DCV-NLM算法比起IDPAD算法有更好的相干斑抑制性能。
SAR(synthetic aperture radar)images have inherent multiplicative speckle noise.It is important to suppress speckle noise and protect edge information.In this paper,a non-local means filtering algorithmDCV-NLM(difference of coefficient of variation non-local means)is proposed.Firstly,the local difference between the variation coefficients in the search windowand the central pixels in the similarity window is used.Secondly,the local difference forms a negative exponential weighting coefficient through the similarity parameters constructed by two norms.Finally,the noise of SAR image is weighted and filtered.Compared with IDPADalgorithm,DCV-NLMalgorithmhas better speckle suppression performance in the visual effect of speckle suppression image than IDPAD algorithm.
作者
高飞
朱磊
冯子金
韩普
Gao Fei;Zhu Lei;Feng Zijin;Han Pu(Xi'an Polytechnic University,School of Electronics and Information,Xi’an 710048,China)
出处
《信息通信》
2019年第10期42-44,共3页
Information & Communications
基金
国家自然科学基金（61401347）
陕西省重点研究发展计划（2019GY-113）
陕西省教育厅专项研究计划（17JK0343）
关键词
SAR图像
相干斑抑制
变差系数
局部差
加权系数
SAR image
speckle suppression
coefficient of variation
local difference
weighting coefficient
