结合非局部先验性与加权核范数最小化的声纳图像去噪  

Sonar Image Denoising Based on Nonlocal Priori and Weighted Nuclear Norm Minimization

作　　者：石建飞[1] 唐玉波 孙裕超 SHI Jianfei;TANG Yubo;SUN Yuchao(No.3 Research Institute,CETC,Beijing 100015,China;Space System Division,Beijing 100000,China)

机构地区：[1]中国电子科技集团公司第三研究所,北京100015 [2]航天系统部,北京100000

出　　处：《电声技术》2020年第7期17-21,共5页Audio Engineering

摘　　要：如何快速有效地抑制声纳图像的噪声,是声纳图像目标识别系统需要解决的关键问题之一。为了解决基于加权核范数最小化的声纳图像去噪过程中易带来过平滑的问题,融合数字图像非局部先验知识,提出了结合非局部先验性与加权核范数最小化的声纳图像去噪算法。首先,利用声纳图像的非局部先验信息建立限制条件,形成基于非局部先验性的矩阵加权核范数最小化的去噪原理模型。其次,交替迭代求解图像噪声模型,重构得出噪声去除后的声纳图像。试验结果表明:提出的算法不仅能够有效减弱声纳噪声,还能够有效改善图像感官效果。How to suppress the noise of sonar image quickly and effectively is one of the key problems to be solved in sonar image target recognition system.In order to overcome the problem of over-smoothness in sonar image denoising based on weighted nuclear norm minimization,a new of sonar image denoising method combining non-local priori and weighted nuclear norm minimization is proposed in this paper.Firstly,a weighted nuclear norm minimization denoising model based on the non-local priori of image is established.Then,the denoising model is iteratively solved alternately to reconstruct the sonar image after noise suppression.The experimental results show that the proposed algorithm can not only effectively suppress sonar noise,but also significantly improve the visual effect of the image.

关 键 词：声纳图像去噪 非局部先验性 核范数最小化 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] U666.7[自动化与计算机技术—计算机科学与技术]