基于多尺度几何变换的导航雷达图象去噪仿真  

Navigation radar image denoising based on multiscales geometric transform

作　　者：韩福江[1] 王秀森[2] 刘瑞鹏[1] 

机构地区：[1]天津航海仪器研究所,天津300131 [2]海军大连舰艇学院,大连116018

出　　处：《中国惯性技术学报》2009年第2期197-200,共4页Journal of Chinese Inertial Technology

基　　金：船舶支撑技术项目(419040312)

摘　　要：导航雷达在采集、传输和显示过程中,由于多种因素的影响导致最终形成的图像中含有大量的噪声,影响了使用者对导航信息的分析和应用。传统的雷达图像去噪算法大多采用小波变换,但这种方法存在边缘模糊等问题。为了去除导航图像的噪声并解决小波变换中存在的边缘模糊问题,本文提出用基于多尺度几何变换的图像去噪方法对导航雷达图像进行处理,并利用基于多尺度几何变换的方法(包括基于Curvelet系数维纳滤波去噪方法和基于Contourlet域去噪方法)和基于小波变换的BayesShrink方法分别对含有模拟杂波和噪声的导航雷达图像进行仿真实验。实验结果表明:与基于小波变换的图像去噪方法相比,基于多尺度几何变换去噪方法能够更加有效去除雷达杂波和噪声。The final images of navigation radar during collection, transmission and displaying contain large amount of noise and could influence the analysis and application of the navigation information. So a wavelet transform was put forward to wipe off the noises and solve the edge blurry problem. The image denoising method based on multi-scale geometric transform was used for processing the navigation radar image. The methods based on multi-scale geometric transform （including denoising methods based on Curvelet and Contourlet） and the BayesShrink method based on wavelet transform were used respectively in simulation experiments for denoising the navigation radar image with artificial clutters and noises. The experiment results show that image denoising methods based on multiscales geometric transform could remove noise and clutter more efficiently than those based on wavelet transform, and have fairly good edge-preserved performance.

关 键 词：多尺度几何变换 导航雷达 图象去噪 小波变换 

分 类 号：U666.1[交通运输工程—船舶及航道工程]