基于支持向量值轮廓波变换的遥感图像去噪

Remote sensing image denoising based on support vector value contourlet transform

提出了一种基于支持向量值轮廓波变换的遥感图像去噪算法。首先利用支持向量机构造支持向量值滤波器,并结合方向滤波器组,构建支持向量值轮廓波变换,再利用该变换将含噪声遥感图像分解成低频部分和高频方向子带部分,最后利用支持向量回归方法对子带系数进行去噪。实验结果表明,支持向量值轮廓波变换具有平移不变、泛化能力好、捕捉奇异性能强等特性,本文提出的去噪算法能在去除噪声的情况下有效保留源图像的边缘信息。

A remote sensing image denoising algorithm based on support vector value contourlet transform is proposed.Firstly,a support vector value filter constructed by support vector machine is combined with directional filter banks,and a support vector value contourlet transform is constructed.Then the transform is used to decompose the noising remote sensing image into low frequency subbands and directional high frequency subbands.Finally,all subbands are denoised based on support vector regression.Experiments show that the support vector value contourlet transform has the advantages of shift invariant,good generalization and strong catching singularity ability.By using this proposed denoising algorithm,the image noise is removed while edges are preserved effectively.