NSCT域分类预处理的改进非局部均值去噪算法  

Improved NLM Denoising Algorithm Based on Classification Preprocess in NSCT Domain

作　　者：王倩[1] 彭海云[1] 秦杰[1] 柴争义[1] Wang Qian;Peng Haiyun;Qin Jie;Chai Zhengyi(School of Computer Science and Technology,Zhoukou Normal University,Zhoukou 466001)

机构地区：[1]周口师范学院计算机科学与技术学院,周口466001

出　　处：《计算机辅助设计与图形学学报》2018年第3期436-446,共11页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(U1504613);河南省高校科技创新团队计划(17IRTSTHN009);河南省高等学校重点科研项目(17A520068)

摘　　要：针对图像去噪领域非局部均值算法存在着计算量过高、复原图像过于平滑等问题,提出一种基于非下采样轮廓波变换(NSCT)域系数分类预处理的改进型非局部均值去噪算法.首先利用NSCT获得图像的高频系数,通过模糊支持向量机将系数分为无噪和含噪2类,只保留含噪系数进行后续非局部均值处理,降低整体算法的计算复杂度;然后利用极谐波变换分解系数取代传统非局部均值中的像素值参与相似度计算,使得计算过程对方向改变具有更好的抵抗能力;最后引入改进双平方函数作为相似度计算的核函数,可以更加符合极谐变换分解系数间的残差特性,使相似度计算得到的权重值更加精确.在标准灰度图像和遥感图像上的实验结果表明,与经典的去噪算法相比,该算法提高了计算速度,拥有更好的边缘和结构保持能力,整体去噪效果也得到了显著的提高.Non-local mean(NLM)algorithm for image denoising exists some problems such as the high computation complexity and the smooth restored image,etc.In this paper,we propose an improved NLM image denoising algorithm based on classification preprocess in the non-subsampled contourlet transform(NSCT)domain.Firstly,high frequency coefficients of the image in NSCT domain are obtained and classified into two classes using fuzzy support vector machine(FSVM),i.e.,non-noisy coefficients and noisy coefficients.Secondly,to reduce the computational complexity of the overall algorithm,only the noisy coefficients are retained for subsequent NLM processing.The polar harmonic transform(PHT)decomposition coefficients are used to replace the pixel values of the traditional NLM to calculate the similarity,which makes the computation process have better resistance to the change of direction.Finally,the modified bisquare function is used as the kernel function of similarity calculation.This is due to the modified bisquare function can be more in line with the residual characteristics between PHT decomposition coefficients,which makes the weight values of the similarity calculation are more accurate.Experiments are conducted on standard gray images and remote sensing images,and extensive experimental results show that the proposed algorithm can not only accelerate the computation speed of the traditional image denoising algorithms,but also has a better ability to preserve edges and structures,and the overall denoising performance of the proposed algorithm has been significantly improved.

关 键 词：非下采样轮廓波 模糊支持向量机 非局部均值 极谐变换 改进的双平方函数 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]