基于多项式拟合的稀疏编码图像去噪算法研究  

Sparse coding for image denoising based on polynomial fitting

作　　者：孙海英[1] 王晓凯[1] 李锋[1] 

机构地区：[1]复旦大学电子工程系,上海200433

出　　处：《计算机工程与设计》2011年第3期991-994,共4页Computer Engineering and Design

摘　　要：为解决传统稀疏编码在图像去噪的过程中无法根据图像数据特点来决定稀疏编码的收缩函数的问题,提出了一种基于多项式拟合的稀疏编码图像去噪算法,该方法采用多项式来拟合稀疏系数的概率密度函数,进而估计稀疏系数的收缩函数,有效地解决了传统的去噪变量的概率密度模型无法反映图像数据特点的问题。仿真计算结果验证了该算法的有效性,并且在效果上要明显优于最大似然估计算法。To solve the problem of determining the shrinking function based on the data of the photos in sparse coding for image denoising effectively,an algorithm based on polynomial fitting is proposed.In this algorithm,polynomial is used to find the appropriate probability density function of the sparse coefficients.Then the shrinking function of the sparse coefficients is estimated.The shrinking function can fit the feature of the photo well.The simulation result of comparing to maximum likelihood method,indicated that this algorithm is effective and better.

关 键 词：稀疏编码 独立分量分析 多项式拟合 图像去噪 收缩函数 

分 类 号：TN911.73[电子电信—通信与信息系统]