|国家预印本平台
首页|一种基于分数阶总变分的乘性噪声去噪快速算法

一种基于分数阶总变分的乘性噪声去噪快速算法

Fast Algorithm for Fractional-order Total Variation Based Multiplicative Noise Removal

中文摘要英文摘要

本文利用算子分裂方法提出了一种针对具有一般保真项的基于分数阶总变分的变分模型的快速交替迭代算法.我们提出了两种具有不同保真项的基于分数阶总变分的乘性噪声去噪变分模型,并利用这种算法进行求解.为进一步提高图像质量,我们提出了参数自适应选择方法.实验结果表明,在固定参数时,本文算法速度快、计算量小,所提出的自适应算法可以在图像非纹理区域有效去除噪声,同时可以保持图像纹理区域的细节信息,可以有效提高图像质量.

In this paper, using the operator splitting technique, we propose a fast alternating iterative algorithm for the fractional-order total variation regularized model with general fidelity term. As an application, we use the new algorithm to solve two models for multiplicative noise removal with different fidelity terms. To improve the performance, we choose the parameters adaptively and propose an adaptive algorithm for multiplicative noise removal. Numerical results show that the new algorithm with fixed parameters has low computational cost. The adaptive algorithms can not only remove the noise and eliminate the staircase effect in the non-textured region, but also preserve the textures well in the textured region, and therefore can improve the result visually efficiently.

张军、韦志辉

计算技术、计算机技术

总变分分数阶微分算子分裂乘性噪声去噪

total variationfractional-order derivativeoperator splittingmultiplicative noise removal

张军,韦志辉.一种基于分数阶总变分的乘性噪声去噪快速算法[EB/OL].(2012-01-10)[2025-08-25].http://www.paper.edu.cn/releasepaper/content/201201-291.点此复制

评论