基于TV模型对含乘性噪声图像的去噪算法

摘要

TV method of removing multiplicative noise is improved by adding typical model of removing additive noise. Different from common explicit scheme or semi-implicit scheme, this paper implement explicit difference scheme by backing up the iterative matrix. It can preserve image contrast while removing noise. Another side, by adding constant Cε in regularized term, the stability of explicit difference scheme is more improved. It predigest implemention of model and improve algorithm speed and reduce staircase effect. The validity of algorithms are proved by experimental results.%在针对乘性噪声的TV模型的基础上,将原本针对加性噪声的典型模型引入到乘性噪声模型中去.在进行数值差分计算时,与常用的显式和半隐式差分实现方法不同,本文采用备份迭代矩阵的方法来实现显式差分,该方法可使模型在去除噪声的同时,能够保持图像的对比度不变.在此基础上,通过在正则化项中加入常数,使得模型在使用显式差分实现时稳定性大为提高,从而简化了模型实现的难度,提高了算法的速度,并且还一定程度上缓解了阶梯效应的出现.试验结果表明,改进的算法效果明显.