一种基于Curvelet变换的红外图像去噪方法

摘要

The wavelet transform has the defect in analyzing the edge characteristic of curves or straight lines in two-dimensional image,it doesn't has better approaching accuracy and spare ability of expression.This paper presents an improved threshold image denoising method based on the Curvelet transform, which combines the soft and the hard threshold together, and forms a new threshold function.Results of experiment to the infrared image and the visible light image show that the method can remove noise and remain edge with good infrared visual effect comparing with the soft and hard threshold denoising algorithm and orthogonal wavelet transform denoising algorithm,and the PSNR is also enhanced.%小波变换在分析二维图像中曲线或者直线边缘特征方面存在明显不足,用于红外图像去噪中没有较好的逼近精度和稀疏表达能力.为解决上述问题,提出一种基于Curvelet变换的阈值改进算法,即采用软硬阈值结合的方式,形成新的阈值函数.通过对可见光和红外图像进行仿真实验.结果表明,该方法与正交小波去噪以及软硬阈值去噪算法相比,在去噪和保持边缘的同时,取得了较好的红外视觉效果,并且峰值信噪比PSNR也得到一定的提高.