基于SWBC变换尺度相关性的红外图像阈值降噪

摘要

As traditional denoising algorithm can't deal with the infrared image well,an infrared image threshold denoising algorithm based on scale correlation of SWBC (Stationary Wavelet-based Contourlet) transform is proposed. The energy distribution characteristics of signal and noise in SWBC transform domain were analyzed. The proposed de-noising algorithm only deals with high frequency sub-bands' coefficients instead of all sub-bands' coefficients in traditional de-noising algorithm. In order to increase the accuracy of SWBC coefficient threshold estimate,different thresholds were set for each coefficient,and combined with the scale correlation,double judgment was carried out for coefficients. In the experiments,different infrared images with noise are used to verify the proposed algorithm. Experimental results show that the proposed algorithm can obtain higher SNR improvement,and the SSIM is closer to 1,compared with the WBC inter-scale hard threshold denoising algorithm,WBC inter-scale self-adaptive threshold denoising algorithm and WBC scale correlation denoising algorithm.%针对传统图像降噪算法无法较好处理红外图像中的噪声问题,提出了一种基于SWBC(Stationary Wavelet-based Contourlet)变换尺度相关性的红外图像阈值降噪算法.本文分析红外图像信号和噪声在SWBC变换域各尺度上的能量分布特性,改进一般降噪算法对所有子带均进行处理的做法,只对高频子带系数进行降噪处理.同时为增加SWBC系数阈值判断的准确性,本文算法对每个系数设置不同的阈值,结合尺度相关特性,对系数进行双重判断.使用不同的含噪红外图像对本文算法进行检验.实验结果表明,相比于WBC尺度间硬阈值降噪、WBC尺度间自适应阈值降噪和WBC尺度相关性降噪,本文算法能获得更高的SNR提升,且SSIM值也更接近于1.