基于小波的红外图像去噪算法研究

摘要

研究红外图像优化预处理问题,针对红外图像噪声分布的特点,提出一种新的去噪算法.对红外图像进行两次小波分解与重构,对分解后小波系数采用不同的阈值策略,分别消除加性噪声和乘性噪声.由于乘性噪声与红外图像相关,利用红外图像与乘性噪声在统计上独立的特点,在进行第二次小波分解之前先对含有乘性噪声的红外图像进行对数运算,这样才能消除乘性噪声.仿真结果表明,无论是视觉效果,还是客观评价指标,采用上述算法均优于传统去噪算法.并可广泛应用于红外成像领域.%Aiming at the characteristic of infrared image's noise distribution, a novel algorithm on infrared image de-noising is proposed.The additive noise and multiplicative noise of the infrared image are eliminated respectively by using twice wavelet decompositions and reconstructions to the infrared image and applying different threshold strategy to wavelet coefficients.Logarithmic operation must be adopted when the multiplicative noise exists in infrared image due to the correlation between the infrared image and multiplicative noise which are statistically independent before the second wavelet reconstruction, then the multiplicative noise can be removed.The simulation results show the presented algorithm is better than previous ones, both in visual effect and objective evaluation criteria.This de-noising algorithm can be widely applied in the field of infrared image.