小波阈值去噪的一种改进方法

An improved wavelet thresholding method for image denoising

白噪声的方差和幅值随着小波变换尺度的增加会逐渐减小,而信号的方差和幅值与小波变换的尺度变化无关。本文在Donoho的软、硬阈值去噪方法基础上,提出了一种新的阈值函数,并把它们应用在图像的去噪上。该阈值函数具有物理意义清晰、表达式简单、计算方便等优点。实际噪声图像测试结果表明,这种经改进的方法可以有效地去除白噪声干扰,无论是在视觉效果上还是在信噪比和均方误差定量指标上均明显优于常用的软、硬阈值去噪算法以及改进的软硬阈值折中算法。

The variance and amplitude of the white noise decrease with the increase in the ,scale of the wavelet transfom,, but the variance and amplitude of a specific image bear no relation to the wavelet transform ,scale. Based on the soft and hard thresholding method which put forward by Donoho, a new class of thresholding function is proposed. This new thresholding function has many advantages over Donoho＇s ,soft and hard thresholding function. It is clear in physics meaning and simple in expression. The results show that this improved method is effective in removing white noise, and gives better NMSE performance and PSNR gains than soft and hard thresholding methods.