小波图像消噪中阈值与信噪比的单峰规律及阈值试探方法

Single Top Value Regulation between Threshold and SNR and a Trial Threshold Method in Wavelet-based Image Denoise

本文研究了小波图像去噪过程中阈值与信噪比的关系,通过大量的仿真得出了小波图像消噪过程中,在不同的噪声强度下,信噪比与小波阈值之间存在单峰性的特征。基于该单峰规律,提出了一种小波图像消噪的阈值试探方法。该方法通过不断地调整小波去噪的阈值,以求得接近最大信噪比时的阈值,从而确定具有接近最大信噪比时的小波阈值,因此可以获得比常用小波阈值估计方法更清晰的图像。仿真结果表明该方法可以得到一个较高的信噪比,去噪效果较好。

The relation between the wavelet threshold and the signal-to-noise ratio in wavelet image denoise is presented. From many simulations of wavelet image denoise, the unimodal property between the wavelet threshold and the signal to noise ratio （SNR） under different noise strength was developed. The rule is very important for this subject and it needs to prove more strictly with mathematical methods. With the unimodal property, which is used to determine the superior wavelet threshold, a trial wavelet thresholding method was developed. With this method different threshold was used and many SNRs were got, so we could find out the best SNR and know which threshold was the best for the special image denoise. If we use that threshold for image denoise, we can get a better denoised image. Experiment results show we can get a better SNR and a clearer image with this method.