小波去噪(大学论文).doc

PAGE PAGE 2 数字图像阈值去噪算法研究与实现 摘 要 我们在现实生活中得到的图像都存在噪声，所以需要图像的去噪。图像去噪的目的是在满足最小均方误差准则的条件下，将原始图像从观测到的含噪图像中还原出来。近年来，在小波变换域实现信号的滤波引起了人们的极大重视，一些滤波方法也相继被提出。传统的小波阈值去噪可分为硬阈值去噪和软阈值去噪，而在其去噪过程中，硬阈值函数在一些不连续点处有时会产生伪吉布斯现象；软阈值函数中估计的小波系数与信号的小波信号之间存在恒定偏差。为了去除这些现象，采用最佳软阈值进行图像处理。实验结果表明，虽然硬阈值去噪方法的性噪比和均方差也比较小，但是对图像的细节方便保护的不好。而最佳软阈值函数消噪后的视觉特性较好，并且信噪比提高，均方根误差有所降低。总体而言，最佳软阈值去噪方法结合了硬阈值和软阈值的优点。 关键词 小波变换；阈值消噪；门限规则 Study and implementation of digital image threshold denoising algorithm ABSTRACT We have inages with the image noise in real life, so it is necessary to image denoising. The aim of image denoising is to recover the original uncorrupted images from noisy ones based on MMSE. In recent years, filtering in wavelet domain became an important research field. Also, some new methods were developed. The traditional method of threshold in denoising has two kinds of ways，one of them is hard thresholding and the other is soft thresholding．In some cases，such as on the discontinuities points，the Gibbs phenomenon will exhibit when we use hard thresholding function to remove noise of signals and soft thresholding method also has disadvantages．In order to remove the shortings，we use the best soft thresholding function(the best soft thresholding method ). The experimental results show that, although psnr and mse of the hard thresholding denoising are relatively small, but the protection of the details of the image is poor.The vision of denoising is better and the RMSE of sign al has been decreased a lot while the SNR has been increased.Overall, the best soft thresholding denoising method combines hard threshold and soft threshold advantages. Key words wavelet transform；thresholding denoising；method of threshold 目 录 TOC \o 1-3 \h \u 摘 要 I ABSTRACT II 前 言 1 1. 图像去噪背景 2 1.1 图像噪声 2 1.2国内外发展现状 3 1.3系统实现概述 3 1.3.1读取图像模块 4 1.3.2加噪模块 4 1.3.4 图像重构 5 2.小波去噪理论 5 2.1小波去噪特点 5 2.2 小波变换原理 6 2.3小波去噪原理 7 2.4阈值函数 8 2.4.1阈值函数(hard-thresholding) 8 2.4.2软阈值函数(