图像小波阈值去噪算法研究

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.在Dohono提出的多分辨率分析小波阈值去噪的基础上,对阈值函数进行了改进.实验结果表明:与传统软阈值和硬阈值函数相比,改进的阈值函数通过调整参数的取值,可以取得较好的去噪效果,证明了在多数情况下,传统阈值函数并不能达到最佳的去噪效果.

Research on image denoising based on wavelet threshold

The aim of image denoising is to recover the original uncorrupted images from noisy ones based on MMSE.On the basis of multi-analysis wavelet threshold denoising method which is proposed by Dohono,an improved thresholding function is presented in this paper.Compared with traditional soft and hard threshold functions,the improved one can obtain the best denoising effects by adjusting the parameter values.Experimental results demonstrate that the proposed scheme removes noise effectively and show that the best denoising effect can not obtained by the traditional threshold function generally.