基于小波和独立成分分析的去噪自适应算法

为了寻求一种能将不同类型和数量的噪声从图像中去除的方法,提出了一种能从图像源中将噪声与信号分离的改进的小波ICA滤波器。该方法首先使用小波降维,用Morlet小波来解决非正交问题;通过ICA规范化降维后的信号,从而发现独立噪声特征;再通过相关性将图像和噪声分离;最后,对图像进行还原,得到去噪后的图像。通过实验与主成分分析(PCA)方法、FastICA方法进行了对比,验证了该方法的有效性。结果显示,本研究提出的方法降噪效果较PCA方法和FastICA方法有大幅提高。同时,复杂度略有上升。

An Adaptive Denoising Algorithm Based on Wavelet Transform and Independent Component Analysis

In order to find a way to remove different types and amounts of noise from the image, an improved wavelet ICAfilter that separating noise and signal from the image source was proposed. The suggested method using wavelet dimensionreduction first and solved the problem of Non-orthogonality by using Morlet wavelet if necessary then normalizing the sig.nal reduced the dimensionality through ICA that found independent noise characteristics. The image and noise were sepa.rated by correlation. Finally, the image was restored to obtain a denoised image. This algorithm was compared with Princi.pal Component Analysis (PCA) and FastICA by experiment to verify the effectiveness of the proposed method. The resultsshow that the method proposed in this paper is much better than PCA and FastICA in image denoising, the complexity isslightly increased at the same time.