基于獨立分量分析的降噪技術

介紹了新興的獨立分量分析技術的基本概念和原理，以及具有代表性的算法，即FastICA算法、EASI算法、非線性PCA算法和基于自然梯度的最大似然估計算法。通過降噪仿真實驗，并采用均方誤差作為降噪的性能指數(shù)，對這些算法與傳統(tǒng)的自適應信號處理算法進行比較。所得實驗結果表明，獨立分量分析算法在降噪上的效果優(yōu)于自適應信號處理算法。因此在降噪上具有較大的應用價值。

The paper introduces a new technology of signal processing: independent component analysis, including its basic concept, principles, and some representative algorithms, such as FastICA, EASI, Nonlinear PCA, and natural gradient algorithm based maximum likelihood estimation. In a denoising simulation experiment with the mean square error criterion, these algorithms are compared to the classic algorithms of adaptive signal processing, such as LMS and RLS. Results show that in denoising application ICA algorithms are superior to the classic adaptive algorithms. Thus ICA algorithms have large value in denoising application, deservnig further study and promoting.