摘要
根据EMD对图像信号进行分解,本文讨论了采用硬阀值去噪声和软阀值去噪的方法的优劣性,并提出了EMD+SG滤波器组合去噪的方法,对前N/2个IMF采用SG滤波器对每一数据点的一个邻域(长度为n的滑动窗口)进行滤波,用一元P阶多项式根据最小二乘法准则,拟合出邻域内的最佳值作为去噪后的数值,再与后N/2个IMF进行图像重构。实验表明,该算法比其它算法具有较好的去噪效果。
According to EMD decomposition for an image signal, the advantages and disadvantages between hardthreshold denoising and soft-threshold denoising are discussed in this paper and the EMD ＋ SG filter combination denoising method is put forward. Firstly, the neighborhood of each data points （the sliding window with length n） is filtered by the SG filter for the former N/2 IMF. Then fit out the best value within neighborhood as the denoised data using a first order polynomial with one variable according to least square methods. Finally, the value and the following N/2 IMF are reconstructed. Experiments show that this algorithm has a good result than other de-noising methods.
出处
《电子测量技术》
2009年第11期58-61,共4页
Electronic Measurement Technology
关键词
经验模态分解
SG滤波器
去噪
empirical mode deeomposition（EMD） SG filter Denoising
