基于Mallat算法脑电信号处理的新方法  

A New Method of EEG Signal Processing Based on Mallat Algorithms

作　　者：李春雪 郭家赫 白雪梅[1] 彭金歌 李永辉 LI Chun-xue;GUO Jia-he;BAI Xue-mei;PENG Jin-ge;LI Yong-hui(School of Electronics and Information Engineering,Changchun University of Science and Technology,Changchun 130022;School of Computer Science and Technology,Changchun University of Science and Technology,Changchun 130022)

机构地区：[1]长春理工大学电子信息工程学院,长春130022 [2]长春理工大学计算机科学技术学院,长春130022

出　　处：《长春理工大学学报（自然科学版）》2020年第6期80-88,共9页Journal of Changchun University of Science and Technology(Natural Science Edition)

基　　金：吉林省自然科学基金项目(NO.20150101013JC);国家级长春理工大学大学生创新创业训练计划项目(2018G021)。

摘　　要：在脑电信号特征提取算法中,常用的是小波包分解和小波分解。“小波包分解”会把信号的高频和低频完全分解。在此过程中,无关信号频率也会被分解,算法复杂度被迫提升。因此,引入只对低频信号进行分解的“小波分解”。假设所需信号中包含高频信号,经小波分解后,得到的信号会以丢失高频信息为代价,换取算法在复杂度上的优势。为了解决传统的小波函数利用Mallat算法进行分解、重构的复杂性,在Mallat算法之上,提出了“半小波包”的概念,形成将小波分解与小波包分解相结合的“半小波包分解”算法,解决了小波分解只对低频信号有效以及利用Mallat算法实现的小波包分解的冗余问题。首次利用半小波包和卷积过程改进的算法对DEAP数据库中的脑电信号进行分解,在保证准确度的基础上,达到了分解信号高速率、分解算法低复杂度的效果。Wavelet packet and wavelet decompositions are commonly used in EEG feature extraction algorithms."Wavelet Packet Decomposition"can completely decompose the signal at both high and low frequencies.In this process,t he frequency of irrelevant signals will be decomposed and the complexity of the algorithm will be increased.Therefore,the"wavelet decomposition"which only decomposes the low frequency signal is introduced.Assuming that the required signal contains high-frequency signal,after wavelet decomposition,the obtained signal will lose high-frequency information at the expense of the complexity advantage of the algorithm.In order to solve the complexity of decomposition and reconstruction of traditional wavelet functionsby using Mallat algorithm,on the basis of Mallat algorithm,in this paper,the concept of"semiwavelet packet"is put forward,which combines the wavelet decomposition with the wavelet packet decomposition.It is solvedthat the wavelet decomposition is only effective for low frequency signals and the redundancy of the wavelet packet decompositionis realized by Mallat algorithm.For the first time,the EEG signals in DEAP database are decomposed by the improved algorithm of semi-wavelet packet and convolution process.On the basis of ensuring the accuracy,the decomposition algorithm achieves the effect of high decomposition rate and low complexity.

关 键 词：脑电信号 小波 MALLAT算法 加速 

分 类 号：R318.04[医药卫生—生物医学工程]