摘要
为解决EMD-IT去噪算法中阈值难以确定的问题,提出一种基于高斯白噪声能量分布的阈值估计方法。将含噪信号进行经验模态分解并估计各固有模态函数(IMF)中噪声的能量;根据模态单元阈值的含义,在各IMF中利用去除掉的模态单元包含的总能量等于噪声能量这一准则估计阈值。合成数据和实际心电信号的去噪仿真实验验证了该方法的有效性,其是自适应的且避免了阈值选择的主观性。
A threshold estimation method based on noise energy distribution for EMD-IT was proposed to solve the difficulty of threshold determination.Firstly,the noisy signal was decomposed using EMD and the noise energy of each level IMF was estimated.Secondly,the criterion that the total energy contained in the removed mode cells is equal to noise energy was used to estimate the threshold based on the meaning of mode cell threshold.The method can avoid estimating threshold subjectively and it is selfadaptive.Finally,simulations results with synthetic signals and real ECG signals verify the validity of the proposed method.
出处
《计算机工程与设计》
CSCD
北大核心
2014年第12期4386-4389,共4页
Computer Engineering and Design
基金
国家自然科学基金项目(61071188)
中央高校基本科研业务费基金项目(3142013026
3142014127)
华北科技学院应用数学重点学科基金项目(HKXJZD201402)
关键词
经验模态分解
固有模态函数
模态单元
高斯噪声
阈值参数
empirical mode decomposition(EMD)
intrinsic mode function(IMF)
mode cell
Gaussian noise
threshold parameter