改进的阈值函数在心音信号消噪中的研究  

Research on Improved Threshold Functions in Noise Reduction of Heart Sounds

作　　者：尹明 刘欣 王燕[2] 张艳[2] 林莉[2] 

机构地区：[1]解放军95538部队 [2]西华大学电气信息学院,四川省信号与信息处理重点实验室

出　　处：《电声技术》2010年第10期48-53,共6页Audio Engineering

摘　　要：为弥补传统阈值函数存在的不足、提高心音信号的消噪精度,提出了3种改进的阈值函数。试验中,首先对传统的软、硬阈值函数及3种改进的阈值函数进行去噪性能对比分析;其次,采用基于小波分析的频带局部能量特征提取法对消噪后的心音信号进行特征提取,并运用模糊Sammon映射算法将六维特征数据映射成二维以显示其散布特性。结果表明,改进的阈值函数二能更有效地消除噪声,其消噪后的心音信号特征向量较全面地反映了心音信号的特征。To compensate for the deficiencies of traditional threshold functions and improve the denoising accuracy of the heart sounds （HSs）, three improved threshold functions are proposed in this paper. First, the traditional soft and hard threshold functions and the three improved threshold functions were used to do noise reduction for many cases of HSs respectively and their denoising effects were assessed. Second, the feature extraction method based on wavelet transform is applied to extract the features from the denoised clinical HSs, and Fuzzy Sammon mapping algorithm is used to project the six-dimensional feature data of HSs into a lower two-dimensional data to visualize the dispersion characteristic of individual feature. The experimental results indicate that the second improved threshold function can eliminate noise more effectively, and the feature vector of the denoised HSs reflects the characteristics of heart sounds comprehensively.

关 键 词：心音 小波变换 阈值函数 消噪 特征提取 模糊Sammon映射算法 

分 类 号：TN911[电子电信—通信与信息系统]