论文部分内容阅读
Recently, many systems and approaches that employ heart sounds as physiological traits for biometric recognition have been investigated.However, there have not been a detailed and complete scheme so far, and the tests just stay in the experimental conditions.This paper designs a heart sound biometric system.This heart sound identification system is consisted of signal acquisition, pre-processing, feature extraction,and identification.The first step is using heart sound sensor HKY06C to extract the heart sounds signal.The second stage is to conduct pre-processing by using Wavelets, etc.Following by the extraction of three high recognition rate features which are (1) Mel-Frequency Cepstrum Coefficient (MFCC), (2) The Marginal Spectrum, (3) The First-to-Second Ratio(FSR).And finally we utilize the classifiers of the Gaussian Mixture Mode (GMM) for matching.We extracted the three features in particular, which have been proven suitable for non-stationary signal processing and capable of achieving correct recognition rate up to 90%.The results prove that the scheme has significant applicability and feasibility.