Font Size: a A A

Cough Sound Analysis Algorithm Design And APP Development

Posted on:2021-01-02Degree:MasterType:Thesis
Country:ChinaCandidate:X FanFull Text:PDF
GTID:2404330611966504Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
Due to the lack of objective evaluation of respiratory function,complications such as respiratory failure,respiratory distress syndrome,and lung infection in the acute stage are the main causes of death in patients with traumatic cervical spinal cord injury.The strength of the cough sound can also reflect the quality of the respiratory function.The analysis and treatment of the cough sound can assist the doctor in diagnosis.However,the traditional evaluation of breathing ability mainly depends on the measurement of the lung function instrument,the main disadvantage is that the operation is more cumbersome and the detection cost is higher.Therefore,the application of cough sound analysis in evaluating the respiratory function of patients has important social value and application prospectsThis article aims to improve the accuracy of cough sound recognition and the ease of use of auxiliary diagnostic software.It is planned to develop a cough analysis APP that assists doctors in diagnosis,solves the problem of cough sound recognition,and lays the foundation for subsequent evaluation of respiratory function.The main research contents include:1)Acquisition of samples: The improved double threshold method is used for endpoint detection,and the original samples are pre-emphasized,framed and windowed,and endpoint detection to obtain experimental samples.2)Research and implementation of true and false cough recognition algorithms: analogy and unvoiced voice classification method,selecting short-term zero-crossing rate and maximum autocorrelation coefficient as features,using a lightweight classifier Fisher to distinguish patients with frequent cervical spinal cord injury Sexual cough.3)Research and implementation of cough sound recognition algorithm: select short-term energy and Mel cepstrum coefficients as feature parameters,first use principal component analysis to reduce the dimensionality of feature parameters,and then use nonlinear support vectors based on radial basis kernel function The machine acts as a classifier and uses particle swarm optimization model parameters.4)Implementation of Android application: develop auxiliary diagnostic software based on the Android platform to realize cough signal collection,real-time display of cough signals,cough feature extraction,true and false cough recognition,cough sound recognition,file management,remote transmission and other functions.
Keywords/Search Tags:Cough Recognition, Respiratory Function, Fisher Discrimination, Support Vector Machine, Android Application
PDF Full Text Request
Related items