A Basic Study of Cough Signal Detection for a Life-Support System

概要

論文の詳細を見る
Aged people who live alone are in particular need of a daily health check, medication, and of warm communication with family and friends. The authors have been developing a life-support computer system with such functions. Among them, a daily health check function with the capability of measuring blood pressure, detecting diseases from coughing, and so on would in particular be very powerful for primary care. As a first step to achieving quick services for a daily health check with a personal computer, utilization of cough information is considered. Features of cough data are analyzed aiming at developing an automatic cough data detection method. This paper proposes a novel method for extracting cough signals from other types of signals. The differential coefficient of a low-pass filtered wave-form is first shown to be an effective parameter for discriminating between vowel and cough signals, and the relationship between cut-off frequency and cough detection rate is clarified. This parameter is then applied to cough signals mixed with vowel signals or white noises to evaluate robustness. The evaluation tests show that the cough feature can be perfectly detected for a 20 dB S/N ratio when the cut-off frequency is set to 24[Hz]. The experimental results suggest that the proposed cough detection method can be a useful tool as a primary care for aged people with a bronchitis like an asthmatic bronchitis and a bronchopneumonia.
社団法人電子情報通信学会の論文
2000-12-25

A Basic Study of Cough Signal Detection for a Life-Support System

スポンサーリンク

概要

著者

関連論文

スポンサーリンク