I.V.Vovk, V.Yu.Semenov Automatic detection and recognition of wheezes based on the analysis of their autocorrelation function |
Acoustic bulletin, Vol. 8 ¹ 3, (2005) |
| An efficient method for automatic detecting and recognizing wheezes in the breath sounds is proposed. A dependence is obtained which relates the acoustic powers of periodic and noise components of the respiratory signal with the peak value of normalized autocorrelation function. A method for wheeze detection, based on comparing the peak value of normalized autocorrelation function with varying threshold, is proposed. An important feature of the method is that it performs not only the classification by the "noise/periodic" property, but allows determining all frequency components present in a given wheeze. An efficient way for visualizing the proposed method's results is developed, that is based on forming the artificial signal that imitates frequency content of the initial breath fragment containing the wheeze. |
|
KEY WORDS: respiratory sounds, wheezes, automatic recognition, autocorrelation function |
| TEXT LANGUAGE: Russian |