Research on the Classification of Audio Doppler Signals Based on SVM
Download as PDF
DOI: 10.25236/icemit.2018.008
Corresponding Author
Donghua Luo
Abstract
In order to analyze Doppler signals effectively, a SVM based classification method is proposed in the paper to classify audio signals output by supersonic Doppler detectors. At first, the percentage method is employed to get the maximum frequency curve of audio signals. Then the characteristics of signals are acquired through the modulus maximum of wavelet transform. These features can reflect the characteristics of the whole signal effectively. At last, according to these features, classification of Doppler signals is realized through SVM. The classification results can display the situation of blood vessels of patients. The simulation results show that, compared with neural network classification method, method introduced in this paper have advantages of fast classification speed and high identification rate. The classification of Doppler signals has important clinical values, since it can provide doctors with recommendations and help them treat patients more effectively.
Keywords
Ultrasonic Doppler Signal, Maximum Frequency Curve, Classification of Signals, Svm, Modulus Maximum