Yazarlar: Türker TUNCER, Emrah AYDEMİR
Konular:Bilgisayar Bilimleri, Bilgi Sistemleri
Anahtar Kelimeler:Ship identification by using sound, 1D-BP, Classification, Sound processing, Machine Learning
Özet: Sound classification one of the most important research issues for machine learning and applied computer sciences. By using sound classification method, many biometric applications/methods have been presented in the literature. This work presents a ship identification method by using sounds. This presented method is very simple and effective. This method has only two fundamental phases and these phases are feature extraction by one dimensional binary pattern (1D-BP) and classification with conventional classifiers phases. 1D-BP extracts 256 features from each sound and these sounds are forwarded to classifiers. To test this ultra-lightweight sound identification method, a ship sounds dataset was collected from YouTube. According to results, this method achieved 97% classification accuracy. This results clearly demonstrated merit of the presented 1D-BP based method on ship sound classification and sound based ship identification.