부산대학교 도서관

Uplifting the conventional method of classifying coconuts based on their meat type is an excellent avenue for developing the post-harvest methodology and utilizing modern technology, thereby increasing detection accuracy. This study intends to design and create Fast Fourier Transform-based coconut meat type detector through sound signatures. Specifically, it aims to (1) differentiate each type of coconut meat based on their acoustic response; (2) test and evaluate the device capability to identify the coconut meat-type according to sound features collected; (3) compare the accuracy of the proposed device against the manual method of determining coconut meat type. The major components include a metal rod, a microphone, a microcontroller, a USB interface, and an LCD. The investigation confirmed that the proposed device successfully detected six coconut meat types with an accuracy of 91.67%, higher than the manual method of detection, which obtained an average accuracy of 76.67%.