Ulla Delfana Rosiani, Mungki Astiningrum, Falah Ahmad Sabila, Oktaviano Andy Suryadi, Dwi Puspitasari, Husnul Hotimah
One of the criteria for quality catfish seeds can be seen in their movement. Agile catfish seeds showed good quality catfish seeds, whereas if not agile, they showed the opposite quality. However, problems arise when these catfish seeds only move a little even though the catfish is in good health. It became difficult for ordinary people to decide. Therefore, we need a way to detect this type of catfish seeds. This study proposes using the Phase Only Correlation (POC) method to observe the movement pattern of catfish seeds to identify their quality. POC is a motion estimation method that can determine the motion vector from one 2D image to another 2D image down to the subpixel level. The subtle movements of a moving object can be captured by this method. The movement of various catfish seeds can produce POC-based movement features that show the movement pattern of catfish seedlings. In this initial study, the data used were video data of catfish breeding movements recorded from catfish cultivators in as many as 40 videos in the form of 20 good-quality catfish and 20 poor-quality catfish. Movement observations were carried out in a container containing catfish seeds and a camera from a mobile phone device at a certain distance. The results obtained by the POC method can produce movement features that can detect the quality of catfish seeds with an accuracy value of 77.5% using the K-NN method and the Naive Bayes method of 85%. © 2022 IEEE.
State Polytechnic of Malang, Department of Information Technology, Malang, Indonesia