Multi-Stage Fish Classification System Using Morphometry

Abstract

The aim of this study is to create a multi-stage fish classification system with high accuracy rate. Classifications are based on biometric points of the fishes that consists of three main phases, data acquisition, feature extraction and classification. In the first phase, fish image database was collected, then features were extracted using morphometry and classified with three stage classifier model. Nearest Neighbor algorithm was used as classifier, and 25 fish species were classified with accuracy of about 99%.