Prawns Identification Based on Edge detection and Feature Vectors

Prawns identification is one of major task in Indian fisheries. In this paper we developed algorithm for prawns identification based on edge detection and feature vectors. By edge detection method and boundary connection we extracted prawn object form background. After extraction of object form background object and divided into eight different regions where the majority of information is available for identification of prawn. For eight areas we applied discrete wavelet transform and extracted feature by considering high-high Sub bands where the edge feature exists. All eight features are combined and formed as feature vector. Finally minimum distance is calculated between for obtained feature vector and database feature vectors. Minimum distance is considered as identified prawn. We conducted experiment of thousand prawns and got 80% accuracy.