Image processing techniques for identification of fish disease

Fish disease diagnosis is a difficult process and needs high level of expertise. Any attempt of developing the system dealing with the fish disease diagnosis and to overcome various difficulties is in it has still not met any extra ordinary success. Identification of diseased fish at early stage is necessary step to prevent from spreading disease. This paper recognizes and identifies the EUS (Epizootic Ulcerative syndrome) disease which is caused by bn Aphanomyces invadans, a fungal pathogen. It is a red spot disease and looks like ulcer hence it is generally misidentified by the people. The paper is divided into two parts, in the first part segmentation has been applied to enhance the image, various edge detection techniques have applied to get the useful information and Morphological operations have been applied on the EUS diseased fish image. In second part features are extracted from the EUS infected fish image through different feature Descriptors i.e. HOG (Histogram of Gradient), FAST (Features from Accelerated Segment Test) and classify the EUS infected and Non-EUS infected fish image through Machine Learning Algorithms and find the classification accuracy through Classifier. If PCA applied after feature extraction then it increases the accuracy as it is a dimensional reduction. The proposed combination of techniques gives better accuracy as compared to the others. The Experimentation has been done on MATLAB environment on real images of EUS infected fish database.