Machine Learning based Approach for Water pollution detection via fish liver microscopic images analysis

This article presents an automatic classification approach for assessing water quality based on fish liver histopathology. As fish liver is a good bioindicator for detecting water chemical pollution, the proposed approach utilizes fish liver microscopic images in order to detect water pollution. The proposed approach consists of three phases; namely pre-processing, feature extraction, and classification phases. Since color and texture are the most important characteristics of microscopic fish liver images, the proposed system uses colored histogram and Gabor wavelet transform for classifying water quality degree. Also, it implemented Principal Components Analysis (PCA) along with Support Vector Machines (SVMs) algorithms for feature extraction and water quality degree classification, respectively. Collected datasets contain colored JPEG images of 125 images as training dataset and 45 images as testing dataset, respectively. Training dataset is divided into 4 classes representing the different histopathlogical changes and their corresponding water quality degrees. Experimental results showed that the proposed classification approach has obtained water quality classification accuracy of 93.3%, using SVMs linear kernel function with 37 images per class for training.