Аннотация:Escherichia coli (E. coli) bacteria is one of the major biological contaminant of water in developing countries. According to WHO nearly 4 billion cases of diarrhoea and 2.2 million deaths in developing countries is because of E. coli. chlorination is an effective and scalable way of decontaminating and making water potable. The correct amount of chlorination is crucial; under-chlorination does not deactivate the E. coli completely and over-chlorination results in poor quality of water. Knowledge of the degree of contamination (measured in cfu) can be effectively used by local administration to decontaminate water by using exact amount of chlorine. Recently, Mobile Water Kit (MWK) has been proposed which can detect the presence of E. coli in water rapidly by manual inspection of pathogenic strains of E. coli. In this paper, we propose a robust and automatic method to quantify E. coli in water using image processing techniques. The choice of pre-processing techniques and identification of specific color based features makes the technique robust. Experimental results on 16 real images of water contaminated by E. coli taken by a mobile phone camera demonstrate the relevance of the features selected to identify the degree of contamination. The main contribution of this paper is the identification of an appropriate image color and structural features to estimate the degree of contamination.
