Ocean Pollution Detection using Image Processing

Ocean pollution is increasing at an alarming rate and causing detrimental effects on the health of underwater flora and fauna. The anthropogenic activities are contaminating the natural water bodies which have adverse effects on humans as well as the environment. Hence monitoring of water bodies such as oceans is a necessity for sustainable water quality. Underwater images are distorted because of scattering of light. They are also faded because the intensity of sunlight decreases with the depth of water. In order to get improved quality of images, color correction method is used to remove casts and histogram equalisation is used to enhance contrast. Pollutants have been determined by feature extraction using gray level co-occurrence matrix. It extracts the features based on entropy, homogeneity, contrast, energy calculations to identify texture of an image for classification. Furthermore, images are compared with the trained neural network which has been classified into four categories namely oil spills, plastic, fishing net and dead coral reefs based on their contribution towards ocean pollution. This is a very simple analysing technique of monitoring oceans with no skilled operator required for the work. No hardware is required for processing and analysing the data, therefore reducing the cost. This paper introduces an easy and economical technique which can be used by scientists and environmentalists to predict measures for mitigating ocean pollution.