Classification of Seagrass Habitat Using Probabilistic Neural Network

Seagrasses are very valuable asset which maintains the ecological and economic components of marine ecosystem. Seagrasses provides Food and shelter for marine animals, purifies water, provides nutrients, and decreases the speed of storm. As the human population is increasing, various types of threads on the seagrass is also increasing. Different activities of humans like sewage input, dumping of solid waste on the shoreline and anchoring of boats are main reasons of reduction in the population of seagrass. Remote Sensing is a technique through which the geospatial data of any location can be captured. So for the proposed research the remotely sensed Seagrass image of Andaman & Nicobar Island of India is collected using Google Earth. The image collected from Google Earth is high resolution image. High resolution image contains features in the form of RGB value. Probabilistic Neural Network (PNN) algorithm is a machine learning algorithm which is used for the classification of Seagrass from the data. Algorithm is applied on the extracted RGB value of the image. After applying PNN algorithm on the remotely sensed data, the classification of seagrass successfully performed with the accuracy of 99% and the Kappa Coefficient value is 0.99. The result shows the very good accuracy of the classification.