Improvement of Differential Crisp Clustering using ANN Classifier for Unsupervised Pixel Classification of Satellite Image

An important approach to unsupervised pixel classification in remote sensing satellite imagery is to use clustering in the spectral domain. In particular, satellite images contain landcover types some of which cover significantly large areas, while some (e.g., bridges and roads) occupy relatively much smaller regions. Detecting regions or clusters of such widely varying sizes presents a challenging task. This fact motivated us to present a novel approach that integrates a differential evaluation based crisp clustering scheme with artificial neural networks (ANN) based probabilistic classifier to yield better performance. Real-coded encoding of the cluster centres is used for the differential evaluation based crisp clustering. The clustered solution is then used to find some points based on their proximity to the respective centres. The ANN classifier is thereafter trained by these points. Finally, the remaining points are classified using the trained classifier. Results demonstrating the effectiveness of the proposed technique are provided for several synthetic and real life data sets. Also statistical significance test has been performed to establish the superiority of the proposed technique. Moreover, one remotely sensed image of Bombay city has been classified using the proposed technique to establish its utility.