Classification of a Satellite Rural Image based on Fractal Dimension using Box Counting Method

Fractal geometry has sparked considerable interest for analyzing remote sensing imageries. Fractal models have been used in several image processing and pattern recognition applications such as texture analysis and classification. Applications of fractal geometry in remote sensing rely heavily on estimation of the fractal dimension (D). When land areas are clustered into groups of similar land cover, one of the most important things is to extract the key features of a given image. The present paper uses box counting method for determining the fractal dimension D for the classification of remote sensing imageries and identifying the various features present in the imageries. The features present in a typical remotely sensed Rural image such as Water bodies, Dry land, Vegetation, Forest, Rocky regions and Housing etc have been studied using the fractal dimensions. The studies show that the box counting method provide a very clear distinction between the fractal dimensions determined for the different features present in the satellite Rural imageries.