Objects classification using fractal dimension and shape based on leaves classification

Abstract

The paper presents the process of creating an automatic clas-sifier based on leaf shape and fractal dimension. Fractal dimension is able to precisely describe any fractal object. The specific case of fractal dimension - the box-Counting Dimension - applies not only to the fractals. We can use it to describe objects that are not really fractals, but have self-similar patterns. Leaves are an example of such objects, but also clouds, cosmic objects, blood vessels, nervous systems, or waves which describe physical phenomena can be considered as those. This paper describes all the steps of creating an automat-ic classifier: the choice of the database to be analyzed, preparation of the clas-sification, classification process itself and the tests . The thesis presents in de-tails the steps which are necessary to undertake in the process of the trans-formation of the images in order to prepare the images for the classification. There are presented methods of features evaluation, and data visualization. Tests using the popular KNN and SVM classifiers answer the question whether the fractal dimension can be used as a feature in object classification.