Face Shape Classification based on Modified Relative Improved Differential Box Count Method

Abstract

The application of Fractal Dimension (FD) is not limited to a single domain; rather its usefulness can be extended to fields like segmentation, classification of shapes and analysis. Through this paper, the idea of classification of face shapes is presented using our proposed method -Modified Relative Improved Differential Box Counting (DBC) method. The method has been compared with the current existing FD calculation method, Improved DBC. The face data set is matched with popular geometric shapes, ellipse, rectangle, oval and triangle and classified according to the minimum FD difference between the face image and the given geometrical shapes. The experiments conducted with face sketch data sets have provided promising results and has also demonstrated that our proposed method can perform better in the field of classification, thus in general indicating that fractal dimension method is effective for classification of face shapes.