An Efficient Box-Couting Fractal Dimension Approach for Experimental Image Variation Characterization

Abstract

This chapter presents a simple approach to compute fractal dimension (FD) on images. In the chapter, elementary experiments demonstrate that the variations between an original image pattern and its reproductions can affect the respective FD. A statistical analysis should be carried out, related to specific class of images, to access the applicability of the FD as a means to find imperfections in practical considerations. The encouraging conclusion is that this approach is faster and simpler than the usual. This can be readily extended to 3D images as well. The chapter proposes an efficient algorithm to estimate FD of images. The chapter suggests its use to identify online image deviation from a standard pattern. The chapter reports on some experiments on textile failings and comparison with four other methods.