Estimation of Informative Features in the Analysis of 2D Images of Bone Objects in Forensic

Abstract

In this paper we propose an approach to solving the problems of extracting and evaluating information features from 2D images of bone fractures and bone objects for further successful classifying fractures. As parameters or features, the textural characteristics of Haralick, local binary patterns of pixels for 2D images, Gabor filters, Laws energy texture characteristics for 2D images are considered. The analysis carried out on basis of information content estimation to select the features that are most suitable for solving the problem of bone fractures classification. This paper also describes the experiments and experimental data.