Examination of Shear Cut Mark Based on Local Multiscale Fractal Analysis

Abstract

Examination and identification of tool marks is traditionally carried out under a comparison microscope by forensic scientists. This manual process is dependent on the experience of the examiner, including subjectivity. In order to improve the reliability and repeatability of the process in criminal investigation and trial, automation and quantification for tool mark examination is demanded. Shear cut tool mark is a classic type of tool marks. Local self-similarity appears on the shear cut mark. A local multi-scale fractal analysis method is developed to classify shear cut marks by using the surface characteristics of artificial marks. A shear cut mark is divided into four regions based on the consistence of the image texture. The local extended fractal dimensions of four scales are calculated along x-direction and y-direction for each region of a shear cut tool mark. Total thirty-two dimensional features for one mark are grouped into a feature vector which represents the texture of a shear cut mark. Finally, the feature vector is input into a Bayes classifier to classify the marks. Experimental results show that the classification rate using local multi-scale fractal features is significantly improved in comparison with using global feature.