search for




 

Forensic shoeprint matching with image feature descriptors
Journal of the Korean Data & Information Science Society 2022;33:223-36
Published online March 31, 2022;  https://doi.org/10.7465/jkdi.2022.33.2.223
© 2022 Korean Data and Information Science Society.

Hojin Yang1 · Moonsoo Jang2 · Soyoung Park3

123Department of Statistics, Pusan National University
Correspondence to: 1 Assistant professor, Department of Statistics, Pusan National University, Busan 46241, Korea.
2 Ph.D student, Department of Statistics, Pusan National University, Busan 46241, Korea.
3 Assistant professor, Department of Statistics, Pusan National University, Busan 46241, Korea. E-mail: soyoung@pusan.ac.kr
This work was supported by a 2-Year Research Grant of Pusan National University.
Received December 31, 2021; Revised January 25, 2022; Accepted January 26, 2022.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Abstract
Pattern evidence such as fingerprint, shoeprint, handwriting is a common type of evidence in Forensic science, which would help identify the suspect to the scene. However, to this day, the pattern evidence relies on the examiners’ subjectivity and experience for the conclusion. We analyze the shoeprint with an objective matching method, MCCOMP, and quantify the similarity of the two shoeprints. The image descriptors of SURF, corner, BRISK and ORB, are extracted first and aligned through the maximum clique. After then, three similarities (clique size, %overlap, median distance) are calculated. Then the differences in mode between mated and non-mated distributions are compared. In addition, as it is common to find the partial shoeprint at the scene, one of the shoeprints in the matching set to be partial (e.g., toe, middle toe, middle, bottom). As a result, SURF and corner were the best image descriptors in the shoeprint analysis. When only partial shoeprint is available for the matching, the middle or bottom has the least information loss compared to the full shoeprint.
Keywords : Forensic science, image analysis, image descriptors, maximum clique, shoeprint.