Shoeprint retrieval: Core point alignment for pattern comparison.

Area of Science:

• Forensic Science
• Computer Vision
• Pattern Recognition

Background:

• Shoeprint analysis is crucial in forensic investigations.
• Existing methods for shoeprint retrieval face challenges in accuracy and stability.

Purpose of the Study:

• To propose an effective shoeprint retrieval method based on core point alignment.
• To enhance pattern analysis for forensic evidence in criminal cases.

Main Methods:

• Detecting contour points and simulating shoeprint sidelines using curve fitting.
• Identifying the core point via concave points on sidelines to partition the shoeprint.
• Calculating Zernike moments for regional pattern description and Euclidean distance for matching.

Main Results:

• Achieved a highest Average Precision (A_PR) of 0.726 using the first four Zernike moments.
• The Zernike method consistently outperformed moment invariant and GLCM methods.
• Demonstrated superior stability of the core point compared to the gravity center.

Conclusions:

• The proposed core point alignment method effectively aligns shoeprints for pattern comparison.
• This technique offers a robust solution for forensic shoeprint analysis.