A novel technique for automatic shoeprint image retrieval

Area of Science:

• Forensic Science
• Computer Vision
• Image Analysis

Background:

• Increasing volume of crime scene images necessitates advanced forensic analysis techniques.
• Shoeprint evidence is gaining importance, comparable to fingerprints and DNA.
• Current shoeprint classification methods often require manual intervention.

Purpose of the Study:

• To develop a robust and automated algorithm for shoeprint matching.
• To evaluate the algorithm's performance under varying image conditions.
• To enhance the efficiency of footwear evidence analysis in forensic science.

Main Methods:

• Implementation of a novel shoeprint matching algorithm.
• Utilizing Hu's moment invariants for image representation and comparison.
• Testing the algorithm's resilience to image resolution reduction and rotation.

Main Results:

• The proposed algorithm demonstrates robust performance in matching shoeprint images.
• Image resolution decrease had a negligible impact on the algorithm's effectiveness.
• Optimal matching accuracy was achieved irrespective of image rotation angles.

Conclusions:

• The developed algorithm offers an effective automated solution for shoeprint matching.
• The method's robustness to resolution and rotation makes it practical for real-world forensic applications.
• This advancement can significantly improve the analysis of footwear evidence in criminal investigations.