Statistical computing for forensic firearm pattern identification: Evaluating a fisher linear discriminant analysis-based algorithm against fixed-impulse noise
Forensic laboratories analyze firearm-related evidence utilizing established ballistic identification systems such as the Integrated Ballistic Identification System (IBIS), Advanced Ballistics Analysis System (ALIAS), EVOFINDER Automated Ballistic Identification System, and CONDOR Ballistic Identifi...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | en |
| Published: |
Faculty of Science and Technology, UKM
2025
|
| Subjects: | |
| Online Access: | https://umpir.ump.edu.my/id/eprint/46590/1/JQMA%20%282025%29_1.pdf https://doi.org/10.17576/jqma.2104.2025.12 https://umpir.ump.edu.my/id/eprint/46590/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Forensic laboratories analyze firearm-related evidence utilizing established ballistic identification systems such as the Integrated Ballistic Identification System (IBIS), Advanced Ballistics Analysis System (ALIAS), EVOFINDER Automated Ballistic Identification System, and CONDOR Ballistic Identification System. However, these systems require physical verification by experts, resulting in the process being time-consuming. Previous studies developed a Fisher Linear Discriminant Analysis (FLDA)-based ballistic identification algorithm to address this limitation, encompassing image pre-processing, feature extraction, and identification. This study evaluates the robustness of the FLDA-based ballistic identification algorithm against fixed-value impulse noise, including pepper, salt, and salt-and-pepper noise. A dataset of ballistic images from five Vektor Parabellum SP1 9mm pistols (Pistols A–E) was contaminated with noise levels ranging from 10% to 90%. The results demonstrate that the algorithm maintains high identification rates exceeding 90% for images with up to 90% pepper and salt noise, utilizing maximum-ranked and minimum-ranked ordered denoising spatial kernels. Similarly, high identification rates of up to 80% were achieved for salt-and-pepper noise. These findings highlight the robustness of FLDA-based statistical computing techniques in forensic firearm pattern identification, reducing reliance on physical verification and expediting forensic investigation. Furthermore, this study aligns with the United Nations Sustainable Development Goals (SDGs), particularly SDG9 (Industry, Innovation, and Infrastructure), by fostering Artificial Intelligence (AI)-driven forensic advancements and SDG16 (Peace, Justice, and Strong Institution) by strengthening forensic accuracy in criminal investigations, ultimately contributing to national security and judicial efficiency. |
|---|