Visible and Thermal Images Fusion Model for Security Threat Detection: A CNN Study

Abstract

As illegal use of fire arms continues to pose security threat across the globe, the need for novel detection technique is timely. In this study, visible and thermal images fusion model for security threat detection: a CNN study was investigated.  The study deployed the fusion of visible and thermal images techniques to design a Convolutional Neural Network (CNN)-based model for security threat detection. Thermal imaging data, preprocessing, supervised learning, detection and classification techniques of concealed objects was utilized in the model design. The study employed performance metrics, such as accuracy, precision, recall, and F1-score, to evaluate the effectiveness of the model. Results reveal that the investigated model demonstrates improved detection rates compared to traditional methods, offering a unique solution for application insecurity threat detection.  Based on these findings, the study recommended among others that in the fast-paced landscape of security, it is vital to establish mechanisms for regular model retraining.