Using Machine Learning to Identify a Camera By Photo Image with On Board Communications

• Published in: Systems of Signals Generating and Processing in the Field of on Board Communications (Online) pp. 1 - 5
• Main Authors: Sineva, I. S., Tkachev, V. D.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 12.03.2024
• Subjects: Analytical models; AutoML; camera identification; Cameras; Filtering; local binary patterns; Machine learning; Machine learning algorithms; Metadata; multi-class classification; Performance evaluation; photo images
• ISSN: 2768-0118
• DOI: 10.1109/IEEECONF60226.2024.10496726

This article explores the use of classical machine learning and Automnl methods to solve the problem of camera identification from photo images in the absence of metadata and limited computational and resource capabilities in on board communications. The experiments were conducted on the well-known Forchheim Image Database dataset using the Python programming language and libraries such as scikit-learn, auto-sklearn, skimage and pandas. The paper discusses the use of a local binary pattern (LBP) descriptor in combination with various filtering methods. A comparison of the classification results for each algorithm using different confidence and performance evaluation metrics was carried out. The experimental results showed that the Auto-sklearn algorithm achieved the best performance indicators when using the bilateral filtering method.