About image security using machine learning

Abstract

Seam carving and Watermarks are specific techniques for use in the area of image processing, where they have as characteristics the achievement of information security, concealment of information within images, as well as the reduction of their size, preserving your characteristics. Nowadays, such techniques are characterized as one of the main problems in the area of image security, and may even harm other areas such as the detection of modified images, an area specifically studied and applied by Computational Forensic Expertise. The present work proposal aims to act in two aspects, the first one in the detection of altered images using the Optimum-Path Forest classifier (OPF) and the Convolutional Reural Networks (CNN), and the second, in the detection of Watermark in images using the OPF classifier. (AU)