Analisis Perbandingan Ekstraksi Fitur CC-PEV dan Subtractive Pixel Adjacency Matrix dalam Steganalisis Berbasis Machine Learning

Abstract

Steganalysis is a technique used to detect hidden messages in steganographic images. It involves detecting the existence of hidden messages in digital media. Recent steganography techniques have gained attention due to their security. This research presents a universal steganalysis method for blocking recent techniques in the spatial domain. The method analyzes using CC-PEV and SPAM feature extractor for classifying whether the suspicious image is stego or normal. SPAM feature extractor successfully detects all stego images with an accuracy of 75,28% while the CC-PEV feature is slightly left behind with an accuracy of 66,95% both using a linear kernel