DCT Image Steganography Analysis for Privacy Preserving Data Mining

Abstract

We live in an information era where sensitive information extracted from data mining systems is vulnerable to exploitation. Privacy preserving data mining aims to prevent the discovery of sensitive information. Information hiding systems provide excellent privacy and confidentiality, where securing confidential communications in public channels can be achieved using steganography. A cover media are exploited using steganography techniques where they hide the payload's existence within appropriate multimedia carriers. This paper aims to study steganography techniques in spatial and frequency domains, and then analyzes the performance of Discrete Cosine Transform (DCT) based steganography using the low frequency and the middle frequency to compare their performance using Peak Signal to Noise Ratio (PSNR) and Mean Square Error (MSE). The experimental results show that middle frequency has the larger message capacity and best performance.