A Smartphone Data Steganography Framework Based on K-mean Image Selection and Random Embedding in Three Image Channels

Concealing information within smart devices during the Internet of Things era presents a formidable challenge. In this study, we introduce a novel steganographic framework designed for smartphones, comprising two distinct layers. The initial layer employs an unsupervised K-means algorithm to categorize smartphone images into three clusters. Images from the first two clusters are selected for the data embedding process. The clustering mechanism aids in the identification of images with the lowest likelihood of being deleted by the user. To improve data availability, two different images are employed for the embedding process. In the second layer, a linear random number is used to embed the data through an XOR operation in the two least significant bits of each channel in the RGB color model. The framework’s performance is assessed and compared to two other frameworks using metrics such as peak signal-to-noise ratio (PSNR), embedding capacity, and mean square error (MSE). Our findings demonstrate that the proposed framework achieved higher PSNR and greater embedding capacity in comparison with the other frameworks.