Steganalysis of Color Images for Low Payload Detection

Steganalysis of Color Images for Low Payload Detection

연구 분야: Strategies

학회: IHIP 2019: Proceedings of the 2019 2nd International Conference on Information Hiding and Image Processing

This paper introduces a steganalysis model that uses statistical texture features and the machine learning approach to detect the presence of hidden data in a benchmark dataset of RGB color images. The work analyzes features of an RGB image in PPM format as a composite unit. The feature set used in this study consists of 120 features per color channel, which includes the basic and extended Gray Level Co-Occurrence Matrix (GLCM). The machine learning binary classifier that is selected for this work is the Support Vector Machine (SVM) algorithm. A public dataset of 10,000 uncompressed PPM clean images is used, and seven stego image datasets of 10,000 images each were created from the clean images, which were embedded with random secret data at payload ratios from 0.01 to 0.5 bit per channel, using 1LSB steganography technique The steganalysis results, showed detection accuracy values ranging from 56.18% for 0.01 bit per channel to 91.00% for 0.5 bit per channel.