Novel color image steganalysis method based on RGB channel empirical modes to expose stego images with diverse payloads

With the rapid development of technology in the modern digital world, covert communication of secret information as a payload without instigating visible attention by using steganography emerged as a possible threat. Steganographic methods select either image contents like edges or random regions of image for hiding payload. Steganalysis methods generally concentrate on content-adaptive algorithms of grayscale images but only few works concentrate on color image steganalysis. To address this issue, a generalized steganalyzer that can identify suspicious content created using steganography methods in digital color images is a need of the hour. In this paper, a novel FroFeat feature extracted from decomposed components of three color channels using empirical mode decomposition process is proposed to augment the existing color rich model features to detect stego images created using five content-adaptive and eight non-content-adaptive steganography methods. These empirical mode decomposed components eliminate the image content including edges and pave way to model the subtle stego noise hidden inside stego images. The proposed method is validated by comparing the performance metrics with existing state-of-the-art steganalysis models. Based on the experimental results, the proposed method achieves an average of 0.484 decrease in detection error for low-volume payload detection compared to the existing methods. Also in this paper, mixed generic color image steganalysis is performed to showcase the generalization ability of the proposed steganalysis method.