A Universal Digital Image Steganalysis Method Based on Sparse Representation
Zhuang Zhang, Donghui Hu, Yang Yang, Bin Su · 2013
With the development of modern steganography technologies, steganalysis has been a new research topic in the field of information security. Since JPEG images have been widely used in our daily life, the steganalysis for JPEG images becomes very important and significant. This paper propose a new steganalysis method based on sparse representation, intending to overcome the shortcomings of traditional classifiers in the field of universal steganalysis for JPEG images. Experimental results show that, comparing with the universal steganalysis method for JPEG stego images based on SVM, our method improves detection accuracy to some extent, and can avoid "over-fitting" problem in the process of classification. Experimental results also prove that our method is more robust than SVM when the detection images meet with Gaussian noises or Salt-Pepper noise.