Detection of block DCT-based steganography in gray-scale images

In this paper, a new approach, the steganography detection system (SDS) is proposed and applied to the detection of block DCT-based steganography in gray-scale images, segmented into 8×x8 blocks. The differences in the coefficients of the block DCT transforms of the original are treated as features. SDS utilizes statistical preprocessing, over an observation region of each image, that generates feature vectors over the regions. These vectors are then fed into a simple neural network classifier. For the experiments conducted here, using 42 images, SDS achieves perfect detection rate with no misclassifications errors.