A perturbative resampling approach to image steganalysis

We study the effect of perturbative operations on images from a steganalytic point of view. It is observed that the rates at which unique colors vary in response to image processing operations performed in a perturbative sense - by adding mild disturbances to the image - provides a mechanism to calibrate the size of the hidden content due to LSB matching steganography in images that were earlier stored in the JPEG format. We also explore possible extensions of this approach to raw images and those that were stored as JPEG and subsequently subjected to image processing operations.