Identify the Presence of Hidden Information Based on Lower Coefficients Value of 2DHWT Sub-bands

This paper introduces an approach which can be discriminated between normal images and images containing hidden information. After applying one scale 2D Haar Wavelet Transform (2DHWT) on the image, a new image sub-band is configured contains the Lower Coefficients Value (LCV) among image detail sub-bands. The differences in three directions between adjacent pixels of approximation and LCV image sub-bands are modeled utilizing 2
^nd
order Markov chains. Components of sample transition probability matrices (TPMs) are then utilized as feature vector for a proposed approach implemented by linear support vector machine (SVM) and Naive Bayes (NB) classifiers. Proposed approach is evaluated by detection accuracy of three common image embedding methods at three quality factors (QFs) with five different embedding rates. The comparison in using approximation sub-band with the other detail sub-bands reveals that the presented feature vector offers an improvement in detection accuracy.