Differentiating between computer generated and natural images usingwavelet based transforms

The technologies making possible to change even the content of the digital media are developing fast. The demand for the development of mathematical and computational algorithms to determine the modifications on digital media has arisen and the studies in this area are gaining speed by the day. In this paper, we propose new statistical models for wavelet based transforms and try to differentiate between computer generated (CG) and natural images. Experimental results show that the features obtained from ridgelet and contourlet transforms are more successful than the wavelet based features in differentiating CG and natural images.