Identifying photographic images and photorealistic computer graphics using multifractal spectrum features of PRNU

A novel identification approach for identifying photographic images (PIM) and photorealistic computer graphics (PRCG) is proposed by using multifractal spectrum features of photo response non-uniformity noise (PRNU). 8 dimensions of mul-tifractal spectrum features of PRNU are extracted to represent the subtle differences between them, and the identification is carried out by using a support vector machine (SVM) classifier. Experimental results and analysis indicate that the proposed method can achieve an average identification accuracy of 98.99%, and has good performance in the ratios between training samples and testing samples. Besides, it is robust against some manipulations such as adding noise, JPEG compression and motion blur.