Image gradient index for classifying digital camera and scanned images

This paper presents a new method for distinguishing images captured using digital camera from that using scanner. The method is based on the concept that the residual noise present in images capture device always distorts the directional change in the intensity or color in an image i.e. image gradient, and the resultant distortions are greatly obtained from printing and scanning process, compared to photography process. Hence, the gradient index values of scanned images derived from gradient values of such image should be lower than the ones obtained from photographic images. Based on the above concept, we propose a method for classifying digital camera and scanned images using image gradient index. The proposed method is tested and its performance is evaluated. The experimental results show that the classification performance up to 95% on average can be achieved using our method achieved.