An enhanced statistical approach for median filtering detection using difference image

In image forensics, detection of image forgeries involving non-linear manipulations have received a great deal of interest in recent past. Median filtering (MF) is one such non-linear manipulation technique which is quite often used in number of applications such as to hide impulse noises. Unlike other linear filtering operations, non-linear characteristics of median filtering makes it harder to detect using traditional forensics methods designed for detecting linear operations. This work utilizes adjacent pixels of difference image corresponding to input as well as MF version of input image for extracting proposed Local Expectation Features (LEF). These features when combined with an existing feature set Global-Local feature show significant improvement in MF detection. Evaluation of the proposed method under various forensic scenarios demonstrate consistent improvement in classification accuracy for a wide range of image sizes and compression ratios as compared to the existing methods for MF detection.