Image Tampering Detection using Local Phase based Operator

Image tampering detection is important due to many incidences of tampered images misuse. In this paper, we propose a hybrid approach for image tampering detection using range filter and texture descriptor. First we highlights important details of the image using range filtering. The range filter highlights the edges, contours and important details of the objects in an image. Further we apply texture descriptor based on local phase of the image in frequency domain is applied to extract crucial features of the image. This texture descriptor has high descriptive ability that provides sufficient image internal statistical information for detecting image forgery. The CASIA v1.0 database is used for performance estimation of our hybrid approach. For classification between tampered and original images Spectral Regression Discriminant Analysis and Support Vector Machine are used as a classifier. Our method outperforms some of the state of the art methods.