Wavelet-based Image Splicing Forgery Detection

Abstract

Digital image processing is a progressive field which has made development over period of time in a way that it becomes easy to play with artifacts of image by manipulating them using transformation such as copy-paste, copy-move, rotation, smoothing of boundaries, scaling, color enhancing, resizing, addition of noise, blurring, compressing etc. Forgery performed with a digital image, raising a doubt about the authenticity of it. Image splicing is one of the most used method for tampering an image by compositing two or many image fragments to create a spliced image. In this paper, a wavelet-based mechanism is proposed to detect image splicing forgery by taking edge information of an image as a distinguishing feature by performing edge analysis using wavelet transform. Haar-based Discrete Wavelet Transform (DWT) is used for edge analysis that decompose an image into four sub-images and it followed by Speed-Up Robust Feature (SURF) method which is a keypoint-based feature extractor technique. SURF extracts features from the decomposed images of DWT and used that features for performing classification using SVM linear classifier.