Improved Approaches of Image Forgery Detection using SURF and 4-level DWT

Abstract

ABSTRACT

Today, there are various types of image tampering techniques, and copy-move forgery is one of them in which a piece of the picture is copied and pasted on another part of a picture. The study reviewed the several methods proposed to achieve this goal. This method generally uses both the feature point-based and the block-based algorithm. We use DWT (discrete wavelet transforms) for the decomposition of an image into several sub-images and to get the coefficients for each sub-image, the study uses4-level DWT. Considering the coefficients, this study determines the underlying size of the super pixel by SLIC algorithm to shape the non-covering sporadic squares. In order to take out the features, SURF algorithm is used to the irregular blocks. The feature is taken out from each irregular block; they are compared by calculating the Dot products between unit vectors. This gives the exactly matched tentacles of SURF feature in each block. Further, (RANSAC) algorithm is used to find the spacious area. A proposed is executed to assess the implementation of the plan the SURF calculation is utilized as a part of this study for feature extraction (FE). SURF is a productive other option to SIFT and it is substantially speedier, more hearty and give more accurate result instead of SIFT.

Keywords: image forgery detection, SURF, RANSAC,4-level DWT