Fake Image Detection Using DCT and Local Binary Pattern

Abstract

In the technological era, digital images occupy an important position in different life's fields and image tampering has become affordable effortlessly, which results in a widespread of tampered and fake images through the internet and social media specifically. There are many techniques for image manipulation, some of the well-known methods are splicing and copy-move. Splicing can be defined as cutting part from an image and pasting it into another picture, while copy-move is about copying part of an image and pasting it into the same picture. This paper challenges splicing and copy -move forgery detection methods on CASIA TIDE databases. The proposed method is based on Local Binary Pattern (LBP), and 2D Discrete Cosine Transform (DCT), which are used for feature extraction. Afterward, a Support Vector Machine (SVM) classifier distinguishes real and manipulated images. Initial performance was increased by applying Local Binary Pattern (LBP) to the whole image rather than in a block-based fashion. The proposed model reaches the-state of- the-art in the CASIA TIDE v1.0 database with remarkable results in terms of accuracy.