Detection of Copy-Move Forgery in Digital Image Based on SIFT Features and Automatic Matching Thresholds

Abstract

Today the technology age is characterized by the spread of the digital images. It’s the most common form of information transmission whether through the internet or newspaper. This huge use of images technology has been accompanied by an evolution in editing tools which makes modifying and editing an image very simple. This paper proposes an effective and fast method for copy-move forgery detection. The paper adopts a SIFT technique for features extraction and wavelet technique to estimate the matching threshold. The low-frequency components are used to compute a dynamic threshold rather than a fixed threshold. Also, a method to remove false positive areas is proposed in order to produce the best possible results. The method can detect accurately and quickly the forgery even after more complex transformations. The experimental results refer that the proposed method can also detect forgery against post-processing operation and multiple copies.