New scheme for fast copy-move detection

Abstract

Image forgery detection is a challenging and important problem for many years. One of the most frequently used types of forgery (copy-move) is copying and pasting image fragment within the same image. There are exist two main types of copy-move forgeries: on the one hand, they are created without applying transformations (plain copy-move) and on the other – different digital transforms can be used to hide the traces of copy-move forgery (transformed copy-move). To solve the task of copy-move forgery detection in a complex way we propose several novel techniques, described further. The carried out research showed high quality of copy-move detection with intensity shift, contrast enhancement and additive noise distortions. In conclusion part we provide limitations of the scheme and the ranges of distortion parameters that can be detected.