Local patterns in the copy-move detection problem solution

Abstract

Embedding of duplicates is one of commonly used methods of image forgery. During this process, an image fragment is copied and pasted to another position in the same image. This is performed to conceal some important part of the image. A copy-move forgery detection algorithm aims to recognize duplicated areas in the image. This algorithm is based on calculating the characteristics in a sliding or overlapping window. In this paper, we compare the performance of copy-move detection algorithms that utilize a local binary pattern, a local ternary pattern, a local derivative pattern, and some extensions thereof. A distinctive feature of the used characteristics is their resistance to distortions inserted into the copy, such as linear contrast enhancement and impulse noise. This method also has low computational complexity.