Discrete polar complex exponential transform for image rotation duplication detection

作者：Zhong, Junliu; Gan, Yanfen

来源：International Journal of Signal Processing, Image Processing and Pattern Recognition, 2015, 8(7): 27-40.

DOI：10.14257/ijsip.2015.8.7.04

• Digital forgery
• Duplication detection
• Exponential transforms
• Polar harmonic transforms
• Rotation invariance

With the aid of powerful photo editing software, region duplication is a easy way to create digital forgery images. Many prior approaches to detect and locate similar or identical regions are introduced in various application fields successfully. But in most case, these existing region duplication detection approaches can only detect some simple forensics which cloned region is moved to a different position in the same image, and are not effective when the duplicated regions have been subjected to geometrical distortions and added noise. In this paper, we describe a new region duplication detection approach that is robust to rotation of the duplicated regions. We propose an algorithm by constructing rotation moment invariants and extracting invariance features of images for Two-Dimensional (2-D)image forgery fields based on Discrete Polar Complex Exponential Transform (DPCET) which is one of three types of Discrete Polar Harmonic Transforms (DPHTs). This Discrete Polar Complex Exponential Transform approach is improved from traditional Polar Complex Exponential Transform (PCET) and this transform also encompasses the features of orthogonality and rotation invariance. We introduce a new 9×9circleinstead of a unit circle in DPCET to calculate features of image more precision. And various experiments were performed to prove that effectiveness of our approach in detecting duplicated for geries, with translation, rotation, noise addition. And the experiment results also show that our approach is more robust to geometric distortions of the duplicated regions than other compared approaches.

• 出版日期2015-8-1