Local Artificial Image's Changes Detection

N.I. Glumov and A.V. Kuznetsov (Russia)

Keywords

Resampling, JPEG, duplicates, FourierMellin transform

Abstract

Algorithms for detection images local artificial changes (forensics) are proposed in this paper. There are various basic properties of images (the presence of duplicates, resampling, JPEG-compression application) analyzing which it is possible to allocate the changes in the image. For the time being the common detection methods of the listed properties do not meet full demands of photos validity neither on reliability, nor on computing complexity. In the current work we solved the task of development and characteristics improvements of these methods with the view of their possible use in detection of local artificial changes on large-sized images. Application of recursive processing methods in the running window allowed to accelerate the images processing considerably, and using the new features in the analysis made it possible to raise reliability of changes allocation, and also to improve stability of algorithms to distortions of images.

Important Links:



Go Back