Main Article Content

Nancy Sahota
Rajnish Kansal


Nowadays, digital images have been used in many areas such as a evidence in a courtroom or for insurance claim, in a scientific fraud, within the medical patient history etc. Development of new powerful image editing software is also increasing parallel to increase on the usage of images in daily life. Two approaches to ratifying the authenticity of a digital image can be categorized as active and passive (blind).The active approach does not use actual digital content. Unlike the active approach, digital image forensics (also called passive image forensics) is a form of image analysis for finding out the condition of an image without the need for a priori information (such as embedded watermarks or signatures). The objective of copy-move forgery is to hide vital and useful information or data and perform region duplication in some section of an image. In this work, block based feature extraction has been proposed in which Fast Fourier Transform (FFT) coefficients has been evaluated for non-overlapping blocks at each pixel neighborhood for similarity matching. First of all, mean has been evaluated for each block and each block is put into a particular box of four pixel intensities such that 256 intensity values has been divided into 64 categories . After those FFT coefficients has been used for similar matching in which the Euclidean distance and the actual coordinate distance of the two blocks has been used so that different matching locations can be detected. Based on these location, whole image is divided into two categories i.e. Forged and non-forged area. For better accuracy morphological operations has been implemented on the segmented forged pixels so that noisy or mis-detected forgery areas can be eliminated. Proposed method gives 99-100% accuracy in copy move forgery detection


Download data is not yet available.

Article Details