Mirror Reflected Cloning Detection

Abstract

Cloning is one of the manipulation techniques in image forgery. Cloning is the process of creating a forged image by copying a region of the image and pasting it into another region in the same image. Mirror reflected cloning is copying a region and flip it horizontally then pasting in the same image. A novel cloning detection scheme using SLIC and MIFT is proposed to detect this kind of forgery. The proposed scheme integrates block and keypoint-based cloning detection methods. First, by using Simple Linear iterative Clustering (SLIC) the host image is segmented into non-overlapping and irregular blocks. Second, Mirror invariant Feature Transform (MIFT) extracts block features from feature points of each block and then similar feature points are matched and are labeled to detect the forgery. Forgery Region Extraction algorithm shows the suspected region by replacing the feature points with small super pixels as feature blocks. By merging the neighboring similar local color feature blocks merged regions are obtained. Finally, a morphological process is applied on the merged regions to detect forgery regions.