An image copy detection scheme based on edge features

This paper presents a novel scheme to detect copied versions of digital images based on edge features of the images. The presented scheme is the first to explore the detection of “merged copies.” To successfully detect merged copies and shifted copies, a center detection mechanism is embedded in our scheme and two kinds of clustering approaches are adopted. Experimental results confirm that our scheme not only withstands most modifications of an original image but also offers good detection performance when dealing with various rotations, 32- or 64-pixels shifting, up to 66% cropping, or merged copies. Therefore, our presented scheme will be very helpful in detecting copies of digital images transmitted in an Internet environment.