Identification of image acquisition chains using a dictionary of edge profiles

This paper presents a novel framework for the identification of an image acquisition chain using dictionaries of edge profiles. We investigate how edges, one of the most common features present in images, are transformed during the image acquisition chain. A dictionary of edge profiles is constructed corresponding to edges obtained from known devices and at different stages in the chain. The processing chain of a query image is then identified by feature matching using the maximum inner product criteria. Experiments have shown that the proposed method is able to identify the sources and acquisition stages of query images. It also has good performance in both recapture detection and chain identification applications for natural scene images.