The block-based analysis for tamper localization is a prevailing mechanism in hash-based forgery detection algorithm. One of the main problems with a block-based analysis is its rough localization stemming from that the forensic hashes attached to the images are lack of contour information. In this paper, we present a novel tamper detection model which can generate an accurate tamper localization result. The main difference to the traditional algorithms is that the proposed model segments the image into closed regions based on strong edges. Then the color and position features of the closed regions are extracted as a forensic hash. Furthermore, a graph-based ranking algorithm is proposed to establish the region correspondence between the received image and forensic one by exploiting their intrinsic structure information. Experimental results demonstrate that the proposed tamper detection model is a promising method for precise tamper localization compared with state-of-the-art methods.

Published in: World Congress on Internet Security (WorldCIS-2017)

• Date of Conference: 11-14 December 2017
• DOI: 10.2053/WorldCIS.2017.0024
• ISBN: 978-1-908320-81-0
• Conference Location: University of Cambridge, UK