Detecting non-stereoscopic to stereoscopic image splicing with the use of disparity maps

Abstract

Splicing is a common image manipulation technique, where the content of more than one image is combined to form a composite image. With recent developments in image editing software, almost anyone can splice images. Splicing can be used to tamper with photographs to depict an event that never occurred. With the developments in digital stereoscopic (3D) cameras and images, it has become a necessity to be able to detect splicing in stereoscopic images. Several techniques have been developed to help a digital forensic investigator identify spliced digital images. However, there is a lack of research on exploiting depth information when detecting splicing in stereoscopic images.This paper proposes a technique to detect simple non-stereoscopic to stereoscopic splicing with the aid of disparity maps. An experiment was performed to test the technique on 35 spliced stereoscopic images. Results indicate that the proposed technique could be used to aid detection. However, detected areas require post-investigation to confirm splicing.