Spherical Panorama Image Watermarking Using Viewpoint Detection

Abstract

Even though interest in spherical panorama content has increased rapidly, few studies have examined watermarking techniques for this content. We present a new watermarking technique to protect spherical panorama images as well as view-images that are rendered with a specific viewpoint. Solving the watermark synchronization problem in the detection process requires finding the viewpoint of a view-image. Scale Invariant Feature Transform (SIFT) and Euclidean transformation matrix are used to find viewpoint information of a detection target viewimage. Using the viewpoint information, a view-image can be recovered to a source image and then we can detect watermark from it. The experimental results show robustness against several attacks such as JPEG compression, Gaussian filter, and noise addition attack.