Digital Image Processing and Recognition in Industrial and Public Environments

Abstract

Because the internet broadcasting, most photos or video streams need copyright protection or forgery detection. Nowadays, the metadata tries to embed and preserve copyright data and general information. The ownership metadata should never be removed, for an efficient protection, but practically it could be relatively easy extracted. This research proposes, based on a comparison between metadata and watermarking methods, a watermarking framework that inserts visual and hide information into digital images, instead of metadata protection. The visible watermark inserted into the host image could be removable or not, based on the objective or non-bijective embedding watermark function. A removable watermark allows the far end receiver to eliminate the visible watermark, whether it uses a framework that uses the invers embedding function. In this way, only the controlled receivers could profit by the clean photos/video stream. More of this, the hidden watermark could embed typical information that identify the owner.