ZHAO Yi 1, LI Jun 2 , HAN Wen 1 , XU Kangkang 1, WU Lidan 3, WANG Hui 3, XU Weibin 3
(1. Jingdezhen Ceramic University , Jingdezhen 333403, Jiangxi, China; 2. Ceramic Research Institute of Light Industry of China, Jingdezhen 333001, Jiangxi, China; 3. Jingdezhen Oceano Ceramic Co., Ltd, Jingdezhen 333000, Jiangxi, China)
Abstract: Due to the rapid development of mobile electronic devices, the design patterns on well-known brand ceramic tile products are prone to being photographed and counterfeited, causing economic losses to the brands. In order to authenticate the copyright information of ceramic tile texture images, a copyright anti-counterfeiting authentication framework based on generative adversarial networks is proposed. This framework converts copyright information into ceramic tile texture images, from which the copyright information can be extracted and visually imperceptible. The generated digital texture images can still correctly extract copyright information even after being printed and photographed. It is experimentally showed that the decoding rate is as high as 93% when capturing images of ceramic tile patterns containing copyright information. Our framework is a new method for copyright protection of ceramic tile products.
Key words: tile image; copyright authentication; adversarial generative framework