The Multi-Image Copyright-Protection Scheme Based on Neural Network

Abstract

Through the rapid development of computer technology and the popularity of Internet, more and more information is delivered through digital way. Because the digital information has the property of easy copy and rapid delivery, any one can distribute and obtain digital information easily. So, the copyright protection of valuable digital (included digital images) becomes an increasingly important research topic. In this paper, we proposed a multi-image copyright-protection scheme based on neural network. Through the use of Neural Network which has the capacity of learning, we can use the multiple original digital images as the input and the copyright pattern as the output. The neural network can learn the relationship between them. Then, some image which is needed to be verified can be as the input of the learned neural network and the output will show the copyright pattern if the image is legal. Our experiments show that this copyright-protection scheme has robustness. It can resist many different attacks of images, it still output the clear copyright pattern to verify copyright in such conditions. In addition, our scheme does not hide digital watermark (copyright pattern) in the original images actually, so our scheme will not cause the loss of the original images. Finally, our scheme can processes multiple images at a time, it cut down the cost of registration.