Improving Prediction based Digital Image Reversible Watermarking by Neural Networks

Abstract

Abstract—Digital image reversible watermarking hides data in a digital image in such a way that it is possible to recover the original image after the watermark extraction. The prediction is very important in the prediction based reversible watermarking. The more accurate prediction yields smaller prediction errors and therefore more efficient embedding procedure. In this paper we propose a high capacity prediction based reversible watermarking through the use of artificial neural networks. The structure of artificial neural network is based on the human brain nervous system and has learning ability so it is a powerful tool to predict. We have used two artificial neural networks for predicting. Artificial neural networks provide higher accuracy than coordinate logical operations. Test results indicate that artificial neural networks predict more accurate compared to coordinate logic operation. Also the results show that the proposed method achieved to higher capacity and quality than many other methods.