Robust digital image watermarking using DWT, DCT and probabilistic neural network

Abstract

Today is the multimedia era. Every hour huge amount of multimedia information like audio, video, image and text, is generated and distributed over internet. Digital images are the major part of multimedia information. With the advancement of technology, there is also a high risk piracy. Hence, methods for content authentication and copyright protection are essential to prevent piracy and make these methods more and more robust. Digital watermarking is a technique to secure privacy of digital information. The proposed approach presents a blind digital image watermarking technique in transform domain that uses probabilistic neural network (PNN) for the watermarking process. The proposed method aims to enhance the performance of watermarking using Probabilistic-Neural-Network including three-level DWT and DCT transformations. In the embedding process, embedding of the watermark is done by training the PNN network. The new trained embedded values are used at the extraction side and inverse three-level DWT and DCT are applied. The inverse transformation will give the extracted watermark. Robustness test is carried out with various attacks on the different images and calculation of performance metrics. The results show better imperceptibility and robustness to common image processing attacks.