Abstract

The paper deals to propose a hybrid watermark embedding and extracting technique. SVD and DWT methods are used for watermark embedding because DWT method is more flexible and provides a wide range of functionalities for still image processing.

Further a significant image attacks are carried out on the watermarked image and the watermark is extracted by the help of proposed watermark extraction algorithm based on Back Propagation Neural Network (BPNN).In the current era of online business such type of technique can play a significant role to identify the copyright of the product even there are numerous image attacks on the watermarked image.

It can lead to save the owner of the product from a heavy loss and restrict the further theft of the products. The proposed watermark embedding and extracting algorithms are tested on MATLAB 8.2 with various parameters such as PSNR and MSE. The result shows that after a numerous image attacks like paper and salts, mean, median, shear, noise, crop and rotation the watermark is identifiable.