International Journal of Soft Computing
ISSN: OnlineISSN: Print 1816-9503
Abstract
In this digital world, it has become very important to save a digital product from illegal copy or reproduction. The technique, which has evolved for this, is known as Digital Watermarking. Several techniques based on spatial and frequency domain have been developed. However, none of them are full proof and exhibit all the desirable properties of watermarking to a satisfactory level. There is a gradual reduction in the fidelity of the cover image with the increase in embedded information content. This study discusses a special method based on Backpropagation Neural Network, which takes a cover image at the input of a Backpropagation Neural Netowrk and the network is trained to produce the desired watermark image. The cover image is taken in the original form and is not fragmented. After training, a random number is embedded in the higher precision bit of the cover image pixel and supplied with the trained network weights for the extraction of watermark. This guarantees authentication also. During the extraction stage, the watermarked image is supplied to the input of the trained network and output watermark image is produced. As the image processing operations do not affect the weights of the Neural Network, so the watermark image is resistant to various image processing operations enhancing robustness of watermarking.
How to cite this article:
Ashish Bansal and Sarita Singh Bhadauria . Generating Watermark Image with Authentication Using Full Cover Image and Backpropagation Neural Network.
DOI: https://doi.org/10.36478/ijscomp.2008.443.450
URL: https://www.makhillpublications.co/view-article/1816-9503/ijscomp.2008.443.450