A Review on Digital Image Watermarking Using Neural Networks

Digital image watermarking technology has been developed rapidly during the recent few years and extensively applied to protect the copyright of digital image. The digital image watermarking scheme can be divided into two categories. They are visible digital image watermarking and invisible image watermarking techniques. The commonly used watermarking applications include copyright protection, authentication, embedded assn hidden information. Furthermore in the class invisible watermarks one may additionally categorize techniques as fragile and robust. Generally, a robust mark is designed to resist attacks that attempt to remove or destroy the mark. On the other hand a fragile mark is designed to detect slight changes to the watermarked image with high probability. A large number of techniques have been proposed and developed for digital watermarking. The traditional methods for digital image watermarking proposed early make use of Discrete Cosine Transform (DCT), Discrete Wavelet Transform (DWT), and Least Significant Bit (LSB) and so on for hiding the copyright information into host image. The major concern is that digital watermarking should provide the qualities like imperceptibility, robustness, security of cover image. However, these techniques suffer from the problems of unsatisfactory value of imperceptibility and robustness to various attacks. They also have issue related to security. This proposed paper presents a survey on various neural networks based digital image watermarking schemes. In addition it also provides directions for future enhancement of digital image watermarking using neural networks. The combination of digital image watermarking and cryptography can protect images against attacks.