The Neural Network Digital Watermarking Algorithm

Today, crypto-graphical techniques had already reached a level of sophistication so that properly encrypted communications could be assumed secure well beyond the useful life of the information transmitted. In addition, the information hiding can fundamentally change the way that we think about information security. Therefore, this artiticle mainly made a research of digital image watermarking.First,I introduced the concept of digital image watermarking, the origin and development of watermarking in recent years; and then I gave the general model of image watermarking system, embedded guidelines and characteristics; after that, I described the typical kinds of image watermarking algorithms, such as the algorithms of the spatial domain image watermarking, frequency domain image watermarking, and so on; Then I focused on the characteristics of digital image watermarking applications. On the basis of the above research, this paper put forward the corresponding improving form of digital watermark algorithm conbining different applications, proseeding from the processing method and form of watermark signal.For details, by analyzing a number of current watermark signal processing methods for a meaningful signal, I firstly introduced the neural network adaptive digital watermarking algorithm and then put forward the process and experimental result digital watermark algorithm, which can make use of image scrambling technology watermark to cancell the noise of the watermark signal. According to human visual system, a combination of using D WT and DCT, digital image watermark was embedded. Then,in the process of watermark abstraction, use of Hopfield associative memory network of the powerful functionality can be achieved poor results on the extraction of the watermark after the treatment, to improve the quality extracted watermark. The wavelet multi-resolution and discrete cosine transform combination of treatment methods, extraction of the human host image eyes are not long sensitive to the watermark information in frequency region, thus ensuring the transparency of the algorithm. All the results are shown in this paper.In the end, some conclusion are draw out, the further researching work are also prospected and suggested.