Based On Discrete Wavelet Transform Digital Watermarking Technology

With the development of modern computer's computation capability, popularization of computers and the quick development of Internet, the security of personal information is being threatened. Traditional data encryption can partly prevent invalid distribution of data, but with the development of decryption technology, almost all the encrypted data are facing the danger of being cracked. Many hackers cannot help trying to crack the encrypted data while finding them, which makes the encrypted data are as dangerous as the unencrypted. In addition to that, when it comes to the validation and ownership check of data in use, traditional encryption cannot help anymore. With the development of modern data conceal technology; information can be treated in the same way as the natural body color to reduce the risk of being found. First, no one except the author himself knows the existence of the hidden information in the data, so there's nothing motivating the hackers'interest of cracking. Second, even if the carrier data are broken, the information hidden inside wouldn't spread over. Digital watermarking is one of the information conceal technologies. It can embeds the secret information into the carrier data, and check it out when necessary. It requires that the embedded data wouldn't affect the normal use of the carrier data, which makes the invisibility requirement of digital watermarking. Besides that, carrier data may be processed while transmitted before used, so digital watermarking should also be robust. Early research of digital watermarking dwelled on space field of images, but it proved not robust. It is proved that human vision has different sensitivity towards different part of images. In order to improve the robustility of digital watermarking, the watermark should be embedded into the important visual parts of the images, such as low frequency part. It leads to the development of transformation field watermarking. Present researches mostly concentrate on transformation field watermarking. The theory of Wavelet Transformation is a new math branch which emerged in recent years. It is a landmark since Fourier put forward Fourier Transformation in 1822, and can solve many problems which are hard for Fourier Transformation. Although Fourier Transformation has been widely used in signal processing field and can describe the frequency attributes of signals well, it meets problems while processing abrupt and non-stationary signals. Wavelet transformation can be treatedas the extension of Fourier transformation, which means it is a local transformation between space (time) and frequency. The general idea of the use of Wavelet transformation in image processing is to perform multi-resolution decomposition with the target image, and then processing the coefficients of the sub-images in different space and frequency. The core is the processing of the coefficients, such as image compress and coding. According to the tower decomposition put forward by S. Mallat, the target image can be divided into 4 frequency bands: horizontal, vertical, diagonal and the low frequency part, and the low frequency part can still be further divided. Images generated by wavelet decomposition have different attributes to the original image. The power of the original image concentrates on the low frequency part, which is the most important part to human vision, which there's much less power in the detail parts of the image. The high frequency parts present the horizontal, vertical and diagonal edge information of the original image. This article first introduces the fundamentals of digital watermarking, and makes some necessary statement on wavelet theory, and then put forward a new transformation field watermarking scheme based on human vision attributes. The watermark embedded is a series of standard normal distributed pseudorandom real numbers. The coefficients of the middle frequency part of the wavelet decomposition are modulated with the watermark series. The examination of watermark is in a way of mathematical statistics, and the extraction of watermark doesn't needs the original image. In the experimental phase, image with watermark has been attacked in several ways, such as image compression (JPEG), Gaussian filtering, median filtering, Gaussian noise, pepper and salt noise, resizing and embedding multiple watermarks. The results show that the new watermarking scheme is fairly robust.