Research On Adaptive And Blind Digital Watermarking Algorithm Based On DCT

As an important branch of data blinding technology, digital watermarking is an effective technology of copyright protection of digital works and safeguard of data. It imbeds symbol information into digital works using the redundancy of Human Visual System, considers the balance between invisibility and robustness, in order to play role in copyright protection.This thesis introduces the basic theory and apply domain of digital watermarking, describes the features of Human Visual System and the application in digital watermarking. In this paper, a blind watermarking algorithm based on discrete cosine which combines with HVS is studied.Base on Human Visual System, in the process of watermark embedding, it combines with features of brightness and texture of images. In this paper, according to the features of brightness and texture to classify the sub-block of images after by DCT, then select the strength of watermark embedding adaptively, to achieve adaptive watermark embedding. The watermark is binary image, which adds scrambling, improving the security of watermark image. It uses DCT-based adaptive algorithm for the host of image carries out twice watermark embedding:first, the watermark image embeds into the host of the DC component of the image which using a parity quantization coefficient method, then the watermark image embeds into the host of the middle frequency of AC part of the image using a positive and negative quantization coefficient method. When picks up watermark from the embedded watermark of the image, it use the method of blind extraction from the DC component and the middle frequency of AC part of the image.The simulation experiments show that:by this algorithm, the watermark has good invisibility, and can resist the cut operation, as well as some common signal processing just as noise, filtering and JPEG compression, also achieve a better robust.