Font Size: a A A

Image Watermarking Algorithm

Posted on:2007-06-20Degree:MasterType:Thesis
Country:ChinaCandidate:X Y DengFull Text:PDF
GTID:2208360185461105Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
The multimedia information security problems including copyrights piracy emerged as side effects over network. In this dissertation, the technologies of digital image watermarking is studied. The main contributions of this dissertation are:1. A digital watermarking algorithm in binary text image:Because of the weak information hiding immunity of binary text image, many watermarking methods based on gray or color images are useless. In this dissertation, a new algorithm is proposed. First, according to the value of surrounding pixels, the probability of the value of center pixel is calculated to confirm the places where the watermark is embedded into. Then, according to the watermarking information, choose an appropriate method to make the less changes of text image. The algorithm is effective widely to all kinds of binary text including the text of handwriting, and the bulks of watermark can be controlled.2. A image blind watermarking algorithm based on integer wavelet transform:A new image blind watermarking algorithm is proposed in this dissertation. The coefficients in three high-frequency domains of integer wavelet transform with an image are adjusted. Then the coefficients are divided by an integer. The remainders according with some rules are used to denote the watermarking. This idea is a non-linear algorithm. One of its advantages is high capacity to watermarking. The watermarking capacity can get 2bit/pixel or more. On the other hand, the watermarking is very sensitive to content tampering. The algorithm is universal and can gain given requisition by combining other methods of image transform or quantization.3. Robust image watermarking algorithm for copyright authentication:We just need judge whether the watermarking exist, if the watermarking is used for copyright certification. The probability distribution of the detection output is discussed in two kinds of embedding model. A new embedding method is designed and analyzed, and the robustness of the method is discussed in the condition of existing noise. Last, we advise an idea of embedding multi-watermarking.4. Image Watermarking Resisting to Geometrical Attacks:Using periodically watermarking, an image watermarking algorithm is discussed in the dissertation, which is robust against geometrical attacks. Based on autocorrelation detection, the appropriate filter is used in image pretreatment, to improve the detection result. The effect of filter is analyzed theoretically, and two filter matrices are proposed. In the end, the examination effect is given, which prove that the method is advanced.
Keywords/Search Tags:digital watermarking, watermarking capacity, robustness, integer wavelet transform, probability distribution of the detection, Geometrical attacks, Autocorrelation detection, whitening filter
PDF Full Text Request
Related items