Imperceptible Watermarking Research Based on Sparse Approximation
| Autor: | YA-HSIUNG HSIEH, 謝亞雄 |
|---|---|
| Rok vydání: | 2014 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 102 Watermark is the general method for information hiding. For digital image, most research is imperceptible watermarking. It can append data secretly and keep image look like complete. Its biggest challenge is damaged by malicious attack or digital compression. Thus, robustness is important for watermarking technique. The current researches usually transform the spatial domain into Frequency Domain and then watermarking. It can hide watermark imperceptibly and have some robustness, but it recently gets targeted attacks which is high-frequency or low-frequency destruction. It caused a lot of problems in the frequency domain watermarking research. Watermarking is less researching on the spatial domain, there are two research directions, one is changing the least significant bit to hide information, and the other is using vector quantization technique. LBS technique has imperceptibility but it also has bad robustness. VQ technique is facing codebook size problem. If it have big codebook, it will spend much time to calculate, otherwise, it will get low quality image. In order to avoid the frequency domain watermark against malicious attacks, and therefore based on the spatial domain watermarking. this paper improve the VQ method by using Sparse Approximation technique to research a new method for imperceptible watermarking on spatial domain. Our method controls the codeword, which select by Sparse Approximation, and corresponding coefficient to hide different watermark. Then, depending on each watermark can hide multiple bits of data. The experiments show that our Algorithms can hide lot of information, and it can against damage and keep image complete. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|