Please use this identifier to cite or link to this item:
Title: 一種利用混合轉換之強韌型數位影像浮水印技術
A Study on Mixed Transform Technology for Robust Digital Watermark of Digital Image
Authors: 莊謙本
Chien-Pen Chuang
Pin-Chun Lan
Keywords: 尺度不變特徵技術
scale-invariant feature transform
discrete wavelet transform
discrete cosine transform
noise visibility function
Issue Date: 2011
Abstract: 本篇論文針對數位影像設計一個具有強韌性質的數位浮水印技術,目的是為了保護數位影像的智慧財產權。一般而言,對於浮水印影像的攻擊可分成幾何攻擊與非幾何攻擊兩大類,一個良好的強韌型數位浮水印演算法要能承受上述的攻擊。然而,在大部分抵抗幾何攻擊的研究中,有著浮水印嵌入容量受到限制的問題,有鑑於此,本篇論文使用尺度特徵不變技術來抵抗幾何攻擊,並且藉由二維條碼本身的高容量和高容錯特性,提升浮水印容量。結合離散小波與離散餘弦的混合轉換,並加入雜訊可見函式的考量,可以改善浮水印影像的視覺透明度。由數個實驗結果來展示演算法的效能,測試的攻擊模式有影像壓縮、縮放、模糊、銳化、對比以及亮度等攻擊,並與另一現存的浮水印演算法做比較,可以擁有較高的影像品質與攻擊抵抗能力。本篇論文的浮水印演算法能有效地兼顧強韌度與視覺透明度,進一步達到保護數位影像的智慧財產權之目的。
This thesis focuses on the study of robust watermarking techniques to protect the intellectual property rights for digital images. In general, watermark attacks can be divided into two classes: geometric attacks and non-geometric attacks. A robust digital watermark algorithm can resist the above attacks. Recently, most of the researches focus on resisting geometric attacks with watermarking techniques but it’s not recommended due to the restricted watermark embedding capacity. To resolve the above problem, this thesis proposes a watermark algorithm based on scale-invariant feature transform. The improvement of watermark embedding capacity depends on two-dimension barcode’s storage of high capacity of information, error-correction ability. The watermark algorithm uses a combination of the discrete wavelet transform, the discrete cosine transform and noise visibility function to improve the perceptual transparency. The experimental results demonstrate the performance of the proposed scheme. The watermark attacks uses in the experiments include image compression, resize, blurring, sharpening, contrast and brightness. Finally, the experimental results show the performance of this scheme to be superior to that of an existing one.
Other Identifiers: GN0698700262
Appears in Collections:學位論文

Files in This Item:
There are no files associated with this item.

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.