完整後設資料紀錄
DC 欄位語言
dc.contributor.author黃祥哲en_US
dc.contributor.authorHsiang-Cheh Huangen_US
dc.contributor.author杭學鳴en_US
dc.contributor.author潘正祥en_US
dc.contributor.authorProf./Dr. Hsueh-Ming Hangen_US
dc.contributor.authorProf./Dr. Jeng-Shyang Panen_US
dc.date.accessioned2014-12-12T02:25:34Z-
dc.date.available2014-12-12T02:25:34Z-
dc.date.issued2000en_US
dc.identifier.urihttp://140.113.39.130/cdrfb3/record/nctu/#NT890428129en_US
dc.identifier.urihttp://hdl.handle.net/11536/67207-
dc.description.abstract本論文中,我們討論了最佳化技術的應用,同時也針對數位浮水印技術進行研究。更將最佳化技術的概念,落實在數位浮水印系統的設計之中。 在第一部分的最佳化技術應用方面,我們採用了基因演算法、擬似退火法、禁止搜尋法、以及最大下降法等四種方式,除了針對各個方法的運作原理深入探討與比較之外,也進行了相關模擬,並且將此四種最佳化方法,分別運用於向量量化器碼簿設計以及運用多重小波轉換的圖像識別系統。由實驗結果得知:採用一個或同時結合多個最佳化原理,訂定適當的代價函數,針對既有的方法進行改良,均能得到良好的效果。 於第二部份中,我們著重於數位浮水印技術的創新與改良。隨著電腦及網際網路的發展,數位資料的運用及傳輸,也隨之更加便捷。然而,此類數位資料的智慧財產權,卻可能在不知不覺中被侵犯。『數位浮水印』為一種保護智慧財產權的良好方法,其基本原理為:原創者將浮水印嵌入創作品之中,加入浮水印之資料,以數位方式傳至使用者處。若原創人懷疑某個使用者擁有盜版的創作品,運用數位浮水印抽出技術,即可證實該創作品之智慧財產擁有人。我們提出了多種創新的數位浮水印嵌入╱抽出技術,並模擬了常用的攻擊手法,以測試我們所提出方法是否強韌。實驗結果顯示:我們所提出的方法,大多能通過攻擊測試;此外,由模擬數據來看,我們方法的保護效果,也較某些既有的方法為佳。 結合本論文第一部分的最佳化背景知識,以及第二部分浮水印的設計方法,在第三部份,我們將最佳化技術的運作原理,應用於數位浮水印的設計概念之中。同時,於初步的基因演算法模擬中,獲得了一個兼顧輸出影像品質與抽出浮水印可辨識度的數位浮水印系統。因此,由目前的研究成果,我們可以預測:結合最佳化原理所設計出的強韌式數位浮水印系統,具有相當潛力,以對抗各種不同的攻擊測試。 在未來的研究方向上,我們將援用既有的成果,除了探討其他的最佳化方法之外,也將研究轉換域中新式的浮水印嵌入╱抽出技術,以期能獲得一個既強韌又有效率的數位浮水印系統。此外,我們也預期將設計現有系統的經驗,拓展到動態影像的浮水印系統,以使數位浮水印的原理,更能切合實際的應用。zh_TW
dc.description.abstractThe purpose of this thesis is to modify and implement several optimization techniques so that they can be applied to a few valuable applications and improve their performance. In addition, with the widespread use of the Internet, copyright protection is becoming an important issue. Therefore, another subject of this thesis is to propose new and effective watermarking schemes. These schemes have been tested with simulated attacks. Combining the concepts of optimization and digital watermarking, we are able to enhance the robustness of watermarks by using the optimization techniques. In the first part of this thesis, we use four popular optimization techniques, namely, the genetic algorithms, simulated annealing, tabu search, and maximum descent algorithms, to improve vector quantizer design algorithm and the pattern classification schemes. Simulation results indicate that our new schemes have better performances compared to the conventional methods. In the second part, we focus on the design and modification of the digital data hiding algorithms. Digital data hiding means embedding additional information, which can be watermark for ownership identification, into the original multimedia data (text). In the copyright application, if the copyright owner can extract the watermark from a suspected copy in question to see whether it is a legal copy. We propose and improve several new algorithms in data hiding for images, and the results show that our algorithms can survive certain types of attacks. Moreover, simulations also reveal that the performance of our algorithms is better than that of many existing algorithms. Finally, in the third part of this thesis, we combine the concepts of the previous two parts described above, and apply the optimization techniques to data hiding. Simulation results show that our system is not only producing the watermarked images with acceptable quality, it is also able to resist certain types of attacks.en_US
dc.language.isozh_TWen_US
dc.subject基因演算法zh_TW
dc.subject影像壓縮zh_TW
dc.subject分類zh_TW
dc.subject浮水印zh_TW
dc.subjectGenetic Algorithmen_US
dc.subjectImage Compressionen_US
dc.subjectClassificationen_US
dc.subjectWatermarkingen_US
dc.title基因演算法用於影像壓縮分類與浮水印zh_TW
dc.titleGenetic-Based Algorithms for Image Compression Classification and Watermarkingen_US
dc.typeThesisen_US
dc.contributor.department電子研究所zh_TW
顯示於類別:畢業論文