李明穗臺灣大學:資訊工程學研究所邱鈺翔Chiu, Yu-HsiangYu-HsiangChiu2010-05-172018-07-052010-05-172018-07-052009U0001-0608200910211100http://ntur.lib.ntu.edu.tw//handle/246246/183367本論文提出了一種新的浮水印技術能使用在回復幾何變形影像與攜帶資訊。我們的浮水印技術中，許多樣板會被藏在小波轉換域中。這些樣板能提供回復何變形影像時所需的參考資訊。我們提出了一個演算法能在幾何變形的影像中出樣板的位置與所藏的資訊，值得注意的是我們的演算法並不需要原始圖片的援。此演算法是建構在貝氏網路的架構上。在幾何變形影像中樣板的位置和所的資訊能夠以機率的方式表示在貝氏網路中。藉由推論貝式網路，樣板的位置所藏的資訊可以被估測出來。估測出樣板的位置後，樣板的位移就得以被計算來。接著再使用差補法估測出每個像素的位移，如此一來便能藉由這些資訊將何變形影像回復。實驗的結果顯示出我們的浮水印技術能很強健的對抗幾何變，並且將幾何變形影像復原。A novel watermarking scheme which recovers an image from geometric distortionnd carries multi-bit binary message is proposed in this thesis. In our watermarkingcheme, several templates are inserted into the discrete wavelet transform domain anderves as the registration references. We provide a blind watermarking extractionlgorithm to extract the hidden message and locate the templates. The locations ofemplates and binary message are represented probabilistically in the Bayesian network.he displacement parameters of each template can further be obtained once theocations of the templates are estimated. The distorted image is recovered based on theisplacements of each pixel which is estimated by interpolation. Experimentsemonstrate that proposed watermarking scheme are robust to local geometric distortionnd is capable of recovering distorted images well.誌謝 i文摘要 iiBSTRACT iiiONTENTS ivIST OF FIGURES vihapter 1 Introduction 1.1 Image Recovery 2.2 Data Hiding 3.3 Paper Organization 4hapter 2 Related Work 4.1 Geometric Distortion Correction 5.2 Geometric Distortion Invariance Watermarking 5.2.1 Exhaustive Search 5.2.2 Image Registration 5.2.3 Geometric Distortion Invariance Domain 6.2.4 Method using a Template or a periodic watermark 6.2.5 Using Feature Point of the Image 7hapter 3 Background 8.1 Geometric distortion 10.1.1 Global geometric transformation 10.1.2 Local Geometric Transformation 12.2 Discrete Wavelet Transform 12.3 Contrast Context Histogram 14.4 Blob Detection 16hapter 4 Methodology 18.1 Overview 18.2 Watermarking Embedding 19.3 Watermarking Extraction 22.3.1 Similarity Measurement Model 24.3.2 Neighbor Relationship Model 25.3.3 The Procedure of Watermarking Extraction 26.4 Image recovery 28hapter 5 Experimental Result 29.1 Image Recovery 30.2 Data Hiding 33.2.1 Capacity 34.2.2 Fidelity 35.2.3 Robustness 36hapter 6 Conclusion 43EFERENCE 44application/pdf8462281 bytesapplication/pdfen-US局部幾何變形盲目偵測數位浮水印資訊隱藏幾何變形回復Local geometric distortionblind detectiondigital watermarkingdata hiddengeometric distortion correctionthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/183367/1/ntu-98-R96922093-1.pdf