以量化為基礎之數位浮水印技術與以小波為基礎基礎之數位簽章技術

Abstract

由於科技的發展與網路的進步，資訊與知識的傳播變得容易，這帶 給人們很大的方便，但也因此產生了盜版與竄改的問題。此外，數位多媒體時代的來臨，知識與創意都可以是商品，更使得版權的保護成為迫不容緩的問題。數位浮水印技術與數位簽章技術是可能的解決方案，已經引起了廣泛的研究。 在本論文中，我們提出了一個以量化為基礎的數位浮水印技術，及 一個以小波為基礎的數位簽章技術。對於以量化為基礎的數位浮水印技術，量化器的設計採用原始的Lloyd-Max 演算法，以保證量化誤差達到最小。此外，對量化器乘上某個倍數，使得此系統能在限制的量化誤差內達到最大的強韌性。在應用於影像上時，區塊DCT 的係數被用來藏入浮水印。 對於以小波為基礎的數位簽章技術，我們使用小波樹中對應相同空 間區塊的係數的統計特徵來決定極性。而數位簽章藉由與極性做互斥或運算藏入，得到的序列用來當作私密金鑰。我們提出的兩個方法對於抵抗一般的攻擊都有不錯的效果。

Due to the rapid growth of technology and computer network, the propagation of knowledge and information becomes more easily. It brings human being much convenience, but also induces piracy and tampering. With the coming of the digital multimedia age, knowledge and creativity can be valuable products, the copyright protection becomes an important problem. Digital watermarking technique and digital signature technique are possible solutions, and there exist many related researches. In this thesis, we will present a new digital watermarking scheme which is quantization-based and a new digital signature scheme which is wavelet-based. For the quantization-based digital watermarking, different messages are embedded by quantizing the host signals with corresponding quantizers. The designing of the quantizers adopt the conventional Lloyd-Max algorithm, which minimize the quantization distortion. Moreover, scaling the quantizers can maximize the robustness of the system under the constraint of the quantization distortion. In the applications on images, DCT coefficients are chosen to training quantizers and embed the watermark. For the wavelet-based digital signature, we use the statistics of the coefficients corresponding to the same spatial area in the wavelet tree to determine polarities. Then a signature is embedded by doing exclusive-or operation with the polarities, and the output sequence is used as a secret key. Both of two proposed method have good performance against common attacks.