最近的幾年由於網路的發達以及數位多媒體資訊的普及，數位浮水印技術已經是主要用來保障數位資訊版權的主要方法，本論文提出了一種不需原始影像(blind)數位浮水印(watermark)技術，先將原始影像經過離散餘弦轉換(discrete cosine transform; DCT)至頻率域，再將片段的浮水印資訊嵌入DCT的係數中，並計算與原本係數的差異，以決定是否要調整嵌入浮水印資訊的DCT係數，此方式能夠提供與過去研究之數位浮水印技術相同的強韌度，卻能提供較佳的影像品質以及能嵌入更多的資訊，.還原的過程中僅需提供兩筆資料即可還原影像，不需要使用原始影像。同時，又因為是以位元角度將資訊嵌入到係數中，因此當影像經過失真式壓縮法壓縮之後，影像的檔案大小也不會有劇烈的改變。 In recent years, as a result of the development of the network and popularization of the multimedia information, watermark technology has been utilized for ensuring the copyright mainly method of digital information. This thesis proposed a new blind watermarking algorithm based on discrete cosine transform (DCT) for digital images. In the proposed method, the difference of DCT coefficients between the original information and watermarked information was calculated and used to determine the adjustment of watermarked DCT coefficients. The results show that the proposed watermark algorithm no only provides similar robust as previous watermarking investigations, but also improves the image quality of watermarked image and the information quantity of hiding ability. Moreover, the original image is unnecessary for acquiring the watermark. In the reconstruction procedure, only two bytes side information is required for digital image codec. The file size of watermarked JPEG compressed image approximates one without watermarking.
