請用此 Handle URI 來引用此文件: http://hdl.handle.net/11455/88049
標題: A Message Embedding Algorithm with Multiple Pixel Clustering Mechanism for High Dynamic Range Images
多像素群機制的高動態範圍影像資訊嵌入演算法
作者: Yen-Ching Ke
Wei-Sung Chen
Chung-Ming Wang
柯衍慶
陳煒松
王宗銘
關鍵字: high dynamic range images
message embedding
distortion-free
clustering mechanism
visualdifference analysis
高動態範圍
資訊嵌入
無失真
多群機制
視覺化差異分析
摘要: This paper presents a novel message embedding algorithm exploiting pixel clustering for high dynamic range images. We take advantage of homogeneous pixel representation and combine M pixels (M ≥ 2) as a cluster to generate sufficient statuses for message embedding. This approach allows us to adopt the triplet coding technique to increase the embedding capacity without incurring any image distortion. We propose a mathematical analysis to determine the most appropriate numbers of pixels in a cluster to achieve the theoretical optimal capacity. In addition, given the constraint of double precision representation, we suggest the affordable numbers of pixel in a cluster to achieve the highest affordable capacity. Our mathematical analysis shows that the highest affordable capacity is slightly smaller than the theoretical optimal capacity with the difference between 0.019% and 0.158%. We compare our scheme of using 2 pixels as a cluster with Ke's results of using a single pixel. The comparison indicates that our algorithm can provide larger payloads in the range of 317 ~ 2,397 bits. We adopt the tone mapping scheme to produce low dynamic range images to quantize the image difference. Besides, we employ the HDR-VDP technique to inspect the visual difference between the cover and stego images. The image difference quantization results show that no distortion is encountered. The probability map produced by HDR-VDP inspection is in grey color indicating that the detection probability of visual difference is null. In conclusion, we present a novel cluster-based message embedding algorithm. Mathematical analysis and experimental results demonstrate our algorithm does not produce any image distortion when concealing messages. It outperforms over our counterparts extending the feasibility of message embedding for high dynamic range images.
本文針對高動態範圍影像,提出一個以多群方式來達成資訊嵌入目的全新演算法,簡稱CLUTCOD演算法。我們利用同質性像素排列組合之特性,同時使用M個像素(M ≥ 2)作為一個群組來嵌入資訊。此外,我們也利用三區間編碼,善用M個像素群組所構成的狀態數,故演算法可以在不增加偽裝影像變動量的前提下,提升資訊的嵌入量。我們使用數學分析,推導出使用多少個像素為一群時,群組內單一像素的理論最佳嵌入量。由於狀態數之數值表示有精準度之限制,因此我們以數學分析,推導出群組內單一像素在精準度限制下,可容許最佳嵌入量。理論分析顯示:可容許最佳嵌入量略小於理論最佳嵌入量,惟兩者僅有0.019~0.158%之差異。我們以2個像素當作一個群組來嵌入訊息,並將結果與Ke學者僅使用一個像素的嵌入方式作比較。兩者之比較結果顯示:我們的演算法在不引起額外失真下,可以比Ke學者方法提升317 bits~2,397 bits的嵌入量。我們使用色調映射技術將高動態掩護與偽裝影像轉換成低動態掩護與偽裝影像,以供量化差異與視覺化差異分析。量化分析結果顯示:低動態掩護影像與偽裝影像間並無PSNR數值的差異;我們使用視覺化差異預測器(HDR-VDP)來量測低動態掩護與偽裝影像之視覺化差異。視覺化差異分析所產出之機率地圖顯示:掩護與偽裝影像之間,被人眼察覺有差異之機率為零。總結本文,我們提出一個以群組來達成資訊嵌入目的全新演算法,數學分析與實驗結果證實我們的演算法不會造成影像失真,優於文獻學者之結果。我們的演算法具有盲擷取之特性,擴展資訊嵌入在高動態範圍影像之應用範疇。
URI: http://hdl.handle.net/11455/88049
顯示於類別:第24卷 第1期

文件中的檔案:
檔案 描述 大小格式 
85637-1.pdf3.25 MBAdobe PDF檢視/開啟


在 DSpace 系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。