植基於影像內插與像素差值之可適應性高動態範圍影像資訊隱藏演算法

Abstract

本文提出一套植基於影像內插與像素差值之可適應性高動態範圍影像資訊隱藏演算法。首先，我們利用先前研提之取樣方法取得參考像素，並藉由最低有效位元取代技術的協助，將訊息嵌入至參考像素中以有效提升訊息嵌入量，並依據門檻值將每一區塊分成低複雜度與高複雜度兩類。低複雜度區塊內的每一像素，我們採用最低有效位元取代技術將訊息嵌入於其中，使得每一像素能保有基本的訊息嵌入量。高複雜度區塊則先使用影像內插技術取得內插影像，接著計算出內插影像與輸入影像的差值來估算訊息嵌入量，最後藉由另一門檻值的協助決定該像素的訊息嵌入方式，以避免失真程度過大。實驗結果顯示，相較於先前演算法，本文所提技術能擁有更高的訊息嵌入量，且偽裝影像有令人滿意的影像品質，使得可適應性資訊隱藏技術在高動態範圍影像上有進一步的發展。This study proposes an adaptive information-hiding algorithm for high-dynamic-range images based on image interpolation and pixel-value differencing. First, we obtain the reference pixels from the input image and embed the secret message into them based on least-significant-bit substitution method. Thereafter, a threshold is used to determine the block complexity and different embedding strategies are adopted for data embedding, either LSB substitution or pixel-value differencing. The embedding capacity for each pixel in complex block is obtained adaptively based on the difference between original and interpolated pixel value. Experimental results show that the proposed algorithm can have more embedding capacity accompanying with satisfactory image quality than previous algorithm.