中間對折變動藏入位元內插技術

Jung 和Yoo 學者於2009 年提出影像放大技術，該方法使用相鄰像素將影像預測放大至原始大小的2 倍大，再將影像切割成2×2 的區塊，每個區塊依左上角的像素計算差值，再依據差值將機密訊息嵌入於放大後產生的像素之中。Jung 和Yoo 的方法每個像素的藏入位元數是固定的，該方法並沒有充分利用影像鄰近像素非常相似，可以藏入較多位元的特性，因此本研究擬利用鄰近像素的變異程度來判斷每個像素可以藏入的位元數。本研究首先使用影像放大技術，計算每個影像區塊的變異程度，統計變異數後，區分區塊級別，再依據區塊級別將不同位元數機密訊息切割並嵌入於預測像素中。針對機密訊息部份，本研究使用中間對折策略，該策略將機密訊息從中間對折，成為有正有負的數值，使其在嵌入時減少對影像品質的破壞程度。

Jung and Yao proposed a image interpolation technique in 2009 which uses the neighboring pixel values and expands a prediction image for two times and then divided the image into 2×2 blocks, non-overlapping , Before hiding secret message can be calculated the difference on the upper left in each block and hide the secret message into the predicted pixel. However , in their method each pixel can embed secret bits is regular, and their method have the advantage in similar pixels so that can hiding secret message more but they don’t use this feature so this study use the neighboring pixel of variance level to judge the payload. At first ,this study use the image interpolation technique and compute the variance degree in each block and then statistics the variance and separate the block level Last, according to the block level hiding the secret message into the predicted pixel for different bits. Focus on the part of secret message, we use the center folding strategy to folding the secret message from center so that let the secret message have positive and negative values and increase the quality when hiding.