本篇研究論文針對灰階影像提出一個可適應性的影像偽裝學技術。其主要的研究貢獻是在嵌入大量的秘密訊息之後,同時也可以產生一張高品質的偽裝影像。本研究技術嵌入訊息容量的多寡,取決於原始影像的區域性複雜度的高低,在我們的方法中,一個影像內容的複雜度將被分成五個等級,等級愈低代表該區域影像紋理變化愈單調,因此嵌入較少的資料量,以免引起明顯地視覺失真;相對地,等級愈高則代表區域影像紋理變化愈複雜,因此將會嵌入較多的訊息。其所引起的失真度對於人類視覺仍然是不可視的。由於區域等級在嵌入完訊息之後,其等級大小仍然可被維持一致性,因此在訊息擷取時是不需要原圖的。最後,在實驗結果中,我們也將和現存可適應性偽裝學技術比較,証明本研究技術所產生的失真會比較少。This paper presents a novel adaptive steganography for digital grayscale image based on the human visual sensitivity so as to reduce the visual quality degradation after the data are hidden. The payload of each pixel is possible different in an image and the size of payload is assessed by its local image complexity. Basically, we hide more data into more busy area than the smooth area in which the original image feature can be maintained. The proposed scheme is blind since the degree of block complexity is hold before and after embedding data. The experimental results demonstrate the visual quality produced by our scheme is better than the existing technique.