本論文提出兩種高容量的影像隱藏技術,分別為(1)三方向模數像素差值之高容量影像隱藏術,與(2)偵測竄改像素差值之影像隱藏術。一般而言決定影像隱藏術的好壞重點有二:遮蔽影像的嵌入容量與偽裝影像的品質。遮蔽影像的嵌入容量越大,對偽裝影像造成的破壞越大,如何在高嵌入容量下還能兼具一定的影像品質是本論文的研究重點之一。在第一個技術中,我們利用三方向像素差值隱藏術的概念提高嵌入容量,並以模數函數方式降低影像失真。雖然提高了嵌入容量以及影像品質,但在擷取秘密資訊時,無法確定所擷取的秘密資訊是否被竄改,因此對所擷取資訊的可信度也需存疑。因此在第二個技術中,我們加入了一個偵測機制,首先利用像素差值隱藏術嵌入秘密資訊,以最不重要位元替換法嵌入檢查碼以偵測秘密資訊是否遭受竄改,並且利用最佳像素判斷隱藏術減少影像失真。實驗結果顯示我們提出的技術,具有高嵌入容量又兼具高影像品質,並且能避開雙重統計分析。
In this thesis, we propose two high embedding capacity steganographic methods for images. The first is a high capacity steganographic scheme integrating TPVD(Tri-way Pixel Value Difference) and modulus function. The second is a digital image steganographic method for secret data authentication. There are two main issues in steganographic schemes: the hidden capacity and the quality of the stego-image. In general, a high capacity embedding scheme will produce a low quality stego-image. In this thesis, we try to keep the stego-image in high quality while affording a relative high embedding capacity. In the first method, we propose to use TPVD steganographic for stego-information embedding, and use modulus function for distortion reduction. This method increases embedding capacity and image quality. However it is unable to detect if the embedded secret data is changed. In the second method, we use a PVD(Pixel Value Difference) steganographic for secret data embedding, a LSB (Least Significant Bit) substitution method for detecting if the embedded data is changed, and an OPAP (Optimal Pixel Adjustment Process) method for reducing the image distortion. The experimental results show that our stego-images are imperceptible to human eyes and are not detectable by the Dual Statistics Method.