| Summary: | 碩士 === 國立臺北大學 === 資訊工程學系 === 101 === We study reversible data hiding for halftone images. A reversible data hiding for halftone images is a method to embed a hidden watermark into a halftone image such that one can recover the original images perfectly when the hidden watermark is retrieved. To obtain this goal, we propose two novel reversible watermark-embedding algorithms called Data Hiding based on Block Density (DHBD) and Data Hiding based on Gray Code (DHGC) which are all based on the block-pattern histogram shifting techniques. In DHBD and DHGC, the used block-pattern histogram shifting techniques rely heavily on two novel block-pattern number assignment schemes called BD and GN, respectively. The block-pattern number assignment BD is constructed via a non-decreasingly block-density approach and GN is designed based on the well-known Gray code.
The main advantage of DHGC is that the number of pixels DHGC alters is smaller than DHBD due to the property of the Gray code. The metric based on the standard peak signal-to-noise ratio (PSNR) is a nice metric to measure the number of altered pixels made by the embedding scheme. The higher PSNR value, the less number of altered pixels. The experimental result shows that DHGC obtains higher PSNR value than DHBD. On the other hand, DHBD is designed to make the watermarked image approximately preserves the block density of the original halftone image. The modified peak signal-to-noise ratio (MPSNR) serves as a good metric to evaluate this effect. The experimental result shows that DHBD obtains higher MPSNR value than DHGC for some test images.
|
|---|
