| Summary: | 博士 === 國立中興大學 === 應用數學系所 === 97 === The rapid development of information technologies has increased the range of circumstances in which information is transmitted over networks. The security of information and the transmission process are important. In recent years, many cryptographical schemes for encryption and authentication have been proposed and implemented to ensure information security. However, the use of cryptography is easily detected in the transmission process. Data hiding or embedding technologies are new solutions that can hide confidential information or encrypt it inside an image—this does not raise suspicions and therefore provides twice the security. However, the greater is the amount of data embedded, the greater will be the distortion of the cover image. Therefore, an important goal in current data hiding research is the realization of both a high embedding capacity and high image quality.
Some applications depend on the accuracy and integrity of the used images; hence, lossless reconstruction of the original cover image is essential to such applications. This goal is called reversible or lossless data hiding. This approach has been adopted in such fields as multimedia content authentication, law enforcement, medical imagery and astronomical work. The present investigation proposes a reversible scheme for hiding data that increases the embedding capacity and reduces the distortion of the embedded image.
This dissertation develops a scheme of reversible data hiding based on a block difference histogram. The scheme divides the cover image into non-overlapping identical blocks. In each block, a pixel is selected as the base pixel and the absolute differences between the gray level of the base pixel and those of the other pixels are calculated. These absolute differences are used to generate a histogram, and the histogram shifting method is applied to embed secret data and overhead information into the pixels of the cover image. This overhead information is generated in the embedding process and is used to extract data and recover the cover image in the extraction process. Both the block determination methods and the difference calculation methods are presented to increase the embedding capacity and reduce the changes in the gray levels of the pixels during embedding. The block determination methods can be used to determine the optimal type of block division in order to maximize the embedding capacity in each embedding process. The difference calculation methods use the median pixel, the maximum pixel or the minimum pixel in the block as the base pixel to calculate the absolute differences between the gray level of the base pixel and those of the other pixels. These differences are known as the median difference, the max difference and the min difference, respectively. The median difference can be used to increase the hiding capacity. The offset distortion method and the distortion reduction method are utilized to cooperate with the max difference and the min difference to reduce the distortion of the embedded image. Experimental results are presented to prove that the proposed methods are valid and outperform many other reversible data hiding methods.
|
|---|
