Article Preview
Top1. Introduction
Reversible Data Hiding (RDH) is known as lossless and it is a scheme which is the key element in many applications in the field of secret communication, copyright protection and content authentication of digital multimedia (Barni, Bartolini, Cox, Hernandez, & Perez-Gonzalez, 2001; Tian, Zhao, Ni, Qin, & Li 2013). Reversibility of such scheme means losslessly reconstructing the original image as well as recovering the secret embedded information from the cover media.
Primarily reversible data hiding scheme has been proposed by Barton (1997). After that many other RDH schemes have been proposed and those can be find in the open literature. Mainly these schemes are classified into the following three categories such as lossless compression (Fridrich, Goljan, & Du, 2001; Celik, Sharma, Tekalp, & Saber, 2005), Difference Expansion (DE) (Tian, 2003; Hu, Lee, & Li, 2009) and Histogram Shifting (HS) (Ni, Shi, Ansari, & Su, 2006; Xuan, Yao, Yang, Gao, Chai, Shi, & Ni, 2006). Lossless compression means to compress the original media losslessly and make the space to embed additional information into it, which can be subsequently recovered losslessly after extraction of embedded data. DE-based embedding idea was initially proposed by Tian (2003) having higher hiding capacity but with a disadvantage as it degrades the quality of recovered image to a large extent. Improvement in the hiding capacity and quality of stego-image can be seen in many successive schemes. Later, several improved techniques for DE-based embedding have come into existence. They include prediction, sorting (Kamstra, & Heijmans, 2005; Sachnev, Kim, Nam, Suresh, & Shi, 2009) and location map reduction (Hu, Lee, & Li, 2009). HS-based scheme was initially proposed by Ni (Ni, Shi, Ansari, & Su, 2006) and proposed the algorithm by them is simpler than DE approach and requires less computation than most of other algorithms. Histogram shifting method uses peak and zero (or maximum and minimum) bins in the histogram of the input image pixel values and then, make the room for data hiding by shifting the bin intensities from peak to zero points. And hiding information has been done by modifying the pixels assuming the peak value. Moreover, HS provides a high visual quality and maintains a high Peak Signal-to-Noise Ratio (PSNR). However, the capacity that Ni's algorithm can provide might not be sufficient for most applications, so many scholars have studied and tried to improve Ni's algorithm. Intensities between the maximum points to raise the hiding capacity is used by (Hwang, Kim, & Choi, 2006), but the threshold of embedding capacity is not sufficient for hiding. Chung, Huang, Yang, Hsu, & Chen (2009) come out with a method used in dynamic programming procedure to maximize histogram shifting to improve the hiding capacity. His method increases capacity indeed, but it has two main drawbacks. It is suitable only for specific types of images. Further it needs much execution time which depends on pair of pixels and zero bins in histogram.