Article Preview
Top1. Introduction
With the rapid growth of the Internet, it is very expedient for people to acquire data from the Internet. Digital Medias namely videos, images, audios are conveyed over the Internet. At the similar time, safety issues termed as interception, interpolation have happened often. Keeping the information diffused over the Internet safe has become very imperative. Data hiding inserts the secret information into the cover image and then diffuses the embedded cover image to the receiver. The embedded cover image is customarily inconspicuous, and will not attract the attacker’s attention. The issue of this type of approaches is that the cover image cannot be improved completely after the secret information extraction. Numerous researchers have projected a lot of reversible information hiding approaches to resolve this issue (Li, Li, Yang, & Zeng, 2013; Wang & Yu, 2012; Nguyen, Chang, & Huynh, 2015).
Lately, to save the storage and bandwidth space of the embedded image, numerous data-hiding systems for compressed images have been projected in the literature. The chief reason is that the sizes of compressed images are much lesser than the sizes of the original images before and after information embedding. Numerous compression methods, i.e., JPEG, JPEG 2000, block truncation coding (BTC), and vector quantization (VQ) has been implemented for embedding information to achieve both a good compression rate and good inserting capacity. If the embedding algorithm has no ability to recuperate the distorted pixels back to their original ones, then this kind of embedding is named lossy embedding. On another aspect, if the stego image can be improved to its original state after extracting the secret information, the consistent embedding system is termed lossless or reversible information hiding (Wang, Lee, & Chang, 2013; Tsai, Hu, & Yeh, 2009).
Reversible data hiding (RDH) entrenches secret information into a host image and can recuperate the original host image flawlessly after the secret information have been extracted from the marked image. The reversibility of RDH is extremely anticipated in some special implementations like military and medical image processing, where no distortion is acceptable. Numerous RDH approaches have been projected in current years. Amongst them, difference expansion (DE) (Li et al., 2013) and histogram shifting (HS) (Wang & Yu, 2012) are the two pivotal methods. In DE algorithm, the host image is alienated into pixel pairs, and the difference value of two pixels in a pair is prolonged to carry one information bit. By adaptively shifting histogram of prediction errors, only designated bins are evacuated to host the payload. This makes the subsequent contributions: (1) performance detailed investigation on the conventional HS based RDH approaches and summarizing their development trends; (2) prioritizing prediction error bins with the help of adaptive shifting and hiding to overwhelm distortion; and (3) adapting prediction errors in group to embed multiple bits (variable length, up to 5 bits) at a time for refining hiding efficacy (Lou, Chou, Tso, & Chiu, 2012; Hong, 2012; Huang, Tseng, Hwang, 2013). The points contiguous to peak point are utilized to embed secret bits. We also adopt the localization to make the histogram of embedded cover image continuous so that the embedded cover image will not attract the attacker’s attention. The capacity is augmented rapidly due to the localization with multilayer embedding (Pan, Hu, Ma, & Wang, 2015).