IEEE SIGNAL PROCESSING LETTERS, VOL. 22, NO. 1, JANUARY 2015 81 Reversible Image Data Hiding with Contrast Enhancement Hao-Tian Wu, Member, IEEE, Jean-Luc Dugelay, Fellow, IEEE, and Yun-Qing Shi, Fellow, IEEE Abstract—In this letter, a novel reversible data hiding (RDH) al- gorithm is proposed for digital images. Instead of trying to keep the PSNR value high, the proposed algorithm enhances the con- trast of a host image to improve its visual quality. The highest two bins in the histogram are selected for data embedding so that his- togram equalization can be performed by repeating the process. The side information is embedded along with the message bits into the host image so that the original image is completely recoverable. The proposed algorithm was implemented on two sets of images to demonstrate its efficiency. To our best knowledge, it is the first algo- rithm that achieves image contrast enhancement by RDH. Further- more, the evaluation results show that the visual quality can be pre- served after a considerable amount of message bits have been em- bedded into the contrast-enhanced images, even better than three specific MATLAB functions used for image contrast enhancement. Index Terms—Contrast enhancement, histogram modification, location map, reversible data hiding, visual quality. I. INTRODUCTION R EVERSIBLE DATA HIDING (RDH) has been inten- sively studied in the community of signal processing. Also referred as invertible or lossless data hiding, RDH is to embed a piece of information into a host signal to generate the marked one, from which the original signal can be exactly recovered after extracting the embedded data. The technique of RDH is useful in some sensitive applications where no perma- nent change is allowed on the host signal. In the literature, most of the proposed algorithms are for digital images to embed invisible data (e.g. [1]–[8]) or a visible watermark (e.g. [9]). To evaluate the performance of a RDH algorithm, the hiding rate and the marked image quality are important metrics. There exists a trade-off between them because increasing the hiding rate often causes more distortion in image content. To measure the distortion, the peak signal-to-noise ratio (PSNR) value of the marked image is often calculated. Generally speaking, direct Manuscript received June 14, 2014; revised July 19, 2014; accepted August 01, 2014. Date of publication August 13, 2014; date of current version August 26, 2014. This work was supported in part by the NSFC under Grant 61100169 and by the Fundamental Research Funds for the Central Universities of China. The associate editor coordinating the review of this manuscript and approving it for publication was Prof. Jing-Ming Guo. H.-T. Wu is with School of Digital Media, Jiangnan University, Wuxi, JS 214122, China, on leave at the Polytechnic School of Engineering, New York University, Brooklyn, NY 11201 USA (with the support of China Scholarship Council) (e-mail: htwu@jiangnan.edu.cn). J.-L. Dugelay is with the Department of Multimedia Communications, EU- RECOM, F-06410 Biot, Sophia Antipolis, France (e-mail: jld@eurecom.fr). Y.-Q. Shi is with Department of Electrical and Computer Engineering, New Jersey Institute of Technology, Newark, NJ 07103 USA (e-mail: shi@njit.edu). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/LSP.2014.2346989 modification of image histogram [2] provides less embedding capacity. In contrast, the more recent algorithms (e.g. [5]–[8]) manipulate the more centrally distributed prediction errors by exploiting the correlations between neighboring pixels so that less distortion is caused by data hiding. Although the PSNR of a marked image generated with a pre- diction error based algorithm is kept high, the visual quality can hardly be improved because more or less distortion has been in- troduced by the embedding operations. For the images acquired with poor illumination, improving the visual quality is more im- portant than keeping the PSNR value high. Moreover, contrast enhancement of medical or satellite images is desired to show the details for visual inspection. Although the PSNR value of the enhanced image is often low, the visibility of image details has been improved. To our best knowledge, there is no existing RDH algorithm that performs the task of contrast enhancement so as to improve the visual quality of host images. So in this study, we aim at inventing a new RDH algorithm to achieve the property of contrast enhancement instead of just keeping the PSNR value high. In principle, image contrast enhancement can be achieved by histogram equalization [10]. To perform data embedding and contrast enhancement at the same time, the proposed algorithm is performed by modifying the histogram of pixel values. Firstly, the two peaks (i.e. the highest two bins) in the histogram are found out. The bins between the peaks are unchanged while the outer bins are shifted outward so that each of the two peaks can be split into two adjacent bins. To increase the embedding ca- pacity, the highest two bins in the modified histogram can be fur- ther chosen to be split, and so on until satisfactory contrast en- hancement effect is achieved. To avoid the overflows and under- flows due to histogram modification, the bounding pixel values are pre-processed and a location map is generated to memorize their locations. For the recovery of the original image, the lo- cation map is embedded into the host image, together with the message bits and other side information. So blind data extrac- tion and complete recovery of the original image are both en- abled. The proposed algorithm was applied to two set of im- ages to demonstrate its efficiency. To our best knowledge, it is the first algorithm that achieves image contrast enhancement by RDH. Furthermore, the evaluation results show that the visual quality can be preserved after a considerable amount of mes- sage bits have been embedded into the contrast-enhanced im- ages, even better than three specific MATLAB functions used for image contrast enhancement. The rest of this letter is organized as follows. Section II presents the details of the proposed RDH algorithm featured by contrast enhancement. The experimental results are given in Section III. Finally, a conclusion is drawn in Section IV. 1070-9908 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.