A reversible data hiding method without rounding based on Haar wavelet transform: NON-R M.Kuslu* 1 and Y. Yalman 2 The techniques that allow to obtain hidden data and recover the image has recently been much sought after. The basic objective of the method provided in this study was to analyse coefficients of digital images achieved by wavelet transform in the frequency space, to hide data, and to recover original image without loss. Wavelet transform is an efficient method to find pixel coefficient characteristic of digital images. The proposed method (NON-R) hides data using shifted histogram high-frequency Haar coefficients without rounding into integer. Thus, both hidden data and cover image can be easily recovered. Reversibility is provided by retrieving the cover image after hiding into high-frequency wavelet coefficients and extracting the message from wavelet coefficients. Experimental results show that the NON-R has a better performance than its counterparts in terms of statistical and perceptual outcomes. Keywords: Wavelet transform, Reversible data hiding, Non-rounding, Histogram shifting Introduction Many data hiding techniques have been developed up to thepresent.Suchtechniquesallowhidingdataintodigital media (e.g. images, video or sound) for purposes such as authenticationandcopyrightproving.Thecoverobjectis definedasthemainobjectinwhichanembeddedmessage will be hidden, and the covered object is defined as the object that contains the hidden data. Short messages, multimedia short messages and digital animations can beusedasacoverimage. 1 These techniques are basically intended to hide confidential information from third par- ties. While such techniques serve to bona fide purposes, but sometimes can be used to spy on and infiltrate into an operating system using security holes. There are there substantial reasons why digital images are used often as a cover image. First, digital images are the most commonly shared data type on the internet and in networks; second, the hidden information is not suspected; and third, it is easier to display than other types of digital media. 2 In some medical and legal cases, it can be critically important to obtain the cover imageintheoriginalformratherthanobtainingthemess- age. As a result, reversible data hiding techniques are necessary for situations where it is not enough only to obtain hidden message from cover image, but also fully obtain the cover image after extracting the message. 3 In frequency domain, it is divided by itself into low- frequency components and high-frequency components for a one dimensional signal depending on the mother wavelet function ψ(t ) and its scale function ϕ(t ) using Wavelet Transform. 4 This is how 1st level expansion occurs. These components are also referred to as low- pass and high-pass frequency. For an image, which is a two dimensional signal, first each line information (data) is divided into low and high-frequency com- ponents,thenagaindividedintolow-andhigh-frequency components by the column information (data) that indi- vidually correspond to these two frequency domains obtained. It is important that data is hidden in which component and using what method. The performance of the proposed reversible data hid- ing algorithm (NON-R) was assessed for the payload andvisualquality.Asaresult,theNON-Rdisplaysabet- terperformanceintermsofnotonlypayloadbutalsoper- ceptual quality as compared to other methods in the literature. Therestofthearticleisorganisedasfollows:studiesin theliteratureisinitiallyaddressedin ‘Reversibledatahid- ing methods’ section, then the wavelet transform method usedisdetailedin ‘Wavelettransform’ section.TheNON- Rdetailshavebeenexplainedin ‘The reversible data hid- ing method: NON-R’ section. Experimental results have beengivenin ‘ Analysisoftheexperimentalresults’ section and final remarks have been denoted in the last section. Reversible data hiding methods A series of message bits can be carried in the original image provide that the original one is adhered. This can be called ‘hiding the message into the main image’. As shown in Fig. 1, reversible data hiding is described as recovery of the original image without any distortion in the obtained specific image after extracting hidden data 1 Computer Programming Department, Celal Bayar University, Manisa, Turkey 2 YY Image Processing Laboratory, Kocaeli, Turkey *Corresponding author, email kuslumurat@gmail.com ©2016TheRoyalPhotographicSociety Received21January2016;accepted20September2016 DOI10.1080/13682199.2016.1240945 TheImagingScienceJournal 2016 VOL 64 NO 8 471