Reversible Watermarking Based on Invariant Image Classification and Dynamic Histogram Shifting

ABSTRACT:

In this paper, we propose a new reversible watermarking scheme. Onefirst contribution is a histogram shiftingmodulation which adaptively takes care of the local specificities ofthe image content. By applying it to the image prediction-errorsand by considering their immediate neighborhood, the scheme wepropose inserts data in texturedareas where other methods failto do so. Furthermore, our scheme makes use of a classificationprocess for identifying parts of the image that can be watermarkedwith the most suited reversible modulation. This classificationisbasedonareferenceimagederived from the image itself, aprediction of it, which has the property of being invariant tothe watermark insertion. In that way, the watermark embedderand extractor remain synchronized for message extraction andimage reconstruction. The experiments conducted so far, on somenatural images and on medical images from different modalities,show that for capacities smaller than 0.4 bpp, our method caninsert more data with lower distortion than any existing schemes.For the same capacity, we achieve a peak signal-to-noise ratio(PSNR) of about 1–2 dB greater than with the scheme of Hwanget al.,themostefficient approach actually.

EXISTING SYSTEM:

The introduction of the concept of reversible watermarking in the Barton patent, several methods have beenproposed. Among these solutions, most recent schemes use Expansion Embedding (EE) modulation, Histogram Shifting(HS) modulation or, more recently, their combination. Oneof the main concerns with these modulations is to avoid underflows and overflows. Indeed, with the addition of a watermarksignal to the image, caution must be taken to avoid gray levelvalue underflows (negative) and overflows (greater thanfor a bitdepth image) in the watermarked image while minimizing at the same time image distortion. Basically, EE modulation introduced by Thodiet al. is a generalization ofDifference Expansion modulation proposed by Tianetal.which expands the difference between two adjacent pixels byshifting to the left its binary representation, thus creating a newvirtual least significant bit (LSB) that can be used for data insertion. Since then, EE has been applied in some transformeddomains such as the wavelet domainor to prediction-errors.EE is usually associated with LSB substitution applied to“samples” that cannot be expanded due to the signal dynamiclimits or in order to preserve the image quality. In, Ni etal.introduced the well-known Histogram Shifting (HS) modulation. HS adds gray values to some pixels in order to shift arange of classes of the image histogram and to create a ’gap’near the histogram maxima. Pixels which belong to the classof the histogram maxima (“Carrier-class”) are then shifted tothe gap or kept unchanged to encode one bit of the message ’0’or ’1’. Other pixels (the “noncarriers”) are simply shifted. Instead of working in the spatial domain, several schemes applyHS to some transformed coefficientsor pixel prediction-errors, histograms of which are most of the time concentrated around one single class maxima located on zero. Thismaximizes HS capacityand also simplifies the reidentification of the histogram classes of maximum cardinalityat the extraction stage. In order to reduce the distortion whilepreserving the capacity, some preprocessing has been suggestedin order to identify pixels, transformed coefficients or prediction-errors that do not belong to the histogram maxima classes(“noncarrier classes”). As we will see later, different schemesworking with prediction-errors do not watermark pixels withina neighborhood of high variance; indeed, these pixelsbelong to histogram classes that are shifted without messageembedding. Recently, Hwanget al.improved the approachof Sachnevet al.. They suggest defining the set of carrier-classesas the classes which minimize, for a given capacity, image distortion. However, their set of carrier-classes is uniquely definedfor the whole image and the execution time of this approach israther high.

DISADVANTAGES OF EXISTING SYSTEM:

  • The modulation is overflow and underflow.
  • The image distortions are more.

PROPOSED SYSTEM:

We propose to adapt dynamically the carrier-classes by considering the local specificitiesof the image. We simply suggest using the local neighborhoodof each prediction-error in order to determine the most adaptedcarrier-class for message insertion.Another refinement we propose is based on the selectionof the most locally adapted lossless modulation. Indeed, reversible modulations are more or less efficient depending onimage content. This is especially the case for medical imageswhere large black areas exist (i.e., the background area). Inthese regions, directly applying HS on pixels may be moreefficient and of smaller complexity than applying it on prediction-errors. Because, the histogram maxima corresponds to thenull gray value; capacityis maximized and underflows simplyavoided by shifting pixel value to the right, i.e. by adding apositive gray value. When working on prediction-errors inthese regions, the management of overflows/underflows ismore difficult because the shift amplitude can be positive ornegative. This is why we suggestconsidering the local contentof the image in order to select the most locally adapted losslessmodulation. This should allow us to optimize the compromisecapacity/image distortion.

ADVANTAGES OF PROPOSED SYSTEM:

  • Directly applying HS on pixels may be more efficient and of smaller complexity than applying it on prediction-errors.
  • The watermark embedder and extractor remain synchronized because the extractor will retrieve the same reference image. Herein, we adapt this process to select the most locally appropriate watermarking modulation.

SYSTEM CONFIGURATION:-

HARDWARE REQUIREMENTS:-

Processor-Pentium –IV

Speed- 1.1 Ghz

RAM- 256 MB(min)

Hard Disk- 20 GB

Key Board- Standard Windows Keyboard

Mouse- Two or Three Button Mouse

Monitor- SVGA

SOFTWARE REQUIREMENTS:

•Operating system : - Windows XP.

•Coding Language: C#.Net

REFERENCE:

GouenouCoatrieux, Member, IEEE, Wei Pan, Nora Cuppens-Boulahia, Member, IEEE, FrédéricCuppens, Member, IEEE, and Christian Roux, Fellow, IEEE “Reversible Watermarking Based on Invariant Image Classification and Dynamic Histogram Shifting” -IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY, VOL. 8, NO. 1, JANUARY 2013.