IJSRSET162643 | Received: 18 Nov-2016 | Accepted: 24 Nov-2016 | November-December-2016 [(2) 6: 149-154]
© 2016 IJSRSET | Volume 2 | Issue 6 | Print ISSN: 2395-1990 | Online ISSN : 2394-4099
Themed Section: Engineering and Technology
149
Enhancing Robustness of Embedded Medical Images with a 4
level Contourlet Transform
J. Samuel Manoharan, G. Jayaseelan, P. Muralidharan
Bharathiyar College of Engineering and Technology, Karaikal, UT, Thiruvettakudy, Puducherry, Tamil Nadu, India
ABSTRACT
Medical Images are an important class in the sense that they deal with real time conditions and are highly sensitive.
Any slightest modification or manipulations done during their processing could degrade the quality of the image
which could severely affect the efficiency of embedding techniques. Hence, Robustness and Fidelity are two
important criteria that need to be adhered to in any data embedding technique. A multi resolution approximation
technique using the Contourlet transform has been introduced in this paper and its effectiveness towards a wide
range of aggressive image processing operations simulating the real time attacks has been found to make it a
suitable transform to embed data into sensitiveness prone medical images. The results have been expressed in terms
of well-known metrics like PSNR, Correlation Coefficient and the structural similarity index.
Keywords : Data Embedding, Multi resolution approximation, Robustness, Attacks, Image Quality.
I. INTRODUCTION
Data Hiding is an age-old technique, which deals with
concealing a data, which may be a audio, video or
image, known as the watermark or payload, inside
another data which may be a audio or video or image
known as the Cover data. It might find various real time
applications such as copyright protection, broadcast
monitoring, tamper detection and covert communication
[1]. Data embedding could be done in the spatial
domain with pixel wise modifications or frequency
domain using coefficient wise modifications with each
having their own merits and demerits. Normally, a
frequency domain technique is preferred if preserving of
image quality is of concern. Three criteria govern the
success of any embedding technique namely, robustness,
perceptual imperceptibility and embedding capacity.
Robustness denotes the capability of the embedded data
to withstand intentional and unintentional attacks while
perceptual invisibility denotes the degree of invisibility,
of any modification made to the cover data, to the
human eye. Embedding capacity measures the amount
of payload that could be packed into the cover data
without causing any degradation to the embedded data.
All three mutually exist and are interrelated and hence a
proper balance is highly essential to preserve the
efficiency of the embedding technique. A general
embedding and extraction scheme is shown in figure 1.
A recent trend is the utilization of data embedding
techniques for storing medical data which may be a
diagnosis report of the patient into a medical image
belonging to the patient itself [2] to serve the purpose of
tele based medical services, labeling and archival, data
base storage for later retrieval and hospital patient
record management systems.
Figure 1: A General data embedding and extraction
scheme