INTRODUCTION

The radical change in information handling has been enhanced by the kind of information security methodologies (Al-Azawi and Fadhil, 2010; Al-Frajat et al., 2010; Amirtharajan and Balaguru, 2009; Amirtharajan and Rayappan, 2012a-d; Amirtharajan et al., 2011) which have been developed over the years. The present internet world works with enormous payload but protection of the payload till it reaches out the concerned is a big challenge (Cheddad et al., 2010; Amirtharajan et al., 2012). We have come across security approaches broadly classified into three. Cryptography (Schneier, 2007), a technique which scrambles the payload with the help of a key to make the payload or data, known to be cipher text. Public Key, Private Key, Block Ciphers, Stream Ciphers and Encryption standards proposed by Forensic agencies are a part of cryptography family. On the other hand, Information copy right protection is achieved through watermarking (Karzenbeisser and Perircolas, 2000; Zeki et al., 2011). Robust and Fragile are the predominantly used categories of watermarking.

Steganography, the third approach is a methodology to hide the secret message in a carrier which is also called cover (Cheddad et al., 2010; Hmood et al., 2010a, b; Amirtharajan and Balaguru, 2009; Amirtharajan and Rayappan, 2012a-d; Amirtharajan et al., 2011). The cover for carrying or hiding the information may be an image, audio file (Zhu et al., 2011) or a video file (Al-Frajat et al., 2010). Steganography on image has been reported in various earlier works (Janakiraman et al., 2012a, b; Thanikaiselvan et al., 2011a, b). Various parameters have been considered in the past such as methods to decide the amount of information to be hidden (ie) capacity, Increasing the robustness of hiding process, Encryption hiding on image, Image processing based hiding techniques, Adaptive embedding, Integrity check mechanism added stegano approach (Amirtharajan et al., 2011), Random pixel embedding, Block based emedding etc., Also FPGA based steganography approaches (Rajagopalan et al., 2012a, b) have been reported in literature.

Simple LSB substitution with optimal pixel adjustment process, fondly called as OPAP has been proposed by Chan and Cheng (2004). Pixel value differencing is proposed by Wu and Tsai (2003) where number of bits embedded is decided by the proposed algorithm. Hiding information on RGB images have been presented by Amirtharajan and Balaguru (2009) and Amirtharajan et al. (2011) and many other works (Padmaa et al., 2011; Mohammad et al., 2011; Thenmozhi et al., 2012; Qi et al., 2010; Zaidan et al., 2010; Zanganeh and Ibrahim, 2011).

We propose a grayscale image steganographic algorithm which uses nibble differencing technique where minimum absolute difference between the Upper nibble of the pixels of a selected 2x2 message embedded image block guides the block bits rotation angle to make the final stego image.

PROPOSED METHODOLOGY

Let I be the grayscale image with A_x and B_y as its row and column dimensions, where 1≤x≤N and 1≤y≤N. Here, I_xy is an element of the image I.

When a message m_j from the {M} has to be embedded into a pixel I_xy, the pixel undergoes the following change:

I’_xy = I_xy-I_xy mod 2^k + m_j

where, k is the No. of bits to be embedded in a pixel I_xy. Grayscale Image of size NxN I is divided into blocks such that:

Ib_mn ∈ I | Ib⊂I

where 1≤m≤2 and 1≤n≤2. Figure 1 shows the Ib subset.

Now Ib is a subset of I (ie) a 2x2 pixel group which belongs to the grayscale image I. Let us assume the Ib with the following elements:

where, i and j are row and column respectively. After embedding encrypted k-bits in the Ib block, the block becomes:

Let this embedded block be an intermediate block Intb.

Fig. 1:	Subblock Ib from the Image I

Calculating the absolute differences between Upper nibble of the Intb block pixels:

where, {D1,D2,D3,D4,D5,D6} ∈ {D}. Finding minimum ({D}) which is the minimum difference between a pair of pixels of the block and {el(minimum({D})}, which is the pair of nibbles producing minimum difference to decide the number of times the embedded k-bits of all the pixels of the block has to be rotated. By concatenating upper nibbles of {el} by doing the following operation we get:

P’ = ((el₁ ∈{el}) & 240) OR (el₂ ∈{el}) >> 4

where, P’ gives a number of times the lower nibbles of the block has to be rotated. The 6th bit b₆ of P’ decides the direction in which the rotation should happen:

P’ can also be expressed as:

P’ = (((el₁ ∈{el}) & 240) OR (el₂ ∈{el})>> 4) mod 4

Because if block is rotated 4 times, the resultant block will be equal to the Intermediate block and also one time rotation of the block clockwise or anticlockwise results in change in orientation of the block by 90°. This holds good for even 1 bit rotation among the pixels of the block.

Therefore P’ times rotation = (P’ mod 4) times rotation. Figure 2 shows the orientation of the Intermediate block Intb after various rotation angles.

Fig. 2(a-d):	(a) Intermediate block, (b) 90° clockwise lower nibble rotation (c) 180° clockwise lower nibble rotation and (d) 270° clockwise lower nibble rotation

Here, Ib”_i-1,j-1 = (Ib’_i,j-1 & 240) OR (Ib’_i-1,j-1 & 15), if k = 4. Similarly other elements of Ib” can be computed for different k values.

After rotation, the procedure has to be repeated for other Ib blocks by selecting the blocks in a random manner or by a specific order defined by the user.

Let us consider an example. Let Ib be:

A total number of 16 bits have to be embedded in this block. The encrypted message to be embedded is 0000111100001111. It is evident that each pixel should carry 4 bits of secret information making k = 4. After embedding the secret message the Ib becomes:

When finding the {D}, the elements of {D} are D1 = 0, D2 = 1, D3 = 1, D4 = 1, D 5 = 1 and D6 = 0. As the minimum of {D} are D1 and D6 and considering the difference pair based on the order of precedence which is D1, {el} elements are 144 and 159. Concatenating the upper nibbles of 144 and 159 we get P’ as 10011001₂ which is equivalent to 153 in decimal. As b6 of P’ is 0, 153 times Anticlockwise rotation or P’ = 153 mod 4 (ie) 90° anticlockwise rotation (or) one time block rotation by left. Now the resulting block will be:

The decryption of message from the stego image can be done as per the following pseudo code:

For I = 1: Total blocks