INTRODUCTION

Power has taken several incarnations ever since its genesis. Right from Stone Age to the present day, power has kept its possessor at the summit of hegemony and thus it is the most sought after commodity. In the electronic epoch power has manifested itself in the form of classified and critical information. Since the human race has succumbed to enticing power to such an extent that iniquity today is skyrocketing, there is a need to protect information from falling into the wrong hands and to prevent clandestine and unscrupulous activities. Providentially, the advancements in technology have begot many techniques to maintain the veracity and variability of the crucial information giving rise to an entire discipline called information hiding. Information hiding is stratified into several subsets namely cryptography (Schneier, 2007), steganography and watermarking (Stefan and Fabin, 2000; Zaidan et al., 2010).

Cryptography (Schneier, 2007) is the art of writing esoteric information in an occult fashion thereby rendering it scrutable only to the authorized receiver. In contrast to cryptography which focuses on keeping the contents of a message secret, steganography (Stefan and Fabin, 2000; Zaidan et al., 2010) focuses on keeping the very existence of a message secret. Steganography is implemented in digital audio (Zhu et al., 2011), video (Al-Frajat et al., 2010) and images (Amirtharajan and Balaguru, 2009, 2010, 2011; Amirtharajan et al., 2012; Bender et al., 1996) of which image steganography has gained much appreciation and commendation in the recent past. In image steganography the vital information is dissembled in a cover image with assiduous efforts resulting in a stego image. The embedded secret information is imperceptible to the human eye thereby rendering the image impregnable (Yang, 2008).

In the available literature many researchers proposed an assortment of approaches to information hiding. These methodologies have different characteristics like capacity, imperceptibility and robustness (Amirtharajan and Balaguru, 2009, 2010, 2011; Kumar et al., 2011). These characteristic are inevitable for different applications, such as secret communication (Stefan and Fabin, 2000), copyright protection (Wang and Lin, 2004; Yen and Tsai, 2008) and tampering detection or integrity check (Lin et al., 2005).

Information hiding techniques could be categorized into two types: methods in the spatial domain and methods in the frequency domain. In the spatial domain approach, the secret messages are embedded by directly injecting secret data in the image pixels (Chan and Cheng, 2001, 2004; Wang et al., 2001; Chang et al., 2003; Yang, 2008; Thien and Lin, 2003). Whereas in the later case, the frequency domain approach the image is first transformed into its frequency domain (Amirtharajan and Rayappan, 2012a,b; Chang et al., 2002) then the secret messages are embedded in the transformed coefficients.

The major concern is about the objective of transmitting secret data, the stego method should possess high capacity, high quality and imperceptibility. More number of research papers have been intended for this theme and performs the embedding operations in the spatial domain either using raster scan or random scan (Amirtharajan and Balaguru, 2009, 2010; Amirtharajan et al., 2011, 2012; Yen and Lin, 2010). A detailed survey on Information hiding till 1999 is available by Petitcolas et al. (1999). A complete survey on image steganography could be found by Cheddad et al. (2010) and on random image steganography and steganalysis in Amirtharajan et al. (2012) three more survey on Field Programmable Gate Array (FPGA) for steganography, middle ware for cryptography/ steganography and Orthogonal Frequency Division Multiplexing (OFDM)+Code Division Multiple Access (CDMA)+stego for secure communication is available by Rajagopalan et al. (2012), Janakiraman et al. (2012a) and Thenmozhi et al. (2012), respectively. There are three kinds of approaches called LSB-based (Chan and Cheng, 2001, 2004; Wang et al., 2001; Chang et al., 2003; Yang, 2008; Thien and Lin, 2003; Amirtharajan and Balaguru, 2009, 2010), PVD-based (Wang et al., 2008; Amirtharajan et al., 2010) and mod-based (Chan and Cheng, 2004; Thien and Lin, 2003; Wang et al., 2008) are commonly available in literature and sometimes it could be combined to offer both capacity, imperceptibility and to improve the security (Chang et al., 2003; Hmood et al., 2010a, b; Xiang et al., 2011; Lin et al., 2005; Janakiraman et al., 2012b; Zaidan et al., 2010, 2011 and Zanganeh and Ibrahim, 2011). The counter attack on steganography called steganalysis are detailed (Xia et al., 2009; Qin et al., 2009). A detailed review on steganalysis is reported by Qin et al. (2010).

In LSB-based approaches, secret data are embedded by directly substituting the least-significant-bits (LSBs) with equal bits of the secret for each pixel. Furthermore, techniques based on pixel-value differencing (PVD) modify the difference value between a pair of pixels to fit the value of the embedded secret. Finally, mod based approaches which use the modular operation, are similar to k-bit LSB-based approaches if the modulus is 2^k.

Motivated by this study, a simple and effective stego method has been proposed to improve the stego image quality and to introduce cryptic effect while embedding.

PRELIMINARY RELATED WORKS

Chan and Cheng (2004) proposed an LSB-based hiding scheme using an optimal pixel adjustment process (OPAP). Their method adjusts each pixel after the message is embedded to improve the quality of the stego object and their experimental results showed that their method yielded quicker results. Yang (2008) proposed new LSB-based approach, named as the Inverted pattern (IP) LSB substitution approach. Later this method combined with OPAP called IPLSB to improve the quality of the stego image. In this study, we have adapted a new LSB-based approach based on Yang (2008), named as the inverted pattern binary and gray (IPBG) LSB substitution approach, to further highlight the quality of the stego-image Before secret messages are embedded, some secret messages are transformed by inverting operation and some secret messages are not. A simple strategy is used to judge whether a section of messages is inverted and a bit string named as the IPKey is used to record these inverting actions. Also, we combine the concept of the OPAP with our approach to improve image quality further. The experimental results show that the proposed approach results in a better image quality than that of the optimal LSB substitution approach (Wang et al., 2001; Chan and Cheng, 2001; Thien and Lin, 2003), the OPAP LSB substitution approach (Chan and Cheng, 2004) and inverted pattern approach (Yang, 2008).

In a normal LSB substitution the RGB (red blue green) image is converted in to gray image and then last few least significant bits of gray image are selected according to key length k and the message which is to be embedded is converted to series of ASCII values of the characters in the message and then to binary. Message is then stored in the cover according to the method of embedding. The series of operations done in LSB substitution are as follows:

Let C be the original 8-bit grayscale cover-image of M_cxN_c pixels represented as:

D be the n-bit secret data represented as:

Suppose that the n-bit secret data D_d (decimal representation) is to be embedded into the k-rightmost LSBs of the cover-image C:

Here, S is the Stego object, C cover object and D_d is the decimal equivalent of the secret data.

In the extraction process, given the stego-image S, the embedded messages can be readily extracted without referring to the original cover-image. The k LSBs of the selected pixels are extracted and lined up to reconstruct the secret message bits. Mathematically, the embedded message bits D can be recovered by:

The OPAP simply improves the stego object after embedding the secret data, either by adding or subtracting 2^k without affecting the rightmost k secret data bits in the stego cover.

THE PROPOSED METHOD

Embedding: A schematic diagram of the proposed method is given in Fig. 1 and 2. Initially the secret data or message is encrypted using Data Encryption Standard DES (Schneier, 2007), is a symmetric key cryptography algorithm. The cover image is split into separate rows. The order of rows considered for embedding data is chosen using a Pseudo random number generator with a chosen seed. For each row, a try is made to embed data, inverse data, gray code of data and the inverted gray code of data. The encoded form of the confidential information on the selected row which offers minimum Mean Square Error (MSE) is chosen and fixed for the same. This binary/ inverted binary/gray and inverted gray data pattern is stored as IPKEY. Thus, on an average MSE is reduced to a greater extent. The Stego image, IPKEY and the seed are communicated.

Mathematical model for row wise, inverted pattern LSB embedding

(1)

Where:
k	=	No. of bits
x	=	Number to be inverted in bits
	=	1’s complement of the number

For example, take a 4 bit binary representation of a number ‘2’ [0010] as x here:

k	=	4 so, -2 = (16-1)-2
	=	13[1101] complement of [0010] ‘2’

(2)

For example, let k = 4. C_i = 16[0001000] and m_i is ‘2’ [0010] as m_i:

Fig. 1:	Proposed schematic diagram for embedding

Fig. 2:	Block diagram for extraction

(3)

where, symbols are same as Eq. 2:

Let S_i = 18[00010010]
To Extract the last 4[since k = 4] bits, we have:

•	Plain data-m (i,j)
•	Inverted data - (i, j)
•	Grey Coded data-g (i,j)
•	Inverted Grey Coded data-g' (i,j)

(4)

where, the M_cxN_c are dimensions of the cover image. Each row is denoted as ri, where, i∈N and I≤R i.e., Set of rows = {ri, ∀ i∈N and i≤R}

(5)

In other words, each row has D pixels.

The complement of m (i, j) is denoted as (i, j)

Let it be divided into R blocks named r_i, r_i,_……., r_iR, each having equal number of pixels D:

(6)

Also, r_i = [r_i1, r_i2,_……., r_iD], where i ∈ N and i ≤ R

Let ‘k’ be the number of LSBs to be replaced in cover pixels.

Let the secret message be a matrix M, where each elements of M is made up of k bits. Then we can denote the message to be embedded in the i^th row, j^th pixel as m (i, j). Let s (i, j) denote the stego value of j^th pixel in the i^th row, when message m (i, j) is embedded in cover pixel r_ij. Alternatively (i, j) is embedded instead of m (i, j) then the stego pixel is denoted as (i, j):

(7)

If we consider R blocks of stego image as s₁,s₂,…………..s_R. Then, s_i = s (i) or (i) or s_g (i) or s_g' (i) where MSE is minimum and s (i) = {s (i, j), j∈N and j≤D}

key matrix is denoted as:

where, K_i is chosen based on the following conditions 00-if m (i,j) is embedded:

01	=	if (i, j) is embedded
10	=	if g (i,j) is embedded
11	=	if g' (i,j) is embedded

The preliminary, unprocessed message m_u (i,j) can be extracted from pixels in stego image as: m_u (i,j) = s (i,j) mod 2^k from Eq. (3) the actual message m (i,j) can be extracted by processing m_u (i,j) as follows:

WORST CASE MSE

The worst case MSE for a block with D pixels is defined as:

(8)

MSE for i^th row, when m (i) (actual data) is embedded, is given as:

When inverted data (i) is embedded, then, MSE for the same parameters is denoted as:

When grey coded data g (i) is embedded, then MSE for the same parameters is denoted as:

When inverse grey coded data g’ (i) is embedded, then MSE for the same parameters is denoted as:

According to the embedding procedure, minimum MSE is chosen. The minimum MSE for a row is defined as:

= D^-1 (2^k – 1)² ( since sum of all s,, s_g and components:

(9)

We know that, if any n numbers x₁, x_2, x_3, x_4,…, x_n add up to produce a total T, then:

(10)

Thus, applying (10) in (9), we get:
MSE_min (i)≤(1/2) MSE_w (i) for all I≤R

Thus, we get MSE for any block to be less than or equal to ½ of the worst case MSE.

Random k-bit Adaptive Embedding

Inputs:

Outputs:

Algorithm for embedding:

•	Step 1: Encrypt the secret data (M) using DES (Data Encryption Standard) with key E
•	Step 2: Let P = length of secret data stream M (in number of bits) got from Step-1
•	Step 3: Split the cover image C into separate rows . Let N = Total number of rows
•	Step 4: Generate a array PRN of N pseudo-random numbers in the range [0,N-1] where each No. occurs only once. Let the seed be stored in a text file
•	Step 5: Invert the bit array M to give . Encode the bit array M using Grey Code to give G and invert G to give
•	Step 6: Let i = 1 (Here, i is the row counter)
•	Step 7: Select PRNG[i]th block and perform the following operations

Selective embedding
•	{a. Let r = pixel index array (for traversal)
•	b. For ( j = 1 to length (r) ) do (Here j is the pixel counter)
	{
Replace k LSBs of jth pixel of the selected block with k bits from M to give O[i,1]
	}
•	c. Compute MSE
•	d. For ( j = 1 to length (r) ) do (Here j is the pixel counter)
	{
Replace k LSBs of jth pixel of the selected block with k bits from to give O[i,2]
	}
•	e. Compute
•	f. For ( j = 1 to length (r) ) do (Here j is the pixel counter)
	{
Replace k LSBs of jth pixel of the selected block with k bits from G to give O[i,3]
	}
•	g. Compute MSEGray
•	h. For ( j = 1 to length (r) ) do (Here j is the pixel counter)
	{
Replace k LSBs of jth pixel of the selected block with k bits from to give O[i,4]
	}
•	g. Compute MSE gray
•	h. If MSE is greatest

KEY[i] = ”00”
Else if is greatest
KEY[i] = ”01”
Else if MSEGray is greatest
KEY[i] = ”10”
Else
KEY[i] = ”11”
Assign MSE[i] = Minimum MSE
}

Recovery process: The same Pseudo random number sequence is generated using the received seed. Using the KEY, the pattern is identified for different rows. Recovery modules are run to recover the secret. The result is then decrypted using DES to get the message back.

Random k-bit adaptive recovery

Algorithm for extraction

•	Get Message M using retrieval
•	B. If KEY[i,1] = ”01”
M[i] = [I]
Else if KEY[i,1] = ”10”
M[i] = MGray[i]
Else if KEY[i,1] = ”11”
M[i] =
Else
M[i] = M[i]