Kubera Kolam: A Way for Random Image Steganography

INTRODUCTION

Communication being inevitable of daily routine has produced uprising right from the Stone period. Technology germinates for our own good, thus becomes better for our own best. Communicating technologies augments at each and every pace, say as of sheer contact to digital communication and to video tête-à-tête to even live discussion. This is all because of some brilliant brains. But this brilliant piece of work is illegally destroyed and killed by many computer professionals (as they call) but the hackers. These marvelous reforms in the technology are expanding, but not exponentially increasing for the same reason. However, this could be raised by only one means and that is information security (Cheddad et al., 2010; Salem et al., 2011; Schneier, 2007; Stefan and Fabin, 2000). Any form of life needs security so does the data. Myriad methods since the past decade are being discovered (Amirtharajan and Rayappan, 2012a-d; Amirtharajan et al., 2012; Bender et al., 1996; Cheddad et al., 2010; Janakiraman et al., 2012a, b; Rajagopalan et al., 2012; Thenmozhi et al., 2012) to save the data but in vain every method, hackers find better means to crack into the data (Schneier, 2007; Qin et al., 2010).

The need for a fulltime settlement of security arose and cryptography emerged as a powerful tool then, the algorithms created were almost the best, not de-cryptable and was providing the feature of protection of intermediate changing means (Salem et al., 2011; Schneier, 2007). Nevertheless, the only drawback to it was the image or content that was encrypted might had completely deformed, wherein one easily figures out that there is some manipulation done on that piece of data (Schneier, 2007; Qin et al., 2010; Zaidan et al., 2010). Now, this was not advisable for the reason that any hacker could poach into the data owing to the fact that something has changed.

This led to the rise of the super power technology of hiding data, Steganography (Al-Azawi and Fadhil, 2010; Al-Frajat et al., 2010; Xiang et al., 2011; Zanganeh and Ibrahim, 2011; Zhao and Luo, 2012; Zhu et al., 2011). Till date flawless, every feature incorporated proved to be the best and many more to be brought (Gutub, 2010; Luo et al., 2011; Mohammad et al., 2011; Padmaa et al., 2011; Thanikaiselvan et al., 2011; Zhao and Luo, 2012). The concept is of the usage of cover image to cover the secret data and create an illusion of nothing being in it. This piece of algorithm in the initial stage was done minimal number of randomizations in the image and data. But of late, several randomizations are made introduced in every fortnight, which for any hacker to actually know the piece of data after unveiling all the randomizations would at least a few years’ time, by which the task would have been accomplished.

This addition to the security since the past few years has again brought back the zeal for inventions of better communication technologies along with information security, starting from Pranav mistry’s sixth sense to thoughts of using 5G technology. Another classification in information security is watermarking (Abdulfetah et al., 2010; Zeki et al., 2011) its objective it to provide authorship through copyright protection. Reviewing the existing literature suggest to implement random image steganography with high capacity, good imperceptibility with additional complexity. Hence, this study proposes a method to accommodate all the necessary requirements of image steganography through kubera kolam.

PROPOSED METHODOLOGY

The indicator channel for the initial process is RED channel. For the embedding procedure, the algorithm is designed for a 6x6 block of an image, as in the image is sub-classified into blocks of 6x6 each and the algorithm is made to work on the each individual block of it. The bare part of Steganography, randomization is incorporated and inculcated with the magic square called Kubera Kolam.

The interesting part of Kubera Kolam is its wide usage in both puja room and may be in steganography algorithms. The magic square of Kubera Kolam is the sum of its any dimensional extension yields to 72 as shown in Fig. 1a. A subtraction of 19 from the Magic Square (Kubera Kolam) will interestingly lead to another magic square but with numbers from 1 extending to 9 and summing to 15 as shown in Fig. 1b.

Extending the 3x3 to 6x6 leads to the square as in Fig. 2.

Pixels of every 6x6 segment of the image is further rearranged and embedded in the same fashion with reference to the numbers formed in the above block, blinding the third party regarding the order of embedding.

Fig. 1(a-b):	(a) The magic square of Kubera Kolam and (b) New magic square with a subtraction of 19 from 1a

Fig. 2:	Extension of magic square from 3x3 to 6x6

Fig. 3:	Digital image of Pulli kolam

Fig. 4:	Embedding capacity offered by the pixels in Pulli kolam

The Kolam design shown below in Fig. 3 is the Pulli Kolam which acts the symmetric key for the embedding of the user’s message.

The number of curves or line on every dot is taken into account and a matrix is formed. For example, the top right dot has just one curve and the adjacent ones consist two each and so on.

Specifically, this dot in the Kolam gives the pixels for embedding as in Fig. 4 and gives the amount of bits for embedding in all pixels. After reading the sender’s message, Kubera Kolam matrix is now referred for the embedding order number and the equivalent position in the Pulli Kolam matrix is matched and the number in that position decides the number of bits to be read from the bit stream and embedded into appropriate place in the formed 6x6 block.

The further randomization is done with the conventional pixel indicator but with some slight modifications. Here, the 5th bit of the data channels (Blue and green) in that pixel is used to decide the reference bits in the indicator channel considered initially. As in, the 5th bits of both the channels turn out to be 0 and 0 then, last 2 bits of indicator channel will be acting as reference bits and if the bits are 0 and 1 then the 2nd and 3rd LSB’s are taken as reference bits and if its 1 and 0, the 3rd and 4th LSB’s are taken and finally if its 1 and 1 then 4th and 5th LSB’s are taken as reference. Now based on the reference bits the mode of embedding is decided i.e., 0 0 implies no embedding and 0 1 means embedding in blue channel and 1 0 in green channel and 1 1 in both. Further methods have also been introduced for the dynamicity of the encryption.

Method 2: Deals with indicator channel being decided by the sender at run time by accepting a value (1, 2 and 3).

Method 3: Proves considerably more efficient than the previous two as the indicator channel for that block is decided by the Mod 3 of the ixj value of the present block under embedding. The MOD result of 0 implies RED, 1 Green and 2 Blue planes.

The algorithm for this scheme is clarified below and the flowchart is given in Fig. 5.

Method 1:
Embedding algorithm

Fig. 5:	Flow chart for proposed method

Method 2:
Embedding algorithm

Method 3:
Embedding algorithm

ERROR METRICS

The quality of the stego image is calculated through two universal parameters viz., mean square error and the peak signal to noise ratio.

The MSE is calculated by the equation:

Where, M is pixels in horizontal dimension and N is pixels in the vertical dimension in X_i and Y_j where i, j constitutes pixels of original and stego images, respectively.

The peak signal to noise ratio is given by the equation:

where, I² max is every pixel’s intensity value. High PSNR values indicate high visual quality.

Fig. 6(a-d):	Cover images (a) Lena, (b) Baboon, (c) Mahatma Gandhi and (d) Temple

Fig. 7(a-d):	Stego images for 3 methods, (a) Lena, (b) Baboon, (c) Mahatma Gandhi and (d) Temple