Abstract: In this study, we highlighted a new and efficient steganographic modus operandi based on pixel indicator routine to infix covert data in an RGB image. Usually when pixel indicator technique is employed on an RGB image, only one among the three planes (R, G and B) is labeled the indicant plane and the other two as depository planes in which the data can be stored. The proposed method is a similar concept by considering all the three planes for storage of the data instead of just two. This is carried out by using 5, 6 and 7 bits of a plane as indicator bits which increases the possible fields (from 4-8) to store the data. Because of this, the data embedding capacity of the image is improved to a greater extent since the indicator plane can also be used for embedding data in it. Optical Pixel Adjustment Process (OPAP) is also used here for reducing Mean Square Error. Furthermore the OPAP technique is not applied on the indicator plane as it modifies the indicator bits for reducing the MSE.
INTRODUCTION
Nobody really owns the internet. It is a global collection of networks for which everyone has access to. These days internet has reached even remote villages. But the million dollar question that arises is ‘How safe is this internet nowadays?’ To make internet more secure and safe, along with the growth of information technology and communication, there has been a tremendous growth in technologies to secure this information too. Information hiding is the best possible way to secure confidential information (Cheddad et al., 2010; Stefan and Fabin, 2000; Qin et al., 2010).
Many different methods are invented to encrypt and decrypt data to keep our data secret. Few among them are Cryptography (Salem et al., 2011; Schneier, 2007), Steganography (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; Thanikaiselvan et al., 2011; Thenmozhi et al., 2012), finger printing and water marking (Zeki et al., 2011). Cryptography is the art of scrambling of data in an unintended format so that no one other than the authorised receiver can decode it. It would look gibberish to any third person viewing it. But it has a disadvantage in that a person looking at it would find out that it is some encoded secret message (Zaidan et al., 2010). And if he gets hold of the secret code then any third person can extract it. Water marking is just for copy right protection and protection of intellectual property (Abdulfetah et al., 2010). The kind of data hidden in objects in the case of watermarking is a signature. This signature helps to signify the authority or ownership of the legal user. This study highlights about steganography and the algorithms used.
Steganography is derived from the Greek word ‘stegos’ meaning secret or something that is covered. ‘-graphy’ means art or drawing or writing, hence both put together means ‘a covered drawing’ (Al-Azawi and Fadhil, 2010; Luo et al., 2011; Mohammad et al., 2011; Zanganeh and Ibrahim, 2011; Zhao and Luo, 2012). Steganography is not new science. It has existed from the ancient times. In the olden days the secret messenger had his message encrypted in the form of tattoo and this was tattooed on his shaven head, thus, hiding the information from the third person. Only when the person’s head was shaven the image and the encrypted message could be decoded.
Ultimate aim of steganography is in the secure communication of the hidden data in a totally untraceable manner and to avoid any attention or suspicion to the transmission of the secret data. Apart from keeping others from knowing the hidden data, it should also prevent third persons from knowing that the secret data even exists. A simple classification is methods in spatial domain (Gutub, 2010; Padmaa et al., 2011) or transform domain (Amirtharajan and Rayappan, 2012d), but the cover object may be text (Xiang et al., 2011), video (Al-Frajat et al., 2010), audio (Zhu et al., 2011) or an image (Amirtharajan and Rayappan, 2012a-d; Cheddad et al., 2010). Aforementioned methods gives proper insight to steganography, in this study, a method is coined to improve the payload, imperceptibility with additional complexity in color image.
PROPOSED METHOD
The familiar method pixel indicator is proposed here by implementing new idea in that, by this way it improves embedding capacity as well as imperceptibility. It reduces the visual distortion by giving good image quality. In this method, number of bits embedded is defined by the user, say k-bit embedding. Indicator plane pixel bits tells that which plane is going to be a data plane. Two methods are introduced here; Red is taken as default indicator in method1. Method 2 uses the indicator plane cyclically. The block diagram of this study is shown in Fig. 1.
The flowcharts for embedding and extraction of the secret message are given in Fig. 2 and 3.
| Fig. 1: | Block diagram for the proposed method |
| Fig. 2: | Flow chart for embedding the secret data |
| Embedding algorithm |
| Fig. 3: | Flow chart for extracting the secret data |
| Extraction algorithm |
RESULTS AND DISCUSSION
Four images are taken as covers namely Lena, Baboon, Mahatma Gandhi and Temple of size 256x256x3. The algorithm is executed in MATLAB 7.1 with k = 1, 2, 3, 4 bit for each image and the results are given in Fig. 4-9. MSE and PSNR values for each iteration along with bits embedded in each pixel and total embedding capacity for method 1 and 2 is given in Table 1 and 2, respectively. The tentative results for method 1 say that it has produced substantially high PSNR values for all the images which is well above the minimum standard of 38 dB. It also conveys that the resultant stego images are of fairly high quality and cannot attract naked eyes’ attention. Of these covers, Lena holds the record of having high PSNR value of 59.0357 for k = 1 bit embedding. BPP is also passably decent. For each k bit embedding sensible amount of bits are entrenched showing that the algorithm works well with good capacity with increased complexity and security as well.
Method 2 results are given in Fig. 7-9, respectively which makes use of cyclic indicator method wherein each plane is termed indicator for subsequent iteration. Thus each plane gets a chance of being the indicator channel. Though one can witness high MSE value in all images, it produces sensibly genuine embedding capacity. Moreover, since OPAP is called the level of distortion is made under control. Stego images as well as the histograms prove this with which it can be concluded that the paper is detected to be good when equated against the subsisting ones. Unlike method 1, method 2 gives equalized grandness to every panorama of steganography.
Both the methods are probed against Chi-square run. The graphical record of Mahatma Gandhi image is shown Fig. 10. The original cover and all the four stego outputs (for k = 1, 2, 3, 4) are represented. It is evident from the graph that with the increase in number of rows the probability decreases and 2, 3, 4 bit embedding curves show almost the same results as that of the original. For 1 bit embedding the probability reduces to zero only after 100 rows in the image. Partially contrary to method 1, method 2 exhibits splendid end results. All the resultants go hand-in-hand with the cover, thus, on seeing the images one cannot even sense that they have some secret entrenched in them. After some good number of rows for all the four embedding processes the probability is zero and remains the same for the rest of the image. Thus, this routine boasts about the well built constructs and is undoubtedly full-bodied against Chi-square test.
Complexity analysis: Advanced Encryption Standard (AES) is adopted for encrypting the confidential information, it acquaints 2^128 intricacy. Of 3 planes, one act as indicator and the other two function as data channels.
| Fig. 4(a-d): | Cover Images for method 1, (a) Lena, (b) Baboon, (c) Gandhi and (d) Temple |
| Fig. 5(a-d): | Stego Images exhibiting maximum embedding capacity, (a) Lena (b) Baboon (c) Gandhi and (d) Temple |
| Fig. 6(a-e): | Sample Results for a single image in method 1 (a) Cover image Mahatma Gandhi. Stego images for ‘K’ bit embedding, (b) K = 1, (c) K = 2 (d) K = 3 and (e) K = 4 |
| Fig. 7(a-d): | Cover Images for method 2, (a) Lena, (b) Baboon, (c) Gandhi and (d) Temple |
| Fig. 8(a-d): | Stego images exhibiting maximum embedding capacity, (a) Lena, (b) Baboon, (c) Gandhi and (d) Temple |
| Fig. 9(a-e): | Sample Results for a single image in method 2 (a) Cover image Mahatma Gandhi. Stego images for ‘K’ bit embedding, (b) K = 1, (c) K = 2, (d) K = 3 and (e) K = 4 |
| Fig. 10(a-b): | Graphical results for checking (a) Method 1, (b) Method 2 against chi-square attack |
This is arranged in 3x2 ways. Of the total of 8 cases, there is no embedding done for 000. This makes the total cases as 7. As a result, the total embedding complexity is given by 2^128x3x2x(8/7)x(32+(64/7)+(32/7)+(128/35)+(32/7)+(64/7)+32+256).
| Table 1: | MSE, PSNR, BPP and embedding capacity for method 1 |
| Table 2: | MSE, PSNR, BPP and embedding capacity for method 2 |
CONCLUSION
The process of embedding secret data based on indicator-plane increases the embedding entropy considerably. OPAP decreases the Mean Square Error (MSE) thus making the stego image indistinguishable with the Cover. Thus, the proposed method which is an amalgam of the above mentioned methods, it incorporates reduction of delectability and increase of entropy at the same time. Imperceptibility, capacity is the major expectation in image steganography both is excellent in this study.