A Modified Least Significant Bit Replacement: Steganography Method for Handling Random Noise Effect

Steganography is one of the security measures employed at file level to cover information in a carrier. The goal of steganography is to send a confidential message over a network in a way that intruders will not notice it. In steganography, the message is wrapped in an innocent carrier such as image, audio or video file. There are always bits tempering in the process and this changes the looks and appearance of the carrier. The more secret message is embedded in any cover file, the more random noise occurs. Secret message bits when embedded drastically modify the Red Green and Blue (RGB) components’ bits of the carrier, hence, the statistical properties and physical appearance of the carrier change and this draws the attention of intruders to suspect that confidential data is hidden there. This research introduced an improved method of least significant bit replacement to cover secret data using 24 bits bitmap image to handle random noise efficiently so that the Human Visual System (HVS) of intruders will not notice the secret communication. The research reviewed some previous related researches that aimed to reduce the effect of random noise and research gap is identified. The research implemented the new Modified Least Significant Bit Replacement (M-LSBR) method to fill the identified research gap. Five (5) experiments were conducted with 5 (five) varying dimension RGB images and the result gathered that the PSNR value, image quality and message capacity of the new method is better than those achieved by the previous methods, hence, the research gap is filled. Finally, conclusion was drawn and recommendation for future research was suggested.


Introduction
Unauthorized access to data is the common threats in a networked system that may cause loss of confidentiality, integrity, and availability of the information technology assets (Kaur, Inderjeet & Duhan, 2017). Steganography is one of the techniques used to safeguard data/information against unauthorized access. Steganography is a method of hiding confidential information in a cover file (Alam, 2016). The cover file may be text file, digital image, audio file or video file. The message wrapped cover, also known as stego is securely sent to a destination over the internet with low chance of intruder noticing it. Steganography is a step further to cryptography which only scrambles the information into incomprehensible form to unauthorized person (Laskar & Hemachandran, 2019). Steganography on the other hand hides the entire existence of the message and therefore it is considered more secure approach (Mortazavian, Jahangiri & Fatemizadeh, 2016). The problem that raised concern in image steganography method is the effect of random noise which is added in the cover during the hiding process and this cause the stego image to degrade in quality (Juneja & Sandhu, 2013). This research is aimed at achieving a detection free technique that efficiently reduces the effect of random noise in stego image. In order to achieve this aim, the conventional Least Significant Bit Replacement method is improved and a 24 bit bitmap image format used as cover. The research is intended to fill the research gaps identified in previous relevant literatures.
This work is published open access under the Creative Commons Attribution License 4.0, which permits free reuse, remix, redistribution and transformation provided due credit is given. Salluri (2020) proposed a method that slightly modified the technique of data embedding into Image Encryption using the Symmetric Key for RDH in Cloud Storage (Fig. 1). The research uses newer modules instead of older ones and provides a better Graphical User Interface for easy understanding and userfriendly experience. Ansari et al. (2020) presents an algorithm called GSA that works for cover images of multiple formats that is designed to apply uniform security policies across all image formats as shown in figure 2. This method can adaptively select the most suitable cover image based on data length, network bandwidth and allowable distortions on the abstract concept of image components that can be adapted for JPEG, Bitmap, TIFF and PNG cover images.

Reviewed Literatures
The Conventional Least Significant Bit (LSB) replacement method of steganography proposed by Champakamala, Padmini and Radhika (2009) is a common and simple approach to embed information in an image file. The method used 640 X 525 sized RGB image as cover to replace the Least Significant Bits (LSBs) of the cover image (C) with the bits of secret message (M) aimed at reducing noise in the stego-image. Every pixel in 24-bit bitmap image consists of Red, Green and Blue (RGB) colour components and each colour component consists of eight (8) bits. This method is easiest to implement but it is characterized by Low PSNR of 19.9987 dB and low quality image that draw intruder's attention to easily sense a secret message is hidden in the stego image. The method also accommodates only 15 bytes or less of message.  Korothan, Kishor and Butey (2016) proposed a technique that involves hiding secret in an RGB image. In this technique, embedding is done through replacing the noisy bit of the image with bits of the secret message. From Human Visual System angle, the stego image looks a bit distorted thereby drawing intruder's attention. The PSNR value achieved by this method is 41.131 dB and the method capacity depends on the amount of noisy bits found in the cover to be replaced with message. Emam, Ali and Omara (2016) proposed a random pixel replacement method to hide bits of the secret message. This method pays attention to hiding a message only in blue and green components of the cover image pixels, neglecting red colour component. This method achieved a high PSNR of 40.434 dB however; the appearance of the stego-image after embedding considerably changes. The method also has a good message hiding capacity as it can wrap up to 50 bytes of message length. Ali and Saad (2019) proposed Secret Message Matching (SMM) method if the embedded bit does not match the LSB of the cover image, then the pixel value of the corresponding pixel is randomly added by ±1. The original image and the stego image looks identical to the Human Visual System and this help reduces suspicion of secret communication. The method achieved 40.1132 dB PSNR but designed to hide only 20 bytes long message.
Abdul-Sada (2017) method is based on LSB-3. The paper exploits the third Least Significant Bit instead of first or second Least Significant Bit and it has achieved great capacity of hiding up to 45 bytes of message and 30.087 dB of PSNR. The original and stego looks a bit different as noise is slightly noticeable to the human eye. Manikandan et al (2021) came up with a combined framework of both encryption and steganography where only the authorized person can send the secret image to the receiver. In the first stage, an image steganography system is processed using confidential medical X-ray images which are hidden under the cover images. On the receiver side, the same process of decryption happens, followed by the two-level verification using email authentication and OTP generation using Pega. This ensures that only the authorized receiver can receive and view the secret image sent by the sender. Figure 3 has richly demonstrated this approach.

Identified Research Gap
Some of the visited literatures achieved low image quality after hiding message, low Peak Signal to Noise Ratio value or low message holding capacity. This research has come to fill these gaps found in these previous researches.

Materials
1. Five varying dimension 24 bit bitmap images 2. Matlab software 3. Microsoft Visual studio v.2010 4. ASCII converter Method Modified Least Significant Bit Replacement (M-LSBR) method takes the following stages: 1. Determine the size of the secret message with that of the cover image such that size of the secret message should be less than the size of cover image. 2. Convert the pixels of the 24-bits bitmap image (cover) into binary bytes using bit and divide it into RGB parts. Each byte represents one of the three components (R,G,B) 3. Encode the secret message into binary using ASCII converter.

Practical of M-LSBR
The method converts the message characters into ASCII binary representations. Table 1 indicates how a secret message 'HELLO' was encoded and hidden in sixteen (16) pixels of a 24 bit bitmap image.

Novelty of the Proposed Method
This research is an innovation of modifying the commonly known Least Significant Bit Replacement (LSBR) steganography method.
The method replaces the least bits of only one component out of the three components (Red, Green and Blue) of every pixel in a 24 bits bitmap image as shown in figure 4, achieving better image quality instead of following the conventional style (adopted by most of the previous methods) that replaces the least bits of all the three components in every pixel as shown in figure 5 which result in less image quality.

Results and Discussion
This research tries to compare the results for of some of the existing works and that of the proposed method (M-LBSR) in order to understand whether the proposed method filled the identified gap found in the previous works or otherwise.
The Conventional Least Significant Bit (LSB) replacement method used by Champakamala, Padmini and Radhika (2009) as seen in the literature review section is characterized with low PSNR value (9.9987 dB only) and low image quality image as shown in figure 6. The method could accommodate only 15 bytes or less of message.
In comparison, the method proposed by this research shows an image hiding a secret message of 52 bytes size in its first experiment with no difference in the appearance of the two images (before and after hiding) as depicted in figure 7 and achieved higher PSNR value (51.8283 dB) when analysed in a Matlab. When the 2 nd experiment was done with the proposed method, the image hides a message of 53 bytes and there is no difference in the appearance of both the cover image and the stego image as shown in figure 9 and the PSNR achieved was 53.0805 dB. The random pixel replacement method by Emam, Ali and Omara (2016) hides bits of secret message in blue and green components of the cover image pixels. The method achieved high PSNR of 40.434 dB however; the appearance of the stego-image after embedding considerably changes as shown in figure 10. The method also has good message hiding capacity as it can wrap up to 50 bytes of message length.
The 3 rd experimental result of the proposed method is compared with the Emam, Ali and Omara (2016) method discussed in the literature review section. The proposed method shows no difference in appearance of both original and after hiding image as shown in figure 11. The method is capable of holding 51 bytes message size and achieved 54.8254 dB PSNR value, hence it is more effective. The Secret Message Matching (SMM) method by Ali and Saad (2019) achieved 40.1132 dB PSNR and is limited to hide only 20 bytes long message with relatively higher image quality that one may not notice any difference between the original and stego image as shown in figure 12. The result of the 4 th experiment carried out by the method proposed by this research is compared with the Ali and Saad (2019) method which yielded almost the same outcome when it comes to image quality as shown in figure 13. However, the message capacity of the Modified Least Bit Replacement method of this research is higher than that of Ali and Saad (2019) method. Third Least Significant Bit Replacement method by Abdul-Sada (2017) achieved great capacity of hiding up to 45 bytes of message and 30.087 dB of PSNR. The original and stego looks a bit different as noise is noticeable (Fig. 14). The 5 th experiment carried out with the Modified Least Significant Bit Replacement proposed by this research indicates that image containing the message remain the same in appearance and the PSNR value achieved is 51.8436 dB (Fig. 15). The fifth experiment indicates how the cover image Figure 10(a) was used to embed a secret message of 53 bytes size. Figure 10(b) indicates the stego image containing the secret message but no difference in the appearance and statistical property of the both the cover image and the stego image. The Human Visual System (HVS) couldn't distinguish between the stego image and original image. The histograms of both original and modified image remain the same. Tables 2 and 3 are summaries of results from previous studies and proposed methods.