AN IMPROVED IMAGE STEGANOGRAPHY BASED ON LEAST SIGNIFICANT BIT MATCHING REVISITED (LSBMR) USING SOBEL EDGE DETECTION

By

Author

Presented To

Department of Computer Science

ABSTRACT

Image steganography is the science of hiding data for securing confidential communication and it is the most popular type of carrier to hold information. Many algorithms have been proposed to hide information into digital images. The least significant bit algorithm (LSB) as one of the algorithms proposed is widely used in message embedding. However, the robustness of the algorithm based on LSB is low. The hidden message is usually destroyed when some image operations like resizing, cropping and rotation are applied to the stego-image. To overcome this limitation, this work proposed an improved image steganography based on least significant bit matching revisited (LSBMR) using Sobel edge detection that withstands image operations like resizing, rotation and cropping. The proposed method employs 2-dimensional discrete cosine transformation (2D-DCT) to transform the detected edges of the cover image pixel value into its coefficient, embeds the secret message in the coefficients of the detected edges of the cover image which was implemented in Netbeans IDE. Experimental results produced better stego-image quality that is robust against multiple image operations such as resizing and cropping. The statistical steganalysis tools such as Virtual Steganographic Laboratory (VSL) and StegExpose cannot detect the presence of secret information in the stego-image. Also, the proposed system generated stego-image with Peak Signal to Noise Ratio (PSNR) that is an image quality metric of 68 decibels (dB) for 8000 bits of secret message as regards to the invisibility over the existing steganography technique.

INTRODUCTION

This chapter discusses the introductory part of this dissertation, which includes the background to the study, problem statements and research motivation, the research aim and objectives, research methodology, contribution to knowledge and finally the organization of the rest of the dissertation.

1.1 Background to the Study

Recently, people exchange information using the existing communication technologies such as the internet and huge volume of data transfer takes place via the plethora of services offered by the web. This information can be very sensitive and need to be protected against any attacker who tries to intercept them during the transmission stage. According to Ratnakirti et al., (2013), data over internet may be stolen, intercepted, illegally modified or even destroyed by an adversary resulting in intellectual property rights infringement, data loss, data leakage and data damage. Transmitting top secret information cannot be solely relied on the existing communication channels because the technologies are vulnerable to attacks Osama, (2005) and exchanged information can be detected relatively easily. Therefore, it is vital to protect the privacy and confidentiality of top secret message during its transit through the internet. To preserve the privacy and confidentiality of important data over the internet it must be provided with a metaphorical envelope such that its contents are revealed only to the intended receiver Ratnakirti et al., (2013) without arousing suspicions. Data hiding techniques such as steganography precisely aim at performing this task.

The steganography technique has been used many years ago to convey secret messages. For instance, Greek historian Herodotus was the first to document the usage of steganography to send messages (Aubrey, 1996). A slave was sent by his master to deliver a secret message tattooed on his scalp. After the message was tattooed, the slave waited until his hair grew back and concealed the message. The most popular steganographic methods between the 13th and 16th century involved written text. One method used a mask, a paper with holes, shared between the sender and recipient. The mask was simply put over the text and the message was revealed. Francis Bacon realized that two different fonts for each letter can be applied to embed binary representations of messages. Holub, (2014), stated that ―Brewster devised a very original technique in 1857, which was later used in several wars‖.