Performance Analysis of LSB, MSB, and Combined LSB – MSB Algorithm in Terms of Image Quality and Encoding Time
Chapter One
AIMS AND OBJECTIVES OF THE STUDY
The purpose of the Performance analysis of LSB, MSB and the combined LSB – MSB algorithm is to:
- Analyse the performance of the aforementioned algorithms on image and image production and compression.
- Discover optimal algorithms to use when performing compression and encoding functions.
- Improve constantly each algorithm and model.
- Combine the MSB and LSB techniques into an hybrid algorithm that embeds secret message bits into the least significant bit and most significant bit of the cover image.
- Compare the three algorithms in terms of encoding time
- Test the algorithms using different image formats and observe the quality of image.
CHAPTER TWO
REVIEW OF LITERATURE
This chapter examines in detail, the history and developments made in the data compression and encoding fields, previous research work on this subject, the characteristics, models, architectures and limitations as pointed out by various scholars and researchers. This will provide the ground work for figuring out an efficient way to compress, encode and decode multimedia forms of data.
HISTORY OF DATA COMPRESSION
Data compression in the modern electronic era began in the 1940s with Clause Shannon and Robert Fano in 1959 designing a systematic way to assign code words based on probabilities of blocks. An optimal method was then created by David Huffman in 1951. The coding in Huffman’s design was top-down compared to the Shannon and Fano bottom-up design. Abraham Lempel and Jacob Ziv in 1971 introduced the LZ77 algorithm and this algorithm was the first to use a dictionary to compress data. It utilized a dynamic dictionary called a “sliding window”. Lempel and Ziv improved on their previous algorithm in 1978 with the LZ78 algorithm and retained the dictionary feature but this time it made use of a static dictionary.
The 1980s came with the Lempel-Ziv-Welch(ZLW) algorithm whereby Lempel and Ziv were joined by Terry Welch who made changes to their earlier LZ78 algorithm. This algorithm became the popular algorithm for data compression at the time. It is still used in modems today.
By the late 1980s, digital images became so popular that image compression standards started evolving. The TIFF file format was published in 1986 and is still used for high colour depth images. In 1987, GIF file format was introduced.
In 1992, the JPEG format was created by the Joint Photographic Experts Group. The degree of compression can be adjusted allowing a trade off between storage size and image quality.
The PNG file format was created in 1996.
RESEARCH WORK ON STEGANOGRAPHY
There has been several researches in hiding data inside an image using steganography technique. In Warkentin et al(2008) proposed algorithm, the idea was to hide data inside the audiovisual files. El-Emam’s(2007) proposed steganography algorithm is based on hiding a large amount of data file inside a coloured bitmap image. In his work, he filtered and segmented the image by using bits replacement on the appropriate pixels. A concept defined by main cases with their sub cases for each byte in one pixel was used to select these pixels randomly rather than sequentially. This concept was both visual and statistical. The result of this concept was that 16 main cases with their sub cases covered all aspects of the input data into color bitmap image. Three layers provided high security which made it difficult to break through the encryption of the input data and also undetectable when steganalysis is applied. It was concluded that a large amount of data that occupies 75% of the image size can be embedded efficiently and the output will be of high
quality.Chen et al.(2009) modified a method proposed by Chang et al.(2004) using the side match method. In this method, data was hidden in the edge portions of the image. The image quality was improved while maintaining the same embedding capacity because the human eyes could rarely see differences in the edge portion. The embedding capacity can also be adjusted based on the demands of individual users. In addition to the improvement on image quality, the proposed approach provided respectable security as well. Wu and Tsai(2003) proposed an algorithm using pixel-value differencing which partitioned the original image into non-overlapping blocks of two consecutive pixels. A different value was calculated from the values of the two pixels in each block. All possible different values were classified into a number of ranges. The human vision sensitivity to gray value variations from smoothness to contrast was used in selecting the range intervals. A new value which replaced the different value was used to embed the value of a sub stream of the secret message. The width of the range that the different value belongs to determines the number of bits that can be embedded in a pixel pair. However, in this method the modification is never out of the range interval. The result produced by this method is more imperceptible than those yielded by simple least significant bit replacement method. The secret message that was embedded can be extracted from the resulting stego-image without making reference to the method of the original cover image. The security of the method was shown using dual statistics attack. Scott’s work on steganographic techniques using digital images used several iterations of replacement strategies during the construction of the application. The aim was to implement are placement and extraction steganography scheme using cover images. To extract the embedded textual information from the image, the image created by the application must be processed. This processing outputs the original message and some extra erroneous information. Comparison between LSB replacement scheme with MSB replacement scheme asserted that MSB produced noticeable differences to the cover during the most significant bit replacement. Rohit and Tarun(2012) compared LSB and MSB based steganography in gray-scale images. It was concluded that the resulting stego-image using LSB shows no distortion when compared with the original image.The performance of LSB was better than that of MSB. Kanzariya and Nimavat(2013) compared various image steganography techniques. The objectives were to identify the requirements of a good steganography algorithm and to determine steganography techniques that are suitable for different applications. In this work,comparison is done using the Mean Squared Error, Peak Signal to Noise ratio and the encoding time.
CHAPTER THREE
METHODOLOGY
This chapter reviews how the existing system works as well as how to produce a better alternative for its improvement. The relationship among actors, entities, platform and information flows within the organization is very important. In a nutshell, system investigation and analysis studies an existing system with the view of improving on it or developing an entirely new system to replace the existing one. The major task here is to design a new system using tested and trusted development methods that is as efficient and probably more efficient than the existing one.
Specifically, this chapter outlines the way the experimentation on each algorithm will be carried out. Each algorithm will be tested using the parameters such as the mean Squared Error, the Peak Signal to Noise Ratio and the Encoding time.
FACTS FINDING
Fact finding is an approach taken to acquire data about a specific or subject with the aim of analyzing and synthesizing the analyzed data to come up with a better system. Fact finding for this study was done by examining related publications, research work, journals and books.
ANALYSIS OF THE EXISTING SYSTEM(S)
Steganography has been a technique from ages past used to send secret messages via images. It offers security features such as authentication, confidentiality and verification. It is the art and science of hiding and communicating data through apparently reliable carriers in attempt to hide the existence of the data. This way, there is no knowledge of the message being sent in the first place.
Below in Figure 3.1 is a typical steganography system.
CHAPTER FOUR
RESEARCH RESULT
INTRODUCTION
This chapter discusses the experimentation process of the three steganographic techniques. In this chapter, we test the Mean Squared Error, Peak Signal to Noise ration and encoding time and see the effects of each algorithm on those parameters.
PROGRAMMING LANGUAGE SELECTION
The programming languages used in this project is JAVA because of its cross platform nature and the power of compilation. The system that is built includes the encoding interface and the decoding interface for hiding and retrieving purposes respectively.
CHAPTER FIVE
CONCLUSION, SUMMARY AND RECOMMENDATION
SUMMARY
This entire project has compared the results of LSB, MSB and the combined LSB – MSB algorithms by calculating the MSE and PSNR. LSB algorithm gives an overall better performance than other two algorithms. However, the combined LSB – HSB algorithm algorithm’s image quality is approximately as good as the stego image quality of LSB algorithm. But the combined LSB – HSB algorithm algorithm’s image quality is better than MSB algorithm’s. As the value of PSNR is more than 40dB and this of MSE is lower, the image quality of the combined LSB – HSB algorithm is better. Moreover, the combined LSB – HSB algorithm algorithm gives better security because it is more complex than other two algorithms. LSB and MSB algorithm are so easy to decode that its security is weak. As complex algorithm such as the combined LSB – HSB algorithm complicates to decode itself, its security is strong. In the future work, it should be experimented that audio and video will be able to embed into the cover image with this combined LSB – MSB algorithm. It is better to be upgraded to the combined LSB – MSB algorithm to get superior image quality and security.
CONCLUSION
In conclusion, the LSB algorithm gives a better performance than the combined LSB-MSB algorithm but a larger file size of the cover image makes the combined algorithm produces images with good quality. However, the proposed Hybrid algorithm produced better image quality than MSB algorithm.
RECOMMENDATION
I will recommend after this project that more work can be done to look into improving the algorithms particularly the combined LSB – MSB algorithm. This can be achieved by working on the compression ratio for stronger embedding procedures and also finding a way to apply this technique on larger sized gray-scale images.
REFERENCES
- El-Emam N. N (2007). Hiding a Large Amount of Data with High Security Using Steganography Algorithm, Journal of Computer Science, Vol. 3, pp. 223 – 232
- Chen P.Y., Wu W.E. (2009). A Modified Side Match Scheme for Image Steganography, International Journal of Applied Science & Engineering, Vol. 7 pp. 53 – 60
- Chang C.C. and Tseng H.W. (2004). A Steganographic Method for Digital Images using Side Match Pattern Recognition Letters, 1431-1437
- Wu P.C, Tsai W.H. (2003). A Steganographic Method for Images by Pixel-value Differencing, Pattern Recognition Letters, pp. 1613 – 1626
