Comparative Analysis Of Image Stegnography Techniques
Abstract
In this paper, discuss about the data hiding using the wavelet approach is better technique in stegnography techniques. optimisation techniques are better results provides for the data hiding in stenography. In Discrete Wavelet Transform, HAAR Wavelet gives the excellent peak signal to noise ratio (PSNR) and less computation time. In optimisation, particle swarm optimisation technique is gives excellent better result in case of PSNR ratio. In spatial domain, common useful technique is least significant bit(LSB) gives better result in case of data payload capacity and less computation time. In this paper, discuss all above techniques with compare to other related techniques useful in stegnography.
Keywords—Image,Secret information,Discrete wavelet transform(DWT), Particle swarm optimisation(PSO), Least significant bit(LSB), PSNR, MSE.
Introduction
Steganography technique is a good technique to keep secure information same as Cryptography. In stegnography, all about to information hide/embed in specific objects like, text, image, audio, video, etc. In cryptography, it is about to encrypt secret information. In some application, steganography is seems to be a best idea than cryptography. The steganography has the advantage that the secret data that is the observer doesn’t know the existence of data because of. Whereas in cryptography we feel the existence of data because human visual system can see even encrypted information publicly if it is detected. Steganography useful in many fields, forensics details to transfer, military important information transfer with security, government information to keep secure, etc. Using of stenography has some benefits,
- In some case, it does not need any key to access secret information.
- It seems to be a simple transfer sender to receiver.
- Detecter can not observe something is hidden in cover objects.
If detecter can assume some information is hidden in cover object than also can not find out which methods are applied. In stegnography, information can hide/embed in text, image, audio, video as a cover object. In stegnography is works on spatial domain technique in which most common method is LSB as least significant bit. Other useful technique is transform domain technique. LSB is simplest techniques in all other methods. For Image cover object, LSB works on pixels to hide/embed information. In LSB technique, Pixel’s bits of cover image is replace with secret information pixel’s bits. LSB is applied to whole pixel of image to hide data according to secret information size. LSB is data lossless method and data is embed in image with less computation time. In disadvantages of LSB, it is easily prone to attack[6]. If secret data easily can detect than no meaning of stenography to use for security propose. so, we need to find out other better with LSB techniques which are provides more security.
Discrete wavelet transform
Discrete wavelet transform domain perform on image for spatial domain to convert in Frequency domain with low and high frequency sub-bands. Apply one level wavelet transform on image decomposes the cover image into four sub-bands. In, respectively horizontal, vertical and diagonal. Discrete wavelet transformed low frequency range pixels are stored in LL sub-band. Some low and high frequency based pixels separate in LH band respectively horizontal dimension. Other pixels in HL and HH sub-bands respectively vertical and diagonal dimension. All sub-bands are separate in equal image size. For simple DWT perform on image it represent by flow diagram, image flow diagram In frequency domain, reconstructed image applied DWT 2D level. To get original image using an inverse wavelet transform. When DWT apply on image it is important to quantify signal-noise-to-ratio. Signal-noise-to-ratio(PSNR) measures quality of image[20]. In DWT domain quality of image ratio is increase because of after separated pixels in frequency unnecessary noise is removed from image. DWT is provides good quality of image and improve security for stegnography because of frequency based image can not easy to detect compare than spatial domain for stegnography[20]. In case of security, wavelet domain is less prone to attack than spatial domain. In comparison survey DWT is good technique with other techniques and provides a multi layer security with good quality of image and more computation time. DWT is trending technique for stegnography as a security purpose it is very useful.
Discrete cosine transform
Discrete cosine transforms(DCT) express a function or a signal with different frequencies. DCT is that the former uses only cosine functions. DCT is perform in low, high, medium frequency. DCT is similar technique as DWT. It is also perform in 1D and 2D or more levels to compress to image in equal size sub-blocks. Discrete cosine transform provides good security for stegnography but less payload capacity than LSB. DCT is a useful in case of security but for limited data information. Feno Heriniaina Rabevohitra and Jun Sang used the PSO and simple LSB substitution based in DCT domain. DCT domain works on image to lossy compression of image.
Optimisation Techniques
In optimisation techniques, it is useful to find optimum solution to problems. generally it uses for the neural network and computer as well as for biological solutions. In optimisation techniques like genetic algorithm, particle swarm optimisation are provides better solution according to there problems. PSO is particle optimisation techniques are newly and trending methods to apply on image stegnography to find best position in image for embed data[5]. The cost function is p evaluate the fitness of pixel position. Cost function factors such as entropy, edge, and individual point edge. PSO is swarm population based algorithm. PSO is perform on particles which are moves to find best position in image for hide information. Particle is a initialise according to there population and moves on one to other places in image by one-by-one generations/iterations and finds best position by pbest and gbest position. By performing personal best and global best positions of particle’s position for embed data in image in x & y co-ordinates. Particles are updates velocity and fitness function value according to their movement for each iterations. PSO is about to search best position by optimising pixels position. In image which Pixels are best or fit for embed information. PSO algorithm gives final results as optimised image. Optimised pixels values are more nearly with original pixels values and that is provides minimum difference between original image and optimised image. So, it is benefits of PSO, after optimisation quality of image is improved. In optimisation,Kennedy and Eberhart, proposed the particle swarm optimisation which is similar to the genetic algorithm. Genetic algorithm has more computation parameters sources as cross over and mutation. It takes more generations than PSO. Genetic algorithm is more complex for stegnography compare than PSO. In fig. 2 perform PSO algorithm by flow diagram. In PSO, fitness function, velocity, best values are updates till the all iterations perform.
After DWT sub-bands by LSB technique data is embedded in HL,LH,HH sub-bands. DWT with LSB techniques are provides multilayer security for data and PSNR and MSE gives good results using DWT and capacity of information embed is more using LSB. Aman Arora, Prateek Thakral. is used LSB technique to hide information with encryption algorithm like AES. Data is encrypted by AES, plain text converts into cipher text. Cipher text data is embedded in image by LSB. Data is replaced with RGB model pixels as red, green and blue pixels bits. Secret Data using this methods can embed maximum size. In grey scale and colour image are used but colour image gives good average PSNR than grey scale image. S. M. Masud Karim, Md. Saifur Rahman, Md. Ismail Hossain. is a proposed method using LSB with any secret key to embed data in image. In paper, randomly “sohel” selected as a secret key to convert in binary image to 1D array with XOR to embed data in RGB model of cover image. In extraction phase of data, using secret key is need to know for get secret data. If secret key is loss or detecter know key then data can easily detect.
In result,LSB technique gives good results in PSNR with 53 db or can be improve. Rafael Lima de Carvalho,Warley gramacho da silva. is a proposed work on PSO optimisation techniques applied to hide data in cover image. After optimisation of image data is embed in optimised pixels. Optimised pixels are nearest with original image pixels and quality of image is improved. PSO with genetic algorithm also compared. Optimisation is good option now and trending method for stegnography. Dr. P. Rajeswari,Ms. P. Shwetha,Dr. S. Purushothaman. is a combine methods like DWT, PSO and LSB. DWT is applied on secret information as a image perform 5 levels of sub bands and resultant output get with 8*8 matrix image. On 8*8 image applied PSO method to find best location with 100 iterations. Maximum iterations get good quality of image. Best location in image of co-efficients are select to embed secret data using LSB technique. Using three methods are provides more security and increase quality of image. In future to calculate PSNR and MSE for this methods. In future it can possible to measures text based secret information embed using methods. Ms Rashmi N,Dr Jyothi K. is a proposed method, encryption and reversible data hiding(RDH) techniques applied with LSB method to compare there PSNR ratio and MSE. For encryption, AES algorithm used. Encrypted data using RDH method and LSB technique embedded in image. Perform good for only grey scale image not for colour based RGB image. In future needs to apply methods for colour image.
Conclusion
In Image steganography, many techniques are presents according to survey but some few techniques are in trending to improve results for quality of image and security purpose,like DWT,PSO and LSB or more. In spatial and frequency domain uses to improve security for less detection. Which are gives good performance. LSB in spatial domain and DWT or DCT in frequency domain can improve results. PSO is optimisation method perform with good quality results with DWT and LSB. In final outcomes, LSB is useful for more capacity, less computation time and DWT method for less detection, PSO useful for increase quality of image.
