With theadvancement in digital communication and data set getting huge due tocomputerization of data gathering worldwide, the need for data security intransmission also increases. Thanks tothe proliferation of multimedia data including text, images , audio and videowhich are the basic elements of multimedia data including text,images,audio andvideo which are the basic elements of multimedia platforms and ever growingrequests for applications based on them.
Multimedia bid data applications include video on demand, 4k displaycontents,interactive video systems,surveillance,health care etc and all of themrequire secured environment during transmission and reception. Data security ensure that information remainsconfidential,no one will be able to change the information thereby increasingintegrity and also make sure data gets transferred to the intended recipientsthrough a secured channel.Cryptography andsteganography are wll known methods available to provide security where theformer use techniques that control information in order to cipher or hide theirpresence and the latter concentrates on data concealment.Steganography isthe practice of masking data especially multimedia data within anotherdata. Visual contents gets moreimportance from people compared to audio contents and moreover visual contentfile is huge when compared to audio file thereby helping increase robustness ofthe hiding algorithms. Due to the hugesize of image and video data, they get compressed before transmission or storageand the commonly used compression technique standards are JPEG and MPEGrespectively. Both the encodingstandards use Discrete Cosine Transformation(DCT) internally for transformingthe signal from spatial domain into frequency domain.
The simplest and widespread imagesteganography method is the least significant bit (LSB) substitutionmethod. It hides data into carrier imageby replacing the least significant bits directly. The embedding capacity can be increased byusing up to 4 least significant bits in each pixel. Though it is popular, it has a common weakpoint as well which is that the sample vlue change asymmetrically. To overcome these limitations,we propose touse statistical methods in identifying the source signal based on the messagesignal and then encrypt the data before embedding.
Most imagesavailable on the internet are JPEG compressed and that is the second domainthat is considered for image steganography. JPEG uses discrete cosine transformation for converting the spatial domainimage into frequency domain before further processing. The DCT transformed image contains DC and ACcoefficients. Here the DC componentrepresents the average color while the AC components corresponds to colorchange across the block. Most of the steganographyalgorithms use DC coefficients for hiding the message while in this work, wetried to use both DC and AC components together which help in concealing thedata in a more robust way there byincreasing the accuracy when compared to traditional algorithms. XOR operations along with Chaotic encryptionis proposed.
A wavelet is amathematical function which is one of the recent developments and useful indigital signal processing and image compression technologies. The use of wavelets for these purposes is arecent development, although the theory is not new. Lifting scheme gives wavelet filter design toperform discrete wavelet transform. Wehave tried to explore the same and use it in image steganography.
Data encryption standard, a symmetric-keyalgorithm in which one-dimensional chaotic map is used to encrypt the imagesignal before getting embedded to the carrier signal. This helps improve the robustness of theproposed algorithm.The proposedmethods can be applied to the following domains including confidentialcommunication and secret data storing, access control system for digitalcontent distribution, protection of data alteration and media database systems. We have tested the proposedalgorithms on the USC-SIPI image database which is a collection of digitizedimages and provided by the university for supporting the research people inimage processing domain. Carrier imagesize used varies from 512* 512 to 1024*1024 while the message signal fallsunder 256 *256 category.We find that theproposed algorithms have a better PSNR value averaging close to 71 while thetraditional algorithms has values of around 56. XOR operations for encrypting the data along with key table usagebetween the sender and receiver helps improve the efficiency in embedding thesecret data. Moreover we have alsostudied different Steganalysis methods and made sure that the proposedalgorithms are robust enough to stay safe from those methods.
Other advantagesof the proposed method include reducing the overall bit-level changes to thecover image for the same amount of embedded data and avoiding complexcalculations. We have achieved betterembedding capacity without compromising on the security factor as well throughthese algorithms. Finally, the new methodsproposed results in little additional distortion in the stego image, whichcould be tolerated.