DCT domain stegasvm-shifted LSB model for highly imperceptible and robust cover-image

• Main Authors: Yahya, Saadiah, Hussain, Hanizan Shaker, M., Fakariah Hani
• Format: Conference or Workshop Item
• Language: English
• Published: 2015
• Subjects: QA75 Electronic computers. Computer science
• Online Access: http://repo.uum.edu.my/15584/1/PID043.pdf; http://repo.uum.edu.my/15584/; http://www.icoci.cms.net.my/proceedings/2015/TOC.html
• Institution: Universiti Utara Malaysia

The importance of information security in protecting and hiding information has increased due to the increased use of computers and Internet.Information hiding technology such as Digital Image steganography embeds secret messages inside other files.Least Square Bit (LSB) is the most popular technique used in image steganography that hides data behind a cover-image in a spatial and discrete cosine transform (DCT) domain.Support Vector Machine (SVM) is another technique that is used to strengthen the embedding algorithm.The main aim of image steganography is to keep the secret-message remain secret regardless of the techniques used.But many of the previously proposed techniques failed to attain this aim.The main concerns to this problem are the non-random changes of a cover-image that constantly occurred after the embedding process and the non-robustness of the embedding algorithm to image processing operation.This study therefore proposes a new model that utilises Human Visual System (HVS) and embedding technique through shifted LSB called StegaSVM-Shifted LSB in DCT domain to preserve the imperseptibility and increase the robustness of stego-images.The proposed technique shows better performances compared to other existing techniques.