Optimization of perceptual steganography capacity using the human visual system and evolutionary computation

Publication Type:
Thesis
Issue Date:
2016
Full metadata record
Files in This Item:
Filename Description Size
01front.pdf140.43 kB
Adobe PDF
02whole.pdf14.03 MB
Adobe PDF
Efficient solutions for the purpose of delivery of information are called for by the revolution of internet. However concerns and problems over security, distribution of digital content and encapsulation of media artifacts have arisen as a result of these phenomenal developments. Hence, it has become necessary to seek capabilities to transport and secure multimedia with its meta-data in a safe way. Steganography has evolved as an enabler of multimedia applications keeping secret communication and embedded captioning secure. There is a tolerable outcome that occurs between imperceptibility and steganographic capacity that fit right into the mix. For instance, the more subtle elements are hidden within the cover object having higher capacity, the more degradation is exhibited towards the carrier file, resulting in an increase in the distortion attributed to the information being concealed and at the same time, decreasing the stego file quality. Suitable use of Evolutionary Algorithm and effective use of the weaknesses of Human Visual System in steganography are investigated in this thesis. Firstly, two high capacity steganography approaches are developed with the use of aforementioned features. The first method aims to overcome the limit capacity of edge based steganography in the spatial domain. The second method proposes a proper threshold selection for each coefficient which increase the capacity of transform domain. An estimate of the embedding rate based on image complexity is also proposed. Moreover, since peak signal-to-noise ratio (PSNR) is largely used as a measure of quality of images of stego, the reliability of current quality assessment metrics for stego images is also evaluated at the third stage. Follow by developing an Anticipatory Quality Assessment Metric for effective imperceptibility measurement. All proposed methods are aimed to assist the optimization of the statical and visual characteristics in the cover images while hiding large size of information. To reveal impressive imperceptibility and capacity of the proposed method over the existing dilemmas, a broad range of requirements have been carried out. To indicate the utility and value of all techniques proposed, they all have been empirically validated. The main aspects of image steganography are improved by the suggestions and methods, and are revealed by the results.
Please use this identifier to cite or link to this item: