Please use this identifier to cite or link to this item:
Title: Ridgelet Transform based Robust Image Watermarking Technique
Authors: JashanJot
Supervisor: Kasana, Singara Singh
Keywords: Ridgelet;RPCA;SVD;PSNR;Arnold Transform;NC
Issue Date: 28-Aug-2018
Abstract: Image watermarking is a mechanism to embed secreat data within the image contents to protect the copyright and prove authentication. Watermarking is basically requirement for improving the robustness and invisibility opposing the various attacks in embedding watermark procedure. Among the two main types of domains used for watermarking, the more robust domain is transform domain as compared to spatial domain. Wavelets are widely used in existed algorithms for transform based watermarking but the wavelets are able to just describe the three image directions namely; horizontal, vertical and diagonal when dealing with the images. Wavelets fail to handle the linear singularities because of their isotropic support and hyperplane singularities of the image more effectively or the point wise singularities. These problems are effectively overcome by Multi-scale geometric analysis. In order to capture the geometric regularity of an image and to find an optimal directional representation; Curvelet, Contourlet, Ridgelet Transforms have been proposed. In this work, hybrid watermarking system is proposed which uses Ridgelet Transform, Robust Principle Components, Singular Values Decomposition in the watermarking process. First input image is transformed into HSI color space and I channel is operated by Robust Principle Components to get the low rank matrix. Further, Singular Values Decomposition is applied on this matrix to decompose the contentS in order to embed the watermark in it. Using Arnold Transform, watermark is scrambled to make it secure. Ridgelet transform is directional sensitive, and provides edge information in the image. Unlike other frequency transform domain, Ridgelet Transform has spatial frequency locality and is robust towards a broad range of attacks. Proposed technique is more robust and imperceptible then the existed algorithms. The quantitative and visual results show the effectiveness of the proposed technique as it is highly tolerant against various geometric and image-processing attacks.
Appears in Collections:Masters Theses@CSED

Files in This Item:
File Description SizeFormat 
thesis_Jashanjot_2018.pdf3.86 MBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.