Font Size: a A A

Research Of Digital Watermarking Against Geometrical Attacks In Feature-based Region

Posted on:2016-02-09Degree:MasterType:Thesis
Country:ChinaCandidate:X JinFull Text:PDF
GTID:2348330536454801Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
How to make the watermark stronger to geometrical attacks has always been a hot area of research in digital watermarking.After deeply studying the second generation watermarking algorithm and invariant feature extraction method,this paper puts forward two novel algorithms.Embedding the watermark into the local feature-based image region can improve the robustness of the color watermarking algorithm based on frequency domain.Firstly,an adaptive watermarking in DC coefficients based on Harris corner detection is put forward.In order to embed the watermark into the region with more Harris corners,this paper proposes a Harris-weighted algorithm which can detect the feature points of the carrier image.Then,classify the sub blocks of the embedding region according to the AC coefficients and choose the proper strength to embed the 1 bit watermark into the DC coefficients of the sub blocks.Compared with common adaptive watermarking in DC coefficients,the proposed algorithm has very strong robustness against rotation attacks owing to the rotation-invariance of Harris.This paper puts forward another novel DCT-SVD based watermarking algorithm against geometrical attacks after focusing on SIFT.First,extract the stable SIFT feature points of the carrier image to construct the local embedding region and then embed the watermark by comparing the DC coefficients and singular values of the image matrix.In order to ensure robustness,common blind watermarking algorithm based on local region embedding always embeds the same watermark into different regions.But the proposed algorithm avoids the limit of embedding capacity in the common blind watermarking algorithm through combining the local regions to embed a whole watermark.By correcting the attacked image through SIFT feature matching,the proposed algorithm can resist the geometrical attacks well.Experimental tests show that the two proposed schemes have perfect invisibility and are robust against kinds of attacks especially geometrical attacks.The complex secret keys can enhance the safety of the algorithms.
Keywords/Search Tags:digital watermarking, geometrical attacks, feature extraction, DCT
PDF Full Text Request
Related items