Font Size: a A A

Research On Algorithms Of Multiple Digital Watermarking And Verification For Handwritten Signature

Posted on:2016-01-26Degree:MasterType:Thesis
Country:ChinaCandidate:X X AnFull Text:PDF
GTID:2308330479489731Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
Digital watermarking is an effective technology to protect the copyright of digital products. Handwritten signature is a traditional authentication way which is widely accepted by public. The traditional watermarks are icons, serial numbers and other identification information. The combination of digital watermarking and biometric authentication is a new research direction. In the digi tal era, paper media is gradually replaced by digital media. How to apply the signature authentication to digital files and electronic bills is an urgent problem to solve.The thesis studies digital watermarking and handwritten signature verification algorithms. The handwritten signature, as watermark, is embedded into digital carrier, which applies the signature authorization to digital files and electronic bills. On the one hand, signature and carrier is one unit, and it is difficult to remove the signature from the carrier. On the other hand, signature is embedded into carrier, invisibly. Only legally authorized users can extract the signature, which protect the security of signature information.In this thesis, we propose the multiple watermarking algorithm based on K-L transformation, which divides the carrier into multiple independent subspace s, and embed watermarks into subspaces. Compared with other frequency domain watermarking algorithms, the algorithm is more suitable for multiple watermarking, with better invisibility, robustness and similar robustness for multiple watermarking. The thesis studies the dynamic time warping algorithm. The on-line handwritten signature verification based on DTW obtains a 7.48% equal error rate. We propose an on-line handwritten signature verification based on DTW and SVM, which use DTW to compute the similarities between test signature and template signatures to form a feature vector. The algorithm obtains 9.00% false accept rate and 15.33% false reject rate. The amount of original signature data is too big, which severely undermine the quality of carrier. We study three methods for feature extraction and data compression based on inflection, PCA and wavelet. The results show that the method based on wavelet is better, which have good performance for data compression and signature verification, to achieve the combination of digital watermarking and signature verification.
Keywords/Search Tags:digital watermarking, on-line handwritten signature verification, K-L transformation, dynamic time warping
PDF Full Text Request
Related items