Technology Combining Fingerprint Recognition With Watermarking Used For Copyright Protection And Authentication

With the development and popularization of computer, multimedia and internet applications, the demand and requirement of network information security grow stronger and stricter increasingly, such as copyright protection of digital multimedia and identity authentication in e-commerce activities. On the basis of systematic analysis of previous research results, the digital watermarking technique used for copyright protection and the automatic fingerprint recognition for identity authentication are studied deeply in this paper. Some of the existing key algorithms of digital watermarking or fingerprint recognition are designed, improved, discussed or implemented. Firstly, taking advantages of nonnegative matrix factorization and scale invariance feature transform, a strong robust image watermarking scheme against geometric attack is designed. The scrambled watermark is embedded into the local feature of contourlet low-frequency subband using odd even quantitative criteria. Secondly, a contourlet domain image watermarking algorithm based on support vector regression is discussed. The non-linear local correlation model of the image is established using support vector machine so as to adaptively embed the watermark. Then, a ridge orientation field estimation method for fingerprint image is implemented in frequency domain. Both the gray scale statistical properties and ridge frequency informations in local fingerprint image region are taken into considered in order to estimate the ridge orientation. Next, a fast image thresholding algorithm based on minimum within-cluster absolute difference criteria is improved. Not only segmentation result is improved by introducing the gray scale-gradient histogram, but also computational efficiency is advanced by means of simplifying the formulae for threshold selection and optimizing the searching procedure using a recursive simple and efficient particle swarm alogorithm. Finally, a fingerprint minutia matching method based improved Tent map chaotic particle swarm algorithm is explored to improve the computational efficiency and matching accuracy.