The Research And Application Of Digital Watermarking In Biometric Recognition System

In recent years, the rapid development and extensive application of biometric recognition technology bring convenience and safety to people’s daily life at the same time. Because of potentially various malicious attacks in the process of biometric recognition system which produce a lot of security problems, enhancing security performance of system is increasingly urgent and important. As a new technology of information hided in the field of multimedia information security, the digital watermarking technology is introduced into biometric recognition system which can effectively solve the part security issues in the process of acquisition, storage, transmission and recognition of biometric images, and it has obtained very good results in this field.Compared with fingerprint, iris and veins of physiological characteristics or other behavioral characteristics, face feature recognition has a lot of advantages, such as direct, friendly, convenient and non-contact etc., and it contains a huge development potential in the field of identity authentication, video surveillance and human-computer interaction. This paper will take face biometric as an example, building a face recognition system and using digital watermarking technology to protect the face feature data in the process of identity authentication. Paper’s main work is as follows:1. This paper describes the general structure process of face recognition system, deeply analyzing AdaBoost algorithm in the process of face detection, K-L transform and principal component analysis in the process of face recognition. Through Visual C++ combined with OpenCV programming to realize face recognition system, this paper gives the simulation results of face detection, the training samples of image acquisition and training, face recognition and others.2. This paper introduces the basic framework of digital watermarking system, and the performance evaluation standards of watermarking embedding and extraction algorithm is respectively presented. This paper simply narrates the flat structure of gray image, giving digital watermarking algorithm based on least significant bit of spatial domain, and describing the basic principle of discrete wavelet transform DWT. Through Matlab programming to realize the digital watermarking algorithm of DWT based on the frequency domain with HVS characteristics, the simulation results of watermark embedding and extraction are respectively given.3. According to the characteristics data in biometric recognition system can be potentially attacked, as well as the study of face recognition system and digital watermarking technology, the digital watermarking technology application in face recognition system are classified. Specific to partial attack of system and not affect the recognition rate is given two solutions:digital watermarking technology combined with qr code to protect data security and digital watermarking technology application in multimode biological features to protect the security of the data. Through the experiment to simulate external attacks that may be presented in the system, such as noise, rotation and cropping etc., the above solutions can be good to resist these attacks, and obtain a good effect.4. The face recognition system is designed and implemented based on DSP platform, and this paper introduces the hardware design scheme of system with related peripherals of the core processor, and the software development environment with the main software module of software design scheme. The experimental results of face detection are given in the SEED-DEC643 board, the face image data transmission and off-line recognition results.