击键特征身份识别技术在光盘库客户端的应用-计算机技术专业论文.docxVIP

华中科技大学硕士学位论文 华 中 科 技 大 学 硕 士 学 位 论 文 II万方数据 II 万方数据 Abstract With the development of Internet technology, the security of users privacy, account, and enterprise business information has increasingly become the focus of attention. As a mass data storage device, Optical Jukebox save a lot of important businesses and individuals data. In the Optical Jukebox’s security environment, user authentication is the most basic blocks and the main line of defense to the access control and user audit to Optical Jukebox client. So to do user authentication is particularly important . In order to strengthen the identity authentication, studied the biological identification technology based on keystroke characteristics, using this technology in combination with password identification on Optical Jukebox client, analyzes the technology’s theory also its advantages and disadvantages. Use C written a software program to collect the users keystrokes and save time characteristic data; To form a training dataset, analyzed and extracted the characteristics of data, for high-dimensional feature data using Principal Component Analysis(PCA) method to reduce the dimensions; Support Vector Machine(SVM) showed a relatively good advantage in solving small sample size, high-dimensional and nonlinear pattern recognition, using SVM to train the training dataset and form the model of the users keystrokes characteristics, used to identification authentication when users log in; In order to improve the accuracy of the identification models, respectively, using Particle Swarm Optimization(PSO) and Genetic Algorithm(GA) to select the optimizing operations of penalty parameter and kernel function parameters for SVM training. After testing, data acquisition module can correctly collect the users keystrokes time data, which is accurate to millisecond, feature extraction module can correctly extract the keystroke characteristic data, the training module can get the right model and identify the logged in use