基于网格计算技术的人脸识别研究-仪器科学与技术专业论文.docx

II II ABSTRACT Because of great foreground in security validation system, detecting crime and human-computer interaction, automatic face recognition already has became a interesting point in many research fields ， such as pattern recognition, artificial intelligence and computer vision. As a new research field, it is attached importance by many research organizations, universities, enterprises and governments in the world. In the past several tens years, many researchers proposed various different face recognition algorithms to enhance the accuracy of face recognition, but by now face recognition has seldom been successful in actual application. It is because of two key issues in the face recognition for huge human face database: One is the ability of handling huge human face image data; The other is that, under different condition such as illumination, posture and complex background, the robustness of face recognition algorithm is very bad. Namely one algorithm is only better fitted for one kind of human face database. There is not a general algorithm for all kinds of human face database. But the central system or the traditional distributional system is unable fundamentally to solve these problems. How does it effectively integrate the different dispersed algorithm and handling ability to establish a general coordinate intelligent virtual processing platform to fully use and share each kind of resources and enhance the processing speed and ability? The appearance of grid technology unfold a very good prospect for solving these problems ，because the final goal of the grid is to realize sharing of resources such as data resources, computing resources and so on. Thus it can effectively use the unused resources. For face recognition two key questions existing in face recognition,the former may be resolved by distributional database based on grid and the experience of practical applications, such as computing grid, data gird, Globus and so on. It d