The Research Of Face Recognition System Based On Two Dimensionality Locality Preserving Projection

For decades of development, face recognition technology has made considerable progress, but practice shows that the face recognition technology is far from mature in general way. To develop a truly robust and practical application systems of face recognition, we also need to research more robust and higher performance of the core algorithm, the core R & D of algorithm involves with a large number of algorithm experiment, and it is a complex and arduous work .So it is particularly important to develop a flexible, fast algorithm verification system from software engineering point of view. In response to these issues, this paper has the algorithms to improve the efficiency and the development of flexible, rapid method of verification system in-depth study. Details as follows:①Rising a improved face recognition method based on two-dimensional locality preserving projection. Firstly it uses 2DPCA(Two dimensionality Principal-Component Analysis) to reduced-order processing, then retain a significant two-dimensional information in maximum. In the reduced-feature space, it uses optimization methods of 2DLPP(Two dimensionality Locality Preserving Projection) to build a optimization projection matrix. As the application of 2DPCA and 2DLPP resulted in fewer parameters to improve the operating efficiency of the entire algorithm. By ORL experiment contrast, the algorithm in the performance has a certain advantage.②To develop a flexible and efficient algorithm verification system of face recognition, it starts with basic links of recognition systems, and introduces the popular system structure of "plug-in" in the software engineering into the recognition systems. To achieve a true parallel algorithm development, the purpose of improving, A unified "interface" was designed by in-depth study of the structure of the algorithm, which provides a unified platform for experiment. At the same time, the platform makes all aspects of face recognition being flexible "assembly", and assesses the efficiency of the face recognition algorithm from multi-angle.③Face recognition system achieved automation test of algorithm and export parameters, which implemented in the same experimental conditions the contrast of efficiency, in order to save time, improve the efficiency and speed up the algorithm improvement.