Face Recognition Based On Data Fusion

Face Recognition has been applied in the fields of security, finance, e-government, e-commerce, human-computer interaction and networks widely. It is very important to improve the performance of face recognition to extend its application fields and boost its commercialization. Generally speaking, there are three key ways to hence the robustness of face recognition, that is, robust feature representation, more discriminative feature selection and more powerful classifier design.In this thesis, based on some relative basic theories and methods of face recognition technology, the wavelet transform and data fusion, we do some researches along the route of the multi-resolution analysis, the wavelet subgraphs selection, data fusion and face recognition. The research contents and innovation points of this thesis are as follows.1. This thesis sums up some important researches of face recognition and makes a systematically introduction some related methods and research status. An overview of the theory of data fusion and its applications on face recognition are introduced in this thesis, from which we obtain the research direction and it provides a research approach for this thesis.2. An algorithm is presented in this thesis to select wavelet subgraphs based on fisher ceiterion. Face image is decomposed by using wavelet transform combined with multi-scale analysis in dealing with the influences of illumination variations. In order to reduce the computational complexity we apply fisher ceiterion in projection of all subgraphs and to select subgraphs with important information.2. A face recognition method based on data fusion is proposed in this thesis. The fusion of selected subgraphs is conducted on data level, feature level and decision level respectively and then we make a recognition and decision on the fusion results.The experiment compared with methods of some other papers on face databases shows that the performance of face recognition based on data fusion is better and the proposed thesis is kind of innovative and makes some significance.