Study On Image-based Face Recognition System Of Tour Buses

An image-based face recognition system for tour buses is described in this paper. The thesis studies the human face feature extraction and dimension reduction techniques, and the design of the face recognition system of tour bus systematically, based on the curvelet known as multi-resolution mathematical tool and subspace method.In feature extraction part, the curvelet transform and other algorithms are used to extract facial features. The curvelet coefficients contain not only the low-frequency information, but also the high-frequency information, which reduces the dependence of illumination. Our feature extraction method can achieve higher recognition rate and a low illumination sensitivity compared to wavelet and PCA method confirmed by experiments.In feature dimension reduction part, we focuses on studying the PCA, LDA, 2DPCA, ICA and other sub-space method in face recognition. The 2DPCA-LDA is used to reduce the dimension of the data generated by curvelet. The experiments result confirmed the 2DPCA-LDA double dimensionality reduction method compared to 2DPCA,LDA,PCA and other dimension reduction method is more suitable for the curvelet dimensionality reduction, which can run quicker, and consume less computing and storage resources.We also give design of the image-based face recognition system for tour buses and some simulation experiments are taken. And the simulation experiment results show our curvelet and 2DPCA-LDA based face recognition method is suitable for face recognition system for tour buses and verify the effectiveness of proposed design.