Research On Discriminant Feature Extraction Methods In Image Recognition

With the development of computer technology,image recognition technology has been more and more widely used,it has become a research hotspot in the field of pattern recognition,and feature extraction is a key issue in image recognition.In the past few decades,researchers at home and abroad have proposed many different feature extraction methods,and most of them have achieved good results in the field of image recognition.In order to further improve the robustness and discriminative power of the extracted features,this paper does the following work:(1)In order to make the neighbor graph better match the subspace learning process and capture the global subspace of the data and the local geometry simultaneously,this paper proposes a low rank discriminant adaptive graph preservation(LRDAGP)subspace learning method for image feature extraction and recognition.Specifically,LRDAGP combines adaptive manifold learning with low-rank subspace learning into a framework,adaptively updating the neighbor graphs in the subspace learning process,and considering the global subspace structure and local geometry of the data simultaneously.In addition,LRDAGP introduces a supervised regularization term in its objective function to enhance the discriminative power of the subspace.This paper designs an efficient optimization algorithm to solve objective function of the LRDAGP.(2)In order to reduce the irrelevant information in the original data while reducing the influence of outliers,this paper proposes a robust version of DLPP based on L2,p-norm with 0 < p < 1,termed DLPP-L2,p,for image feature extraction and recognition.DLPP-L2,p learns an optimal projection matrix by maximizing the L2,p-norm-based locality preserving between-class dispersion and minimizing the L2,p-norm-based locality preserving within-class dispersion simultaneously.Furthermore,by imposing an L2,p-norm penalty on the projection matrix to achieve row-sparsity,DLPP-L2,p can discard irrelevant features and transform relevant features simultaneously.(3)The experimental results on the ORL,CMU PIE,Extended Yale B, AR,COIL-20,LFW and our real-world crop leaf disease datasets demonstrate that the two feature extraction methods proposed in this paper can effectively extract the discriminative features of images.