Feature Extraction Research On The Face Recognition Process

Feature extraction is an important issue in the classification of data with a large dimension. The purpose of feature extraction is to generate a set of features that have a smaller dimension than the dimension of the original data, while retaining the data characteristics sufficient to classify the data. These extracted features can reduce the computation for classification which is related to the number of features, and improve classification performance by removing irrelevant characteristics in a dataset. The major content of this paper is as follows:(1)Face recognition using wavelet transform and adaptive class-augmented PCA First of all, human face images are compressed using Discrete Wavelet Transform. Then the features of low-frequency component are extracted by the adaptive class-augmented PCA. Different from traditional class-augmented PCA, the method does not require constructing the between-class information, so it will be used more flexible. Besides, because of the wavelet transform, the recognition rate and time-consuming are both improved. Experiments show the effectiveness of the algorithm.(2)Face recognition algorithm fusing 2DPCA and fuzzy 2DLDAIn the beginning, 2DPCA is used to extract the optimal projective vectors from face image. Then, the membership degree matrix is calculated by fuzzy K-nearest neighbor, and merged into the process of 2DLDA.Finally,fuzzy between-class scatter matrix and fuzzy within-class scatter matrix can be obtained. Compared with (2D)~2PCALDA, the method makes full use of the advantages of (2D)~2PCALDA. It not only effectively extracts the row and column recognition information, but also makes full use of the distribution information of samples. Experiments show that the method can achieve better recognition effect.(3)Two-dimension class-augmented PCAFirstly, face features are obtained by 2DPCA and normalize them. Then combine these normalized features with the class information. Finally, process the above combined data with 2DPCA again to gain the goal feature. The method considers not only the structure of human faces, but also the class information of the training data, and it improves the recognition rata.