Classification of human face images using principal components
Zahid Riaz, Khurram Shehzad Khattak · 2006
This paper describes a novel approach of using principal components analysis (PCA) for recognizing the human faces in an efficient manner. The idea builds the concept of eigenfaces. With minimal additional effort PCA provides a roadmap for how to reduce a complex data set to a lower dimension to reveal the sometimes hidden, simplified dynamics that are underlying. Also the results obtained in this paper are quite acceptable at the initial level. The experiments have been performed on the fifteen subjects (150 compressed and normal images each) of ORL-datasets producing 91% accuracy in case of normal while 87.39% in case of compressed images. Discrete wavelet transform (DWT) is used to compress images for classification. Euclidean distance is used to classify the principal components of the face images. For real time system, it is suggested to use this technique as preprocessing step for near ideal results. In the end a comparison is performed between the two classifications and the other statistical techniques