Human Face Recognition System A Framework

J. K. Keche, Dr. Mahendra P. Dhore


Face recognition is one of the hot topics in biometrics which is used to identify the face image by matching the main features of face image. This paper discusses the feature extraction techniques for human face recognition with k-Nearest Neighbor classifier. The proposed framework uses hybrid feature extraction technique which is combination of PCA, LDA, and DWT. The features of face are generated using PCA, LDA, and wavelet family 'db1'. The test image features are compared with training face features using Euclidian Distance (ED). The experimentation is performed on ORL and Face94 databases. Here 200 images from two databases are taken and calculates the correct and wrong recognitions of human face. The results obtained for the proposed framework are promising and the system is able to achieve good performance in human face recognition.

Keywords: Face Recognition, PCA, LDA, DWT, k-NN, Euclidean Distance

Full Text:



  • There are currently no refbacks.