One of the important steps that must be considered in developing a robust facial recognition is feature extraction. The rate of recognition in the facebased biometric system can be determined by the amount of measurable and relevant features extracted from the face image. Several feature extraction algorithms in appearance-based technique such as Linear Discriminant Analysis (LDA), Independent Analysis (LDA) and Principal Component Analysis (PCA) have been used in face recognition. This paper applied Contrast Limited Adaptive Histogram Equalization (CLAHE) before three appearancebased feature extraction algorithms: PCA, LDA and combined PCA/LDA for face recognition system. A comparative analysis was conducted on the three techniques, experimental results showed that the PCA technique recorded the best recognition accuracy (RA) of 95.65% for ORL database, the best False Rejection Rate (FRR) of 0.1250 in LDA for FERET database and the best False Acceptance Rate (FAR) of 0.5000 in PCA / LDA for FERET database.