Inferior quality images often pose a major challenge in the domain of face recognition. This paper presents a scheme for face recognition which can not only work efficiently on standard databases but also on inferior quality images having low and medium resolutions. The proposed framework uses Local Gabor Phase Quantizers (LGPQ) to compensate for the quality of the images. In this framework, a probe face image is taken as input, which undergoes preprocessing for photometric corrections. The preprocessed image undergoes Gabor transformation, where Local Phase Quantizers are applied independently on each frame to obtain the signature histograms. These histograms are finally matched with the gallery image weights using Principal Component Analysis (PCA). The framework has been tested on images selected from the CMU, AR, Yale databases.