Nowadays face recognition has much importance in surveillance systems and human in computer interaction. It is difficult to recognize face under varying lighting conditions. This can be done by combining the strengths of robust illumination normalization, local texture based face representations, kernel based feature extraction and multiple fusion. There are three methods (LBP, LTP, GABOR WAVELETS) are proposed here for recognition of input image. Input images at different lighting conditions, are first preprocessed and then LBP, LTB are calculated. For improving Robustness kernel PCA is used which makes use two complimentary sources that are Gabor wavelets gucci borse imitazioni and LBP. Here compare the output from Gabor wavelet selection for the test image with database images and the image can be verified / recognized or not. This method provides good accuracy face recognition results.