Abstract: In this paper, a novel image projection analysis method is developed for image feature extraction Compared with the Liu’s projection analysis method, the proposed method has a desirable property, i e , the projective features vectors are mutual uncorrelated What’s more, the proposed method is directly based on image matrices That is to say, it need not to convert the image matrix into high dimensional image vector like the existing image vector based linear discriminant methods Thus, much computational time will be saved if the method is used for feature extraction Finally, the proposed method is tested on ORL and NUST603 face databases The experimental results indicate that the proposed method is more powerful than Liu’s, and a recognition accuracy of 95 5% on ORL and 99 4% on NUST603 are achieved with ordinary classifiers The experimental results also show that the proposed method outperforms Fisherfaces and its speed for feature extraction is nearly 15 times faster than Fisherfaces Moreover, the experiments also demonstrate that the proposed method is robust in the uncontrolled lighting condition