In this paper, we combine two moments of Hu and Zernike to try improve the recognition accuracy. And we apply PCA on the Zernike moments to speedup of recognition without degrading the recognition accuracy. We compare the recognition performance between Hu and Zernike moments. Also we show the recognition accuracy of new methods with two modified features. We evaluate the recognition accuracy with two datasets of hands and digits. For hand recognition, we localize hands from hand candidates with improved method against our preliminary study [2].