Rotary Face Recognition Based on Pseudo-Zernike Moment
With regard to the catastrophe problem of the face image misalignment from random angle for rotating, this paper proposes a feature extraction method of the images based on pseudo-Zernike moment. First transform the image into polar coordinates, and then calculate the multi-stage pseudo-Zernike moment of the image. Next come normalization to remove the gray effect and selection of appropriate order of the moment. Finally the invariant image rotation feature vector can be obtained and the dimensionality, by LDA, can be reduced. Experiments prove that, on the ORL image database with rotation change from a random angle, the recognition rate of alignment is 89%, which, at the same time, has a strong robustness on the white noise.
Keywordspseudo-Zernike moment rotation invariant feature extraction face recognition
Unable to display preview. Download preview PDF.
- 2.Shan, S.-G., Gao, W., et al.: Curse of Mis-Alignment Problem in Face Recognition. Chinese Journal of Computers 05, 782–791 (2005)Google Scholar
- 3.Wu, T.-H., Zhou, C.-L.: Study on Face Detection Under Rotation in Image Plane and Facial Features Localization. Acta Electronica Sinica (09), 1714–1718 (2007)Google Scholar
- 4.Sun, X.-M., Su, F., et al.: An eye location method of big angle rotation face at normal illumination. Journal of Optoelectronics. Laser (1), 94–98 (2009)Google Scholar
- 10.Pang, Y.H., Teoh, A.B.J., Ngo, D.C.L.: A discriminant pseudo-Zernike moments in face recognition. Journal of Research and Practice in Information Technology 38(2), 197–211 (2006)Google Scholar