A robust Human Face Recognition algorithm for flexible situations using SIFT(International Forum on Medical Imaging in Asia 2009 (IFMIA 2009))

スポンサーリンク

概要

• 論文の詳細を見る

• There are many methods for face recognition such as Principal Component Analysis(PCA), Independent Component Analysis(ICA) and so on. Those methods use a kind of algorithm compressing information which is could be feature information of faces. Because of this, they are not flexible enough to be used variable situations such as face images with background. In this paper, a new technique for variable face images is developed using Scale Invariant Feature Transform(SIFT). Originally, SIFT is used to extract matching points from among two or more similar images. Nevertheless, it is hard to pull out matching points from face images because there are not many feature points in them. To solve this problem, new method analyzing Difference of Gaussian(DOG) of images into histogram is used. This paper shows higher rate of accuracy than presented method with erratic condition. AR face database is mainly used and Caltech Faces as well.

• 2009-01-12

著者

• CHOI Seokyoon

  Dept. of Medical Engineering, Korea University

• OH Jangseok

  Dept. of Electronics & Information Engineering, Korea University

• KIM Mingi

  Dept. of Electronics & Information Engineering, Korea University

• Kim Mingi

  Department Of Electronics & Information Engineering Korea University

• Oh Jangseok

  Department Of Electronics & Information Engineering Korea University

• Choi Seokyoon

  Department Of Electronics & Information Engineering Korea University

• CHOI Kwanghee

  Dept. of Electronics & Information Engineering, Korea University

• Choi Kwanghee

  Department Of Electronics & Information Engineering Korea University

関連論文

• A robust Human Face Recognition algorithm for flexible situations using SIFT(International Forum on Medical Imaging in Asia 2009 (IFMIA 2009))
• L1 minimization based EM algorithm for PET Reconstruction(International Forum on Medical Imaging in Asia 2009 (IFMIA 2009))
• Improvement of snake model for ventricle segmentation in MRI(International Forum on Medical Imaging in Asia 2009 (IFMIA 2009))
• A robust method of feature extraction from noised endoscopic images(International Forum on Medical Imaging in Asia 2009 (IFMIA 2009))

スポンサーリンク