Face Recognition Using Fuzzy Moments Discriminant Analysis
Hussein Attya Lafta, Farah Mohammed AL-Shareefi · Journals & Books Hosting (International Knowledge Sharing Platform) · 2014
In this work, an enhanced feature extraction method for holistic face recognition approach of gray intensity still image, namely Fuzzy Moment Discriminant Analysis is used. Which is first, based on Pseudo-Zernike Moments to extract dominant and significant features for each image of enrolled person, then the dimensionality of the moments features vectors is further reduced into discriminant moment features vectors using Linear Discriminant Analysis method, for these vectors the membership degrees in each class have been computed using Fuzzy K-Nearest Neighbor, after that, the membership degrees have been incorporated into the redefinition of the between-classes and within-classes scatter matrices to obtain final features vectors of known persons. The test image is then compared with the faces enrollment images so that the face which has the minimum Euclidean distance with the test image is labeled with the identity of that image. K eyword: Zernike Moments, LDA, Fuzzy K-Nearest Neighbor.