Fuzzy LDA for face recognition with GA based optimization

The paper addresses the face recognition problem by modifying the Fuzzy Fisherface classification method. In conventional methods, the relationship of each face to a class is assumed to be crisp. The Fuzzy Fisherface method introduces a gradual level of assignment of each face pattern to a class, using a membership grading based upon the K-Nearest Neighbor (KNN) algorithm. This method was further modified by incorporating the membership grade of each face pattern into the calculation of the between-class and with-in class scatter matrices, termed as Complete Fuzzy LDA (CFLDA). Both Fuzzy Fisherface and CFLDA methods utilize the Fuzzy-KNN algorithm. The present work aims at improving the assignment of class membership by improving the parameters of the membership functions. A genetic algorithm is employed to optimize these parameters by searching the parameter space. Furthermore, the genetic algorithm is used to find the optimal number of nearest neighbors to be considered during the training phase. The experiments were performed on the ORL (Olivetti Research Laboratory) face image database and the results show consistent improvement in the recognition rate when compared to the results from other techniques applied on the same database and reported in literature.