Multi-feature-Based Type-2 Fuzzy Set Induced Parallel Rank-Level Fusion in Face Recognition

Abstract

AbstractThis paper projects a new decision-making model for multi-feature and multi-classifier fusion by integrating the type-2 fuzzy set-based approach to mitigate the factors that pretend the face recognition accuracy. A new scheme is expended in the present research to generate a type-2 fuzzy set-based ranks for each of the top-ordered classes, which are selected on the basis of weights or confidence factors, yielded by the classifier. These type-2 fuzzy sets are shaped for each of the top-ordered classes from the feature vectors of the test image and those of the training samples of the respective class. Type-reduced defuzzified value is the conceived as the fuzzy rank of the particular class. Considering the three different feature vectors and the respective outputs of the classifier, the fuzzy ranks are fused weighted by the confidence factors for each of the regarded classes to generate the fused scores. Based on these scores, the class of the test image is recognized. Well-known databases are used to evaluate the performance of the proposed model. The experimental results have manifested an improved performance than other state-of-the-arts methods despite varied occlusion, pose, and illumination present in the databases.KeywordsRank-level fusionFace recognitionType-2 fuzzy set