Premium
A novel geometric fuzzy membership functions for mouth and eye brows to recognize emotions
Author(s) -
Priya R. Vishnu,
Bharat Rawal
Publication year - 2019
Publication title -
concurrency and computation: practice and experience
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.309
H-Index - 67
eISSN - 1532-0634
pISSN - 1532-0626
DOI - 10.1002/cpe.5610
Subject(s) - vagueness , quadrilateral , face (sociological concept) , artificial intelligence , fuzzy logic , mathematics , measure (data warehouse) , computer science , pattern recognition (psychology) , data mining , finite element method , social science , physics , sociology , thermodynamics
Summary In this article, the novel geometrical fuzzy membership functions are proposed to accurately recognize the emotions. The four corner features are taken from both the eyebrows and mouth regions of the expressive static image without considering reference face. The quadrilateral shape is defined from the features and it failed to match with the basic geometric shapes. This quadrilateral is given to the proposed novel fuzzy membership functions to calculate the degree of impreciseness and vagueness exists in the defined quadrilateral. As the result, the 12 fuzzy features are obtained and evaluated by the various classifiers. Both CK and JAFFE datasets are used in the experiment to prove the performance of the proposed method compared to some of the recent methods. It is observed the proposed method gives the highest accuracy rate of 98.4% and 99.15% against contemporary methods. The proposed work achieves the statistical measures of 97.8% of sensitivity, 99.2% of specificity, 97.9% of positive predictive value, and 97.8% of F‐measure. Moreover, the proposed work achieves encouraging results compared to the recent Google API using publicly available datasets. This achievement is due to the fact of exactly identified the impreciseness and vagueness in the defined quadrilateral. The proposed approach avoids reference face and uses the least number of fuzzy features compared to state‐of‐the‐art approaches that reduces the pre‐processing, computational, and learning times.