Document Detail


Frequency study of facial electromyography signals with respect to emotion recognition.
MedLine Citation:
PMID:  24402883     Owner:  NLM     Status:  Publisher    
Abstract/OtherAbstract:
Abstract Emotional intelligence is one of the key research areas in human-computer interaction. This paper reports the development of an emotion recognition system using facial electromyogram (EMG) signals focusing the ambiguity on the frequency ranges used by different research works. The six emotional states (happiness, sadness, fear, surprise, disgust, and neutral) were elicited in 60 subjects using audio visual stimuli. Statistical features were extracted from the signals at high, medium, low, and very low frequency levels. They were then classified using four classifiers - naïve Bayes, regression tree, K-nearest neighbor, and fuzzy K-nearest neighbor, and the performance of the system at the different frequency levels were studied using three metrics, namely, % accuracy, sensitivity, and specificity. The post hoc tests in analysis of variance (ANOVA) indicate that the features contain significant emotional information at the very low-frequency range (<0.08 Hz). Similarly, the performance metrics of the classifiers also ensure better recognition rate at very low-frequency range. Though this range of frequency has not been used by researchers, the results of this work indicate that it should not be ignored. Further investigation of the very low frequency range to identify emotional information is still in progress.
Authors:
Jerritta Selvaraj; Murugappan Murugappan; Khairunizam Wan; Sazali Yaacob
Related Documents :
22979623 - Modulation difference limen for spectral center-of-gravity signals.
23505773 - Noise disturbances in daycare centers before and after acoustical treatment.
11039483 - Spatial attention: different mechanisms for central and peripheral temporal precues?
Publication Detail:
Type:  JOURNAL ARTICLE     Date:  2014-1-9
Journal Detail:
Title:  Biomedizinische Technik. Biomedical engineering     Volume:  -     ISSN:  1862-278X     ISO Abbreviation:  Biomed Tech (Berl)     Publication Date:  2014 Jan 
Date Detail:
Created Date:  2014-1-9     Completed Date:  -     Revised Date:  -    
Medline Journal Info:
Nlm Unique ID:  1262533     Medline TA:  Biomed Tech (Berl)     Country:  -    
Other Details:
Languages:  ENG     Pagination:  1-9     Citation Subset:  -    
Export Citation:
APA/MLA Format     Download EndNote     Download BibTex
MeSH Terms
Descriptor/Qualifier:

From MEDLINE®/PubMed®, a database of the U.S. National Library of Medicine


Previous Document:  Pathway to Clozapine Use: A Comparison Between a Patient Cohort from New Zealand and a Cohort from t...
Next Document:  Simulation Education Approaches to Enhance Collaborative Healthcare: A Best Practices Review.