Finger Braille is one of the tactual communication media of deafblind people. Deafblind people who are skilled in Finger Braille can catch up with speech conversation and express various emotions. The objective of this study is development of a Finger Braille supporting device which assists not only verbal communication but also non-verbal (emotional) communication between deafblind people and non-disabled people. In this paper, to develop emotion recognition system, we analyzed features of emotional expression (neutral, joy, sadness and anger) and derived an algorithm of emotion recognition using accelerometers worn by receiver. The selected features were the duration of dotting and amplitude of acceleration by dotting. The classifier was canonical discriminant analysis (CDA). A sentence was classified into the emotion for which the mean of posterior probabilities of dotting in the sentence was highest. The results of the evaluation experiment showed that the algorithm of emotion recognition was practicable and independent of sentences

Resumen ralizado por los autores