Reconocimiento de patrones mioeléctricos para la interpretación de un conjunto de letras del alfabeto dactilológico icónico

Abstract

In the present work, a technological advance was developed guide towards the development and human growth of the community in condition of hearing impairment. To contribute a grain of sand towards this objective, a device was proposed for the recognition and interpretation of a set of dactylological alphabet gestures. Then, a database of myoelectric signals associated with the interpretation of dactylological alphabet gestures A, O, M and R executed by 20 people was constituted. Subsequently, the descriptive characteristics of the myoelectric signals were extracted using Multiresolution Analysis. Finally, in order to maximize the performance in the classification of four gestures of the Colombian dactylological alphabet, three supervised learning algorithms were used to tune the parameters of a classification system based on fuzzy logic. As a result, it was possible to reveal representative patterns that allowed to recognize to a certain extent; the link between the myoelectric signals associated with the extension and flexion of phalanges of the hand and the dactylological alphabet gesture that the person interpreted.