Construcción de un prototipo para el reconocimiento y traducción del lenguaje de señas a texto utilizando el sensor kinect.

On the following research work it is introduced a system the language of Ecuadorian signs specially the alphabet this language is based on automatic learning techniques, to facilitate in a way, the communication of people who suffer from a kind of hearing impairment for those who do not know this language. In order to meet the objective a bunch of ideas and various authors investigations were combined in order to define the procedure of which this prototype is, it consists of four phases: hand detection, obtaining the descriptive characteristics of the same, classification and interpretation of the information obtained in previous stages. The data acquired by Kinect depth were stored in a database so then it can be processed, adapting it to a fixed size, and transforming them into binary images so that the computational cost is as low as possible. The training data is applied for child employee for the Training of Machine Learning called Support Vector Machine (SVM) after feature extraction through descriptor Histograms Oriented Gradients (HOG), once the detector is obtained images may be applied to identify hands and perform the translation accordingly. The translation can be visualized in real time through a guide made in Matlab which shows the sign made by the user, resizing it and finally the translation corresponding to the signal originally issued.

• Κύριος συγγραφέας: Mingo Morocho, Leydi Maribel (author)
• Μορφή: bachelorThesis
• Γλώσσα: spa
• Έκδοση: 2016
• Θέματα: <MÁQUINAS DE SOPORTE VECTORIAL (SVM)><HISTOGRAMA DEGRADIENTES ORIENTADOS (HOG)>; <SENSOR KINECT><LENGUAJE DACTILOLÓGICO><MATLAB>
• Διαθέσιμο Online: http://dspace.unl.edu.ec/jspui/handle/123456789/10907