Resumen

VILLA, Betsy; VALENCIA, Valeria  y  BERRIO, Julie. Digital image processing applied on static sign language recognition system. Prospect. [online]. 2018, vol.16, n.2, pp.41-48. ISSN 1692-8261.  https://doi.org/10.15665/rp.v16i2.1488.

Sign language L.S.C. (Lengua de Señas Colombiana) is a native language which chiefly uses manual communication to convey meaning. This can involve simultaneously combining hand shapes, movement and orientation of the hands, arms or body, and even facial expressions to convey a speaker’s ideas. Currently in Colombia, there is an absence of technology focus on teaching and interpreting this language; for this reason, it’s interesting and a social commitment to e carry out initiatives that promote life quality improvement for the country’s deaf-mute population. In this article, the design and implementation process of a static hand gesture recognition system is shown, for this task we used Matlab as computing environment and the Scale Invariant Features Transform (SIFT) method to extract characteristics from the image. Our system allows the acquired image visualization and its corresponding translation to the colombian sign language. Through key points identification and their comparison with SIFT features in the system data base makes possible to retrieve the translation. The system can recognize 20 static letters from the Colombian Sign Languages, a graphical interface was implemented in Matlab that provides better visualization, simple access to the system and high usability. A better response of the system is noticed when a standardized element of the image is used, in our case, a surgical glove. As future work, we propose to apply neural networks to improvement of the tool, and a real time implementation, which can generate a greater impact for the current needs of the colombian population.

Palabras clave : Colombian Sign Language; Matlab; Image processing; Gesture recognition; Static recognition; Image segmentation; SIFT.

        · resumen en Español     · texto en Español     · Español ( pdf )