Implementación de un sistema de visión artificial en android, para la detección y reconocimiento de barajas

Abstract

The present work consists in the development of a system for automated detection and recognition of playing cards used in the Ecuadorian game called “Cuarenta”, with the objective of supporting blind people for their inclusion in this social and leisure activity. This system consists of a mobile application, that interacts with a Web service which pre-processes detect and recognizes playing cards, and subsequently sends the recognition result for it to be reproduced in audio format by the mobile application. For the pre-processing of the image, techniques were used like: gray scale, Gaussian filter and thresholding. In card detection, a technique of edge search was used. For the recognition and classification was applied: central point search, image transformation to flattened perspective, segmentation, and template matching. In recognition stage two training sets were used, reaching results of up to 97,84% accuracy. Also, in the real situation test stage an accuracy of 98,33% was obtained.