LSM static signs recognition using image processing

Abstract

Object recognition is a widely field in artificial vision application, because now the machines are intended to become autonomous. This article presents the methodology for recognizing objects in an image, tecniques used are: segmentation, feature extraction and classification object within the image. Fuzzy c-means algorithm was used for segmentation, which is a fuzzy classification algorithm in which a data can belong to multiple groups in different degree of membership. For feature extraction were used Hu moments as a geometrical descriptors, which are a mathematical tool that provides seven moments that identify geometric features of the objects, main characteristics of Hu moments is its invariance to rotation, scaling and translation. Finally, the geometric features that provide the seven moments are used as input to a classifier, delivering results: these moments are used to identify the signs of the alphabet of the Mexican Sign Language (LSM).