Conference Paper

Clasificación de objetos rígidos a partir de imágenes digitales empleando los momentos invariantes de Hu

Conference: X Congreso Internacional sobre Innovación y Desarrollo Tecnológico. ISBN: 978-607-95255-4-5, At Cuernavaca, Morelos


The aim of this study was to develop software that allows, from one image, classifying objects in the image, based on size, shape and position of the object. In order to feature extraction of objects in the images the method Hu's moment invariants was used, which, as its name implies, are invariant to rotation, scaling and translation of objects. In the classification of these same objects the K-means classifier was used. The tests performed for the validation of this software were: numbers, letters, tools, screws, keys shapes and coins. The results of these tests are reported to a percentage of certainty of 97.76% and an error rate of 2.24%. According to the proposed results, the same accuracy variables can be applied in industry for quality control in medicine and classification of cancer cells; in the military as enemyartillery detection, among many other uses.

Download full-text


Available from: Jorge Ochoa Somuano,
805 Reads