SciELO - Scientific Electronic Library Online

 
vol.25 número43Análisis de señales acústicas de pulmón basado en coeficientes cepstrales de la escala Mel y mapas auto-organizadosEfecto de la infiltración de lluvia en la respuesta hidráulica y en los mecanismos de falla de modelos de taludes arenosos índice de autoresíndice de materiabúsqueda de artículos
Home Pagelista alfabética de revistas  

Servicios Personalizados

Revista

Articulo

Indicadores

Links relacionados

  • En proceso de indezaciónCitado por Google
  • No hay articulos similaresSimilares en SciELO
  • En proceso de indezaciónSimilares en Google

Compartir


Revista Facultad de Ingeniería

versión impresa ISSN 0121-1129

Resumen

BASTIDAS-RODRIGUEZ, Maria Ximena; PRIETO-ORTIZ, Flavio A.  y  ESPEJO-MORA, Édgar. Fractographic classification in metallic materials by using 3D processing and computer vision techniques. Rev. Fac. ing. [online]. 2016, vol.25, n.43, pp.83-96. ISSN 0121-1129.

Failure analysis aims at collecting information about how and why a failure is produced. The first step in this process is a visual inspection on the flaw surface that will reveal the features, marks, and texture, which characterize each type of fracture. This is generally carried out by personnel with no experience that usually lack the knowledge to do it. This paper proposes a classification method for three kinds of fractures in crystalline materials: brittle, fatigue, and ductile. The method uses 3D vision, and it is expected to support failure analysis. The features used in this work were: i) Haralick’s features and ii) the fractal dimension. These features were applied to 3D images obtained from a confocal laser scanning microscopy Zeiss LSM 700. For the classification, we evaluated two classifiers: Artificial Neural Networks and Support Vector Machine. The performance evaluation was made by extracting four marginal relations from the confusion matrix: accuracy, sensitivity, specificity, and precision, plus three evaluation methods: Receiver Operating Characteristic space, the Individual Classification Success Index, and the Jaccard’s coefficient. Despite the classification percentage obtained by an expert is better than the one obtained with the algorithm, the algorithm achieves a classification percentage near or exceeding the 60% accuracy for the analyzed failure modes. The results presented here provide a good approach to address future research on texture analysis using 3D data.

Palabras clave : Artificial Neural Network; brittle fracture; ductile fracture; fracture due to fatigue; Support Vector Machine; 3D data.

        · resumen en Español | Portugués     · texto en Inglés     · Inglés ( pdf )

 

Creative Commons License Todo el contenido de esta revista, excepto dónde está identificado, está bajo una Licencia Creative Commons