The previous neural network based on the proximity values was developed using rectangular pavement images. However, the proximity value derived from the rectangular image was biased towards transverse cracking. By sectioning the rectangular image into a set of square subimages, the neural network based on the proximity value became more robust and consistent in determining a crack type. This paper presents an improved neural network to determine a crack type from a pavement surface image based on square sub-images over the neural network trained using rectangular pavement images. The advantage of using square sub-image is demonstrated by using sample images of transverse cracking, longitudinal cracking and alligator cracking. © Springer Science+Business Media B.V. 2010.
CITATION STYLE
Lee, B. J., & Lee, H. (2010). Improved crack type classification neural network based on square sub-images of pavement surface. In Innovations in Computing Sciences and Software Engineering (pp. 607–610). Kluwer Academic Publishers. https://doi.org/10.1007/978-90-481-9112-3_105
Mendeley helps you to discover research relevant for your work.