| Titre : |
To automate the rating of paint blister |
| Type de document : |
texte imprimé |
| Auteurs : |
S. Swarnamala, Auteur ; T. S. N. Sankara Narayanan, Auteur |
| Année de publication : |
2001 |
| Article en page(s) : |
p. 16-21 |
| Note générale : |
Bibliogr. |
| Langues : |
Anglais (eng) |
| Résumé : |
The present work aims to automate paint blister failure rating using image processing and neural network techniques. Photographic images of paint blisters after ASTM D 714-87 are acquired and processed using image processing techniques to extract the relevant features, specifically, the paint blister contours. The blister contours are used to estimate the feature histogram which is subsequently used as input to a two-stage trained neural network (Backpropagation Network) to assign the closest rating of the paint blister failure. Accuracy exceeding 70% is obtained in most of the cases. |
| Note de contenu : |
- Paint failure rating system
- Human eye can pick the relevant information from a scene and discard the remaining
- For an automated system it is important that the size/distribution factors be quantified
- The size-spread histogram determines the category the individual blisters belong
- Multilayer-preceptron with backpropagation learning rule is used for classification and prediction tasks
- Two stages of network : the 'Size Net' and the 'Dist Net'
|
| En ligne : |
https://drive.google.com/file/d/1_NcqxeOljeJS0TtOBAukIUPIxeZ7dJIm/view?usp=drive [...] |
| Format de la ressource électronique : |
Pdf |
| Permalink : |
https://e-campus.itech.fr/pmb/opac_css/index.php?lvl=notice_display&id=10939 |
in EUROPEAN COATINGS JOURNAL (ECJ) > N° 9/01 (09/2001) . - p. 16-21
Aller sur edunet
Accueil
To automate the rating of paint blister in EUROPEAN COATINGS JOURNAL (ECJ), N° 9/01 (09/2001)
