Sie befinden Sich nicht im Netzwerk der Universität Paderborn. Der Zugriff auf elektronische Ressourcen ist gegebenenfalls nur via VPN oder Shibboleth (DFN-AAI) möglich. mehr Informationen...
Automatic texture characterization using Gabor filters and neurofuzzy computing
Ist Teil von
International journal of advanced manufacturing technology, 2011, Vol.52 (1-4), p.15-32
Ort / Verlag
London: Springer-Verlag
Erscheinungsjahr
2011
Quelle
Alma/SFX Local Collection
Beschreibungen/Notizen
Natural manufactured materials are highly heterogeneous. The remarkable variability exhibited from one sample to another makes it impossible to find two samples with the same morphological distribution in their texture. This fact makes automatic quality measurement of the properties of naturally textured objects a desirable though challenging problem in industrial applications where costs are paramount. Cork, as a natural and heterogeneous material with fully seven quality classes, defies easy automatic quality classification. This paper describes an improved computer vision system for measuring the quality of industrial cork samples by incorporating techniques based on the biological configuration of human vision and human reasoning, such as Gabor filtering techniques and neurofuzzy computing. These techniques have provided important texture detection improvements over previous results. Neurofuzzy classification system performance has proved to be better than results obtained with other classification algorithms. The proposed neurofuzzy classifier includes in its feature space three wavelet-based texture quality features. The results obtained from the final proposed system showed an error rate of only 4.85% that is lower than previous similar systems designed for this application. The system proposed in this paper has the potential to reduce costs, time, and current conflicts in the cork industry.