M.R. Chandraratne, S. Samarasinghe, D. Kulasiri, C. Frampton, and R. Bickerstaffe (New Zea
Image analysis, Artificial neural networks, Co-occurrencetexture features, Discriminant analysis, Computer vision,Lamb grading
This paper presents a lamb carcass classification system
based on texture analysis and artificial neural networks
(ANNs). Digital images of lamb chops were used to
calculate twelve image geometric variables. In addition, a
set of ninety texture features was used to extract the
textural information. Texture analysis is based on the co
occurrence matrix method.
Principal component analysis (PCA) was used to reduce
the dimensionality of feature spaces. The reduced feature
space, with six geometric variables and eight co
occurrence texture variables, was used for the analysis.
Three-, four- and five-layer multilayer perceptron (MLP)
networks and Discriminant function analysis (DFA) were
performed on the data. The classification accuracy from
three-layer MLP (93.1%) was 14% better than that from
DFA. This study shows the potential of neural networks
combined with texture analysis for lamb grading.