In recently, there is a large increase in the development and research of automatic classification methods and systems especially in wood-rich countries. One of the main reasons for this increase is knots which are found in wood obtained from the trees. Allocated to the class according to the different types of knots by an expert is a huge waste of time and constitute failure. To eliminate this problem, classify the knots in the wood floorboards using artificial neural network in this study is aimed. As the first to do this, features of knot images with two-dimensional discrete wavelet method are extracted. Then, these features are classified with back propagation multilayer neural networks. Haar, Daubechies, Bior, Coif, Symlet type wavelet methods in feature extraction are used and effects on their classification are determined. Classification rate with used methods are tried to highest level, at the same time reviews are conducted in terms of process time by taking into importance of calculation time in the industry applications.
Journal Section | Articles |
---|---|
Authors | |
Publication Date | November 7, 2016 |
Published in Issue | Year 2016 Volume: 5 |
As of 2021, JNRS is licensed under a Creative Commons Attribution-NonCommercial 4.0 International Licence (CC BY-NC).