Wood Recognition System Using Neural Network

Saiful Nizam , Mohd Padzil (2008) Wood Recognition System Using Neural Network. Project Report. UTeM, Melaka,Malaysia. (Submitted)

	PDF (24 pages) Wood_Recognition_System_Using_Neural_Network_Saiful_Nizam_Bin_Mohd_Padzil_QA76.87.S24_2008_-_24_pages.pdf - Submitted Version Download (508kB)
	PDF (Full Text) Wood_Recognition_System_Using_Neural_Network_Saiful_Nizam_Bin_Mohd_Padzil_QA76.87.S24_2008.pdf - Submitted Version Restricted to Registered users only Download (2MB)

Abstract

Neural networks, as used in artificial intelligence, have traditionally been viewed as simplified models of neural processing in the brain. Automated visual inspection has been employed in various industries for decades to replace human dependent job with intelligent machines. Artificial intelligence techniques have been incorporated with image processing tools in the design of automated visual inspection systems. In this thesis, an automatic visual inspection system for recognition of tropical wood species based on neural network system has been proposed. In order to strengthen Malaysia's role as top exporter in wood products, wood recognition is needed to keep exporting wood products runs smoothly. Currently, there have been cases of wood species being misclassified, which cause problems to the wood industry. A solution is proposed to use the wood anatomy as a way to identify the species of the timber which in this thesis, the extract data base used in this project is provided by Central of Artificial Intelligence and robotic (CAIRO). The unique anatomy of the woods acts as a 'fingerprint' to each wood species. The objective of this research is to design a computer system that able to recognize several types of wood using neural network to replace the laborious manual task of wood identification performed by human. A multi-layer neural network based on the popular feed-forward algorithm has been used for classification. Besides using the conventional FF Neural network, the Multilayer network is also implemented in the simulations for further analysis on the efficiency of the network. The results obtained show a high rate of recognition accuracy proving that the techniques used are suitable to be implemented for commercial purposes.

Actions (login required)

View Item