Fuzzy data management on pores arrangement for tropical wood species recognition system

A human-decision based classification wood recognition system is designed to classify 52 tropical wood species. The system is designed based on visual inspection of the wood anatomy textures, which can actually be presented as image data processing using statistical parameters representing the texture and the grey values. There are thousands of wood images being processed in the wood database. In order to overcome the large processing time needed to process the large wood database and the nonlinearity problems of the wood texture, an efficient fuzzy data management technique is proposed. The fuzzy data management technique is implemented based on the size and quantity of pores on each wood image which mimics the human interpretation on wood features. Finally, a multilayer feedforward neural network is used to classify the wood species. This paper involves comparison of the system´s performance with and without the implementation of fuzzy data management. The results show that the inclusion of the fuzzy data management has improved the classification accuracy by approximately 4.0% and reduce the processing time for training and testing.