Classification Of Wood Defects Using Generalized Feed-Forward Neural Network

In this paper a new classification algorithm is proposed for the Wood Defects. In order to develop algorithm 158 different wood defect images With a view to extract features from the images after using matlab, an algorithm proposes (FFT) Fast Fourier Transform coefficients. The Efficient classifiers based on Generalized Feed-Forward Neural Networks (GFF NN). A separate Cross-Validation dataset is employed for correct evaluation of the proposed classification algorithm with reference to important performance measures, like MSE and classification accuracy. The Average Classification Accuracy of GFF Neural Network comprising of hidden layers1 with 50 PE’s organized in a typical system is found to be superior (97.5 %) for Training. Finally, optimal algorithm has been developed on the idea of the simplest classifier performance. The algorithm will provide an effective alternative to traditional method of wood defects analysis for deciding the best quality wood.