Wooden Surface Classification based on Haralick and The Neural Networks

Texture describes the pattern of any surface. The texture analysis is concerned with the understanding of any object surface. The texture analysis has wide application in computer vision and pattern recognition. Haralick features describe the different gray scale features. For the wooden surface, each wooden surface has its own unique pattern. In this paper, a system is proposed to identify the wooden surface using the textural features (Haralick features) presented in its surface. This system was designed based on feature extraction and by correlating the features of those wooden and non-wooden surfaces for their classification. A multi-layer feed forward neural network was then trained with fast-back propagation to identify wooden and non-wooden surface. The proposed system was able to classify the wooden and non-wooden texture up to 94.10% accuracy. The obtained result showed a high rate of classification accuracy using very 13 features comparing to the most related work.