Abstract: In order to improve the low rate of detecting the front side of a bamboo in chip in the bamboo automatic detection field, the paper present an algorithm of identifying the front side of the bamboo chip based on BP neural network. A model of BP neural network wa built with three layers of three input and one output. The operation flow is as follow: extract the front and reverse side’s textural features and characteristic value of gray value, then import these data into the trained model, and at last calculate the predicting results. The experiment results show that the algorithm can identify 96% of the reverse side and 95% of the front side under the principle of minimum risk. Therefore, the stability of the algorithm is stronger than any other method with signal feature.

Key words: the front and reverse side of bamboo, neural network, textural features, characteristic value of gray value