Wood microscopic image identification method based on convolution neural network

To prevent the illegal trade of precious wood in circulation, a species identification method based on convolutional neural network (CNN), namely PWoodIDNet (Precise Wood Specifications Identification) model, is proposed. In this paper, model for rare tree constructed to reduce parameters by decomposing kernel, overfitting, enrich diversity features, and improve performance model. The results showed that can effectively generalization ability, characterization ability detail recognition accuracy, classification identification. was used analyze accuracy microscopic images 16 kinds wood, reached 99%, which higher than several existing classical models. addition, analyzed verify feasibility effectiveness as method, provide new direction technical solution field