基于神经网络的珍稀红木图像识别技术

Recognition of Rare Wood Based on the Neural Network

木材识别是木材交易和珍稀植物保工作中的一项重要技术。该文提出了一种木材图像识别的方法,首先利用30X放大镜获取木材横切面的图像,设计了一个5层的CNN神经网络,对12种木材的材种进行识别。实验中训练数据的分类正确率达到了80%,验证数据的分类正确率最高达到了70%。

Wood species recognition is important in the trade activities and the protection of rare plants.This article proposed a classifi cation method using the wood images,first,acquire the cross-section image of the wood with 30X amplify-glass,second,classify 12 wood species with a neural network which using the CNN method.The result shows that the accuracy of the training dataset reached 80%,the accuracy of the validation dataset reached 70%.