基于深度学习的茶叶状态智能识别方法  

Intelligent Identification for Tea State Based on Deep Learning

作　　者：王琨[1] 刘大茂[2] 

机构地区：[1]福州大学阳光学院电子信息工程系,福州350015 [2]福州大学物理与信息工程学院,福州350108

出　　处：《重庆理工大学学报（自然科学）》2015年第12期120-126,共7页Journal of Chongqing University of Technology：Natural Science

基　　金：福建省教育厅项目(JA13366)

摘　　要：目前对茶叶状态的识别主要依赖人工完成,规模化、产业化生产制造较为困难。作为机器学习的一个新兴方向,深度学习以其准确性和高效性得到广泛关注。将深度学习中的卷积神经网络算法应用于茶叶状态识别中,根据实际的应用场合对经典的卷积神经网络算法进一步优化,利用提取到的茶叶图像集训练网络,最终使网络能够正确识别茶叶状态。实验结果表明:该算法能够对茶叶状态的判别标志——茶叶嫩芽进行有效的3D识别,且对于不同的影响因素具有一定的鲁棒性。Currently the identification of tea state is still done manually, and it is difficult to form a large-scale and industrial production. As a new direction in machine learning, deep learning is getting more attention for its accuracy and efficiency. So the Convolutional Neural Network of deep learning was applied to tea state identification. According to the practical applications, some improvements was made for classical Convolutional Neural Network, using the extracted tea image to train the network, and finally the network can identify the tea state correctly. Experimental results show that the algorithm can make 3 D identification for the symbol of tea-tea buds effectively and has robustness for different factors.

关 键 词：深度学习 卷积神经网络 图像分割 特征提取 3D识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]