| 人工神经网络技术在鲜茶叶分选中的应用 |
| |
| 引用本文： | 陈怡群,常春,肖宏儒,宋卫东,张佩.人工神经网络技术在鲜茶叶分选中的应用[J].计算机与农业,2010(7):37-40,43. |
| |
| 作者姓名： | 陈怡群 常春 肖宏儒 宋卫东 张佩 |
| |
| 作者单位： | [1]农业部南京农业机械化研究所,南京210014 [2]贵州省山地农业机械研究所,贵阳550002 |
| |
| 基金项目： | 贵州省“十一五”茶叶重大专项（黔科合重大专项字[2008]6015） |
| |
| 摘 要： | 将人工神经网络技术应用于鲜茶叶的分类,茶叶图像面积、周长、凸壳面积、凸壳周长、等二阶距椭圆长轴长度、短轴长度、椭圆偏心率等几何参数和R、G、B三个彩色空间分量的均值、标准偏差、平滑度和一致性等纹理参数可以作为茶叶分类的特征值。试验表明,BP网络用于茶叶分类能够取得较好的效果,分类判断的正确率达到90%。网络的隐藏层和输出层为多个神经元时,其可能达到的分类效果要略好于隐藏层和输出层只有单个神经元的网络,但前者训练出的网络会出现权值不能收敛到全局误差最小值的情况,其可靠性不如后者。
|
| 关 键 词： | 人工神经网络 图像识别 茶叶分选 特征值 |
Artificial Neural Networks Technology in the Fresh Tea Sorting |
| |
| CHEN Yiqun,CHANG Chun,XIAO Hongru,SONG Weidong,ZHANG Pei.Artificial Neural Networks Technology in the Fresh Tea Sorting[J].Computer and Agriculture,2010(7):37-40,43. |
| |
| Authors: | CHEN Yiqun CHANG Chun XIAO Hongru SONG Weidong ZHANG Pei |
| |
| Institution: | 1.Nanjing Research Institute for Agri.Mechanization,Nanjing 210014;2.Guizhou Research Institute of Upland Agri.Machinery,Guiyang 550002) |
| |
| Abstract: | In this paper,discusses the use of artificial neural networks technology in fresh tea classification study.Geometric parameters such as tea image area,perimeter,convex hull area,convex hull perimeter,major axis length the same second-order moment ellipse,minor axis length,eccentricity of ellipse,and texture parameters such as R,G,B three color space components of the mean,standard deviation,smoothness and consistency can be used as classifying characteristic values of tea.BP network for classification of tea can achieve good results.Tests showed that the correct classification rate is more than 90%.When hidden layer and output layer of the network is more than one neuron,it is possible that the classification results is slightly better than the hidden layer and output layer is only a single neuron network.But the former appears to train the network weights can not converge to the global error minimum,so its reliability is Relatively poorer than the latter. |
| |
| Keywords: | artificial neural networks image recognition tea sorting characteristic value |
| 本文献已被 维普 等数据库收录！ |
|
