摘要
现有的植物识别方法容易出现过拟合现象,导致识别率过低。为此,笔者提出基于多源信息融合的植物识别方法,通过对植物图像进行预处理,划分训练集和数据集,采用卷积神经网络(CNN)和灰度共生矩阵(GLCM)的识别结果作为信息源输入DS证据理论推理,以获得更好的识别效果。最后,将本文方法与融合前两个信息源单独的识别率进行对比,发现本文提出的方法能够提高识别率。
The existing plant recognition methods are prone to over fitting,which leads to low recognition rate.For this reason,the author proposes a plant recognition method based on multi-source information fusion,which divides the training set and the data set by preprocessing the plant image,and uses the recognition results of convolutional neural network(CNN)and gray level cooccurrence matrix(GLCM)as the information source to input the DS evidence theory inference,so as to obtain better recognition effect.Finally,the method proposed in this paper is compared with the recognition rate of the two information sources before fusion,and it is found that the method proposed in this paper can improve the recognition rate.
作者
付波
刘合琛
赵熙临
马霁旻
Fu Bo;Liu Hechen;Zhao Xilin;Ma Jimin(School of Electrical and Electronic Engineering,Hubei University of Technology,Wuhan Hubei 430068,China)
出处
《信息与电脑》
2020年第9期135-136,共2页
Information & Computer
关键词
植物识别
多源信息融合
卷积神经网络
灰度共生矩阵
D-S证据理论
plant recognition
multi-source information fusion
convolutional neural network
gray level co-occurrence matrix
D-S evidence theory
