基于多维特征数据库的玉米长势自动监测车辆设计  

Design of Corn Growing Automatic Monitoring Vehicle Based on Multidimensional Feature Database

作　　者：罗元成[1] 汪应[1] 

机构地区：[1]重庆工程职业技术学院,重庆402260

出　　处：《农机化研究》2017年第11期212-216,共5页Journal of Agricultural Mechanization Research

基　　金：重庆市教育委员会重点项目(1202086)

摘　　要：为了提高农作物长势预测的精度和实时性,提出了一种新的基于双目立体视觉的玉米长势自动化监测车辆,并将图像多维重构技术引入到了车辆的设计中,采用自主导航技术在无需人员进入农田的情况下,实现了玉米长势的智能远程监控。为了解决玉米叶面积采集特征数据的冗余导致信息处理速度不高的问题,提出了改进的LPP的降维方法,并对算法进行了验证。测试结果表明:采用LPP算法,能够完成对作物多维特征信息的优化降维,具有较高的实用性和准确性。对玉米长势自动化监测车辆的性能进行了测试,对生物量的预测结果表明:采用监测车辆生物量反演模型得到的长势预测量和实测量的误差较小,从而验证了监测车辆设计的可行性。In order to improve crop growth forecast accuracy and real-time performance, a binocular stereo vision is based vehicle automatic monitoring of corn growing new, and the image reconstruction of multidimensional technology is introduced into the design of the vehicle. The autonomous navigation technology is without the need for workers to enter the farmland under the condition, which can realize remote monitoring of intelligent the growth of corn. In order to solve the problem of redundant data acquisition characteristics of maize leaf area which resulted in information processing speed is not high, the dimension reduction method of LPP is improved, and the algorithm is verified by the test. The results show that by using LPP algorithm to complete the optimization of crop multidimensional feature information for dimension- ality reduction, which has high practicality and accuracy. The performance of the vehicle automatic monitoring of corn growth was tested by the biomass prediction results show that the growth forecast and monitoring vehicle biomass inversion model and the measurement error is small, which verifies the feasibility of monitoring vehicle design.

关 键 词：多维数据库 玉米长势 监测车辆 LPP降维 生物量 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置] S311[自动化与计算机技术—控制科学与工程]