农田活体害虫动态特征获取系统研究  

Acquisition System and Research on Dynamic Characteristics of Live Pests In Farmland

作　　者：钱晶 聂余满 曹平国 王勇平 QIAN Jing;NIE Yu-man;CAO Ping-guo;WANG Yong-ping(Institute of Intelligent Machines,Chinese Academy of Sciences,Hefei Anhui 230031,China)

机构地区：[1]中国科学院合肥智能机械研究所

出　　处：《计算机仿真》2019年第5期403-406,共4页Computer Simulation

基　　金：基于压缩感知的多比特核磁共振退相干参数层析研究(61403362)

摘　　要：针对传统害虫监测系统中由于害虫虫体破损造成的监测误差问题,提出了一种基于机器视觉技术获取农田活体害虫动态特征的方法。首先通过相机获取目标害虫的原始图像信息,而后利用目标颜色特征与阈值化方法将监测目标与背景图像分割,使用高斯滤波法获取害虫形状特征并记数,由害虫形状特征随时间产生的变化量得到害虫的运动量,最终实现对活体害虫动态信息的监测。实验结果表明,活体害虫的运动量信息可为智能农业中的适时适量施药提供依据。Aiming at the error of traditional monitoring system caused by pest body damage.A method for obtaining the dynamic characteristics of pests in farmland based on machine vision was presented.First,pest original image information was acquired with camera.Then,the pests and the background images wese segmented by color feature and thresholding methods.The Gaussian filter method was used to obtain pest shape features and count.The quantity of pest motion was obtained based on frame-to-frame differencing method.Finally,monitoring of dynamic information on live pests was achieved.The experimental results show that the information of the amount of physical activity of pests can provide the basis for timely and proper application of pesticides in intelligent agriculture.

关 键 词：图像分割 害虫识别 特征获取 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]