基于无人机多光谱影像的棉花苗期株数提取方法
Extraction method of cotton seedlings number based on UAV multispectral images
郝帅帅 1李永可 1陈燕红1
作者信息
- 1. 新疆农业大学计算机与信息工程学院,乌鲁木齐 830052
- 折叠
摘要
准确、快速获取棉花苗期株数对棉花管理早期决策起着至关重要的作用.文中利用无人机遥感获取棉花苗期多光谱影像,通过植被指数NDVI剔除土壤、地膜等背景,再进行灰度运算、阈值分割、形态学变换等图像处理操作提取棉苗轮廓,并运用OpenCV图像识别计数方法得出棉苗数量,统计误差在0.23％～1.45％之间.结合地面调查,采用线性回归分析方法,建立无人机图像识别株数与实际株数的线性关系模型.结果表明,模型精度指标R2、RSME和nRSME分别为0.9865、1.6124和2.137％.该方法可为后续田间大面积测定出苗率和棉花高产的准确评估提供技术支持.
Abstract
Accurate and rapid acquisition of the number of cotton seedlings plays a vital role in the early de-cision-making of cotton management.This study uses UAV remote sensing to obtain multi-spectral images of cotton seedling stage,removes soil,plastic film and other background through vegetation index NDVI,ex-tracts cotton seedling contour through image processing operations such as gray scale operation,threshold segmentation,morphological transformation,and uses OpenCV image recognition and counting method to obtain cotton seedling quantity,with statistical error between 0.23％and 1.45％.Based on the ground sur-vey,the linear regression analysis method is used to establish the linear relationship model between the number of UAV image recognition plants and the actual number of plants.The results show that the model accuracy indexes R2,RSME and nRSME are 0.9865,1.6124 and 2.137％,respectively.This method can provide technical support for the subsequent determination of seedling emergence rate and accurate eval-uation of cotton high yield in large areas.
关键词
出苗率/图像识别/植被指数/线性关系模型/无人机Key words
emergence rate/image recognition/vegetation index/linear function relation model/UAV引用本文复制引用
基金项目
新疆维吾尔自治区重大科技专项(2020A01002_4)
出版年
2024
NETL
NSTL
万方数据