基于遥感图像的复杂种植区施肥均匀性监测  

Monitoring of Fertilization Uniformity in Complex Planting Areas Based on Remote Sensing Images

作　　者：杨中杰[1] 包永红[1] 王立中[1] YANG Zhong-jie;BAO Yong-hong;WANG Li-zhong(Inner Mongolia Agricultural University Department of Computer Technology and Information Managemen,Baotou Inner Mongolia 014109,China)

机构地区：[1]内蒙古农业大学计算机技术与信息管理系,内蒙古包头014109

出　　处：《计算机仿真》2022年第11期244-247,370,共5页Computer Simulation

摘　　要：利用目前方法监测复杂种植区施肥的均匀性时,忽略了对遥感光谱角度的匹配,导致施肥均匀性监测结果与实际数据相差较大问题。为此提出基于遥感图像的复杂种植区施肥均匀性监测方法。在光谱角度匹配的基础上提取复杂种植区的遥感图像。融合多光谱和全色数据,提取遥感图像中第一主成分单色图,通过分水岭分割提取超像素图像,构建特征组,采用随机森林法实现复杂种植区施肥的均匀性监测。仿真结果表明,所提方法的种植面积估算准确率高、种植位置识别精度高,且施肥均匀性与实际数据的一致性较理想。When monitoring the uniformity of fertilizer in complex plant areas,some methods ignore the matching of remote sensing spectral angle,resulting in a large difference between the monitoring results of fertilization uniformity and the actual data.Therefore,a method of monitoring the uniformity of fertilization in complex plant areas based on remote sensing image was put forward.Firstly,remote sensing images of complex planting areas were extracted based on spectral angle matching.After integrating multispectral and panchromatic data,the monochrome image of the first principal component in remote sensing images was extracted.Then,the watershed segmentation was used to extract super-pixel images and construct feature groups.Finally,the random forest method was adopted to monitor the uniformity of fertilizer in complex planting areas.The simulation results show that the proposed method has high accuracy in estimating planting area and identifying planting position;Meanwhile,the uniformity of fertilizer is in good agreement with actual data.

关 键 词：光谱角度匹配 植被面积 遥感图像 图像特征分类 随机森林分类法 

分 类 号：TP33[自动化与计算机技术—计算机系统结构]