水稻农学理化参量无人机遥感反演研究现状与展望

Research Status and Prospect of Unmanned Aerial Vehicle Remote Sensing in Inversion of Rice Agronomic Physicochemical Parameters

近几年,无人机遥感技术在水稻理化参量反演上得到了广泛应用,并逐渐发展成为水稻田块尺度遥感信息获取的主要途径之一。深入分析基于无人机遥感的水稻农学理化参量(指在农业领域中可确定某种物理、化学性质的参量)反演研究现状及存在问题,有利于更好地把握水稻无人机遥感未来发展趋势。综述无人机遥感技术在反演生化组分含量、结构参量、生产力等方面的研究现状,其中生化组分含量的反演研究主要集中在氮素和叶绿素方向且目前仍然以数据驱动的方法为主,例如用于反演氮素的窄波段植被指数NDRE,通过对极限学习机与偏最小二乘回归耦合对水稻叶绿素含量的反演等,而基于物理模型的反演方法较少;结构参量的反演研究主要包括叶面积指数、生物量等,方法有用于反演叶面积指数的辐射传输机理模型PROSAIL,用于反演生物量的基于冠层光谱特征的优化高斯过程回归方法;生产力的遥感重点在水稻的产量估算、病害和倒伏检测,方法有用于水稻估算的利用RGB影像使用K-Means与核相关滤波算法融合。对无人机遥感平台、设备、方法进行了总结,梳理近10年水稻农学理化参量无人机遥感反演的研究进展和成果。最后综合国内外的研究现状进行水稻无人机定量遥感讨论分析与展望,以期为今后无人机遥感技术在水稻定量遥感研究中提供参考。

In recent years,Unmanned Aerial Vehicle(UAV)remote sensing technology has been widely used in rice physical and chemical parameter inversion,and has gradually developed into one of the main ways to obtain remote sensing information at paddy field block scale.In-depth analysis of the research status and existing problems of rice physical and chemical parameters inversion based on UAV remote sensing is conducive to better grasping the future development trend of rice UAV remote sensing.This article reviews the current research status of using drone remote sensing technology to invert biochemical component content,structural parameters,and productivity.The inversion research of biochemical component content is mainly focused on nitrogen and chlorophyll,and currently data-driven methods are still dominant,such as narrowband vegetation index NDRE used for nitrogen inversion,and the use of extreme learning machine and partial least squares regression for rice chlorophyll content inversion.There are fewer inversion methods based on physical models.The inversion research of structural parameters mainly includes leaf area index and biomass,and the methods include the radiation transfer mechanism model PROSAIL for inversion of leaf area index,and the optimized Gaussian process regression method based on canopy spectral characteristics for inversion of biomass.The key focus of remote sensing for productivity is the estimation of rice yield,disease and lodging detection,and the methods include the use of RGB images and the fusion of K-Means and kernel correlation filtering algorithms for rice yield estimation.The drone remote sensing platform,equipment,and methods are summarized,and the research progress and achievements of drone remote sensing inversion of rice agronomic and physical parameters in the past decade are sorted out.Finally,the research status is comprehensively analyzed and prospected for quantitative remote sensing of rice by drones,aiming to provide reference for future drone remote sensing technology in rice quantitative remote sensing research.