基于无人机遥感影像的玉米冠层温度提取及作物水分胁迫监测

针对当前无人机热红外遥感提取冠层温度不准确、监测作物水分胁迫状况精度不高的问题，该研究以不同水分处理的拔节期夏玉米为研究对象，利用无人机获取试验区域热红外和可见光图像资料，分别采用Otsu算法、EXG-Kmeans算法和Otsu-EXG-Kmeans算法获取冠层区域图像，并对提取结果进行精度评价，而后采用最优算法求得对应作物水分胁迫指数（Crop Water Stress Index，CWSI），通过分析CWSI同土壤含水率相关关系以及CWSI日平均变化趋势来监测玉米水分亏缺状况。结果表明：1）相比于其他方法，Otsu-EXG-Kmeans算法对冠层温度提取精度更高（用户精度为95.9%），提取的冠层温度更接近实测温度（r=0.788），可以准确获取图像冠层温度。2）相比于冠层温度，CWSI与土壤含水率的相关性更高（r= -0.738），CWSI日平均变化趋势更符合实际情况，可更加精确地监测玉米缺水状况。该研究为无人机遥感精准监测作物水分胁迫状况提供参考。

Temperature extraction of maize canopy and crop water stress monitoring based on UAV remote sensing images

Unmanned aerial vehicles (UAVs) with thermal infrared remote sensing have been used to rapidly extract the temperature of the crop canopy, further to monitor the water stress condition of the crop. The removal of soil background from the thermal infrared images can be an effective way to improve the monitoring accuracy of crop moisture. But there is also a great challenge on the thermal infrared image processing. In this study, an accurate and rapid temperature extraction was implemented to monitor the water stress using the UAVs with remote sensing images. The summer maize at the jointing stage was selected as the research object. There were also four completely irrigation treatments with three replicates and a total of 12 experimental plots. Specifically, the four water treatments were: T1 (50% of the field capacity), T2 (65% of the field capacity), T3 (80% of the field capacity), and T4 (95%-100% of the field capacity). The thermal infrared and visible images were captured by the UAVs on July 26, July 27, July 29, July 31, and August 2, 2021. The data of canopy temperature and soil water content at 20cm depth were also collected at the same time. After that, the Otsu algorithm, EXG-Kmeans, and Otsu-EXG-Kmeans were selected to determine the canopy area in the images. The accuracy of the extraction was evaluated from four aspects, including the schematics of canopy area, classification accuracy, temperature histogram, and correlation with the measured canopy temperature. An optimal combination of parameters was achieved to extract the thermal infrared images of canopy temperature. Subsequently, a crop water stress index (CWSI) was also achieved during this time. The relationship between CWSI and soil water content was then established to monitor the water deficit of maize, according to the daily average trend of CWSI. The results showed that the Otsu-EXG-Kmeans presented a higher extraction accuracy for the canopy temperature (User''s accuracy: 95.9%>83.2%>66.6%), and the extracted canopy temperature was closer to the measured temperature (r:0.788>0.762>0.750>0.737), indicating that the Otsu-EXG-Kmeans was an effective way to accurately extract the canopy temperature. The CWSI presented a higher correlation with the soil water content (r:-0.738<-0.666), compared with the canopy temperature. The daily average variation trend of CWSI was more consistent with the actual situation. Consequently, the improved extraction can be widely expected to extract the canopy temperature, thereby accurately monitoring the water shortage situation of maize. This finding can provide a strong reference for the UAVs to accurately monitor the crop water stress under remote sensing.