| 摘要: |
| 【目的】小麦倒伏是造成产量和质量下降并且影响农业机械自动化收割的重要原因。 基于无人机遥感平台获得的数据能够为小麦倒伏提供及时准确的监测结果，为智慧农业智能 化育种、栽培和管理提供科学依据。【方法】基于无人机搭载的消费级相机，获取可见光影 像结合生成的DSM数据，采用随机森林的分类方法进行小麦倒伏识别。【结果】基于可见光 和DSM数据获得的小麦倒伏分类总体精度为98.41%，Kappa系数为0.97，相较于仅依靠可 见光谱信息的分类结果具有显著提升，识别结果也更加可靠，能够显著改善小麦倒伏识别效 果。【结论】证明了基于无人机搭载消费级相机获取的DSM数据在农作物倒伏识别中的可行 性，提供了一种自动识别小麦倒伏的新思路和新方法。 |
| 关键词: 倒伏 DSM 无人机遥感 小麦 随机森林 |
| DOI：10.12105/j.issn.1672-0423.20190404 |
| 分类号: |
| 基金项目:国家重点研发计划“粮食作物监测诊断与精准栽培技术”子课题“基于无人机的玉米生长参数实时监测”（2016YFD0300602） |
|
| Wheat lodging identification using DSM by drone |
|
Zhao Licheng, Duan Yulin, Shi Yun, Zhang Baohui※
|
|
Institute of Agricultural Resources and Regional Planning，Chinese Academy of Agricultural Sciences/Key Laboratory of Agricultural Remote Sensing，Ministry of Agriculture and Rural Affairs，Beijing 100081，China
|
| Abstract: |
| ［ Purpose］Wheat lodging is an important cause for yield and quality. It affects the automated harvesting of agricultural machinery. The data obtained from UAV platform can provide timely and accurate monitoring results for wheat lodging，and scientific basis for intelligent breeding，cultivation and management of smart agriculture and digital agriculture. ［Method］Using the consumer-grade camera equipped with drone，the DSM generated by the combination of visible light images was obtained，and the random forest classification method was used to identify wheat lodging. ［Result］The results showed that the overall accuracy of wheat lodging classification using visible RGB and DSM data was 98.41%，and the Kappa coefficient was 0.97. Comparing to the classification results using visible spectrum alone，the recognition results were more reliable and precise，improving wheat lodging recognition. ［Conclusion］The research proves the feasibility of crop lodging recognition using DSM acquired by UAV equipped with consumer cameras，and proposes a new idea and method to automatically identify wheat lodging. |
| Key words: lodging DSM UAV wheat Random Forest |
