基于单时相Landsat8_OLI影像的棉花提取技术研究

针对利用单时相遥感影像快速、高精度地提取棉花信息研究较少的现状,根据作物的物候历,选取棉花生育期的单时相Landsat8_OLI影像。通过对棉花和其他地物的光谱特征与植被指数的差异性分析表明:近红外波段(NIR)和差值植被指数(DVI)是棉花识别的特征指数;并由此构建了OLI数据的棉花信息提取的特征指数模型。经实地验证,运用此模型提取的库尔勒市棉花信息的生产精度为97.96%,用户精度为98.63%。基于单时相遥感影像的棉花提取思路可以为我国主要棉产区的棉花信息遥感提取、长势监测和产量估测提供新思路和技术参考。

Study on Cotton Information Extraction Based on Single-temporal Landsat8_OLI image

In order to establish a rapid and accurate cotton information extraction method using single-temporal Landsat8_OLI image, this paper took Korla city as an example in Xinjiang Uygur Autonomous Region, used Landsat8_OLI satellite image on August 28, 2013 according to the cotton phenological calendar. We analyzed the difference of spectral characteristics and vegetation index between cotton and other main land cover types, and established a model of cotton information extraction based on the near infrared band(NIR) and difference vegetation index(DVI), then the model was used to extract cotton information in Korla city. The results of statistical accuracy assessment suggested that the model was effective with production accuracy is 97.96% and user accuracy is 98.63%. The results show that the process of the cotton identification method in this paper can provide a technological support for understanding cotton planting situation, monitoring cotton growing condition and estimating the yield of cotton in other cotton-producing areas.