基于单时相双极化TerraSAR-X数据在高原山区烟草的识别

Recognition of Tobacco in Plateau Mountain Based on Singe-Temporal and Dual Polarized TerraSAR-X Data

为了探讨多极化雷达影像在贵州高原山区农作物上的识别应用,采用单时相双极化Terra SAR-X数据展开烟草识别研究。选择贵州省清镇市流长乡烟草示范基地为研究区,在烟草生长期内获取了TerraSAR-X数据,通过对图像进行预处理并分析典型地物的后向散射系数,建立地物训练场,采用最大似然法进行分类,同时利用GPS采集的样方数据对烟草识别进行精度验证。结果表明,其识别精度可达80.52%,为贵州高原山区的典型农作物监测提供了理论基础与技术支撑。

To explore the application of multi-polarization radar image in crops recognition in plateau mountain of Guizhou province, the single phase dual polarized TerraSAR-X data were used to recognize the tobacco. Taking the Liuchang, Qingzhen City of Guizhou Province as research area, the TerraSAR-X data were obtained during the growing period of tobacco. Tile ground training field was established by preprocessing the image and analyzing the backscattering coefficient of the typical objects. The accuracy of the tobacco recognition was verified by maximum likelihood method and quadrat data acquisition based on GPS. Results showed that the accuracy of recognition was 80.52%. It will provide theoretical basis and techni- cal support for monitoring crops in plateau mountains of Guizhou province.