The efficiency of crop recognition on ENVISAT ASAR images in two growing seasons
The efficiency of crop recognition on ENVISAT ASAR images in two growing seasons
复制标题
DOI：
10.1109/tgrs.2006.864380
复制
发表时间：
2006-03
影响因子：
8.2
通讯作者：
K. Stankiewicz
中科院分区：
文献类型：
--
作者：
K. Stankiewicz
本项目的目的是评估基于从ENVISAT-1获取的微波先进合成孔径雷达(ASAR)图像进行作物识别的效率。调查是在2003年和2004年连续两个生长季进行的。所选季节的农业气象条件差异很大，这导致了作物冠层相关特征的年际变化。利用多时相ASAR交变偏振图像序列进行作物分类。使用每年单独训练的神经网络分类器进行分类。在波兰西部进行的实地观察为分类器的培训、验证和测试提供了数据集。尽管在两个数据集上注意到分类器性能上的一些差异，但2003年和2004年获得的结果显示出高度的相互一致性。
The aim of the presented project was to assess the efficiency of crop recognition based on microwave Advanced Synthetic Aperture Radar (ASAR) images acquired from ENVISAT-1. Investigations were conducted during two consecutive growing seasons, in 2003 and 2004. The agrometeorological conditions during the selected seasons differed markedly, which induced year-to-year variations regarding the relevant characteristics of crop canopy. Multitemporal series of ASAR alternating polarization images were used for crop differentiation. Classification was performed using a neural network classifier trained separately for each year. Field observations conducted in the western part of Poland supplied datasets for training, validation, and testing of the classifier. Despite some differences noted in the classifier performance on two datasets, the results obtained for 2003 and 2004 showed high mutual consistency.