论文部分内容阅读
棉花是中国重要的经济作物,快速、准确地提取棉花的种植面积和分布信息,对于优化棉花种植空间格局、科学指导棉花生产及提高其管理水平具有十分重要的意义。为了探讨多时相中高分辨率影像在棉花种植面积监测中的可行性,本文以江汉平原为研究区,根据棉花物候特征,选取2012年、2014年江汉平原棉花生长关键期的多时相HJ-1A/1B卫星数据,通过分析研究区棉花不同生育期的光谱特征和归一化植被指数(NDVI)时序变化特征,对分类影像进行阈值分割、掩膜处理,最后利用决策树算法提取研究区2012年、2014年棉花种植面积。通过计算混淆矩阵评价分类精度的方法和提取面积精度方法对棉花提取结果进行评价,总体精度达到95.96%,Kappa系数为0.93,以农业局统计数据为参考,2012年、2014年HJ数据提取的棉花种植面积精度分别达到了97.91%、91.27%。因此,在不受云和降水等因素的影响下,基于江汉平原区域关键时相HJ卫星CCD影像数据,可利用该方法进行棉花种植面积监测。
Cotton is an important cash crop in China. The rapid and accurate extraction of cotton acreage and distribution information is very important for optimizing the spatial pattern of cotton planting, scientifically guiding cotton production and improving its management level. In order to investigate the feasibility of multi-temporal high and medium resolution images in the monitoring of cotton acreage, this paper took Jianghan Plain as the study area and selected the HJ-1A / 1B satellite data, the spectral features of the different growth stages of cotton were analyzed and the temporal variation characteristics of normalized vegetation index (NDVI) were analyzed to classify the images for threshold segmentation and mask processing. Finally, the study area was extracted by using the decision tree algorithm. In 2012, 2014 cotton acreage. The accuracy of the method was evaluated by calculating the confusion matrix and the precision of the extraction area was used to evaluate the cotton extraction results. The overall accuracy was 95.96% and the Kappa coefficient was 0.93. Based on the statistics from the Agricultural Bureau, the cotton extracted from the HJ data of 2012 and 2014 The acreage precision reached 97.91% and 91.27% respectively. Therefore, under the influence of cloud and precipitation and other factors, based on the HJ satellite CCD image data of key phases in the Jianghan Plain, this method can be used to monitor cotton acreage.