基于决策树模型分类多时相早稻信息提取研究

Study on Information Extraction of Multi-temporal Early Rice Based on Decision Tree Model Classification

决策树模型分类算法为早稻遥感分布信息的获取提供技术方法。研究直接利用归一化植被指数计算,配合卫星影像纹理分析,逐步剔除非耕地的信息,得到耕地的初步分类结果;再结合多时相归一化植被指数和地表水分指数曲线等综合判析,进一步区分出早稻和其他作物。通过相关资料的验证,早稻的用户精度达到了87.50%,生产者精度为85.71%,总体精度达到了93.50%,Kappa系数为0.823,证实该方法在提取早稻信息的可靠性、准确性,也为后期水稻产量的预估提供科学决策依据。

Decision tree model classification algorithm provides technical basis for obtaining remote sensing distribution of early rice.In this study,NDVI calculation and satellite image texture analysis were used to gradually eliminate the information of pseudo-paddy fields,and the preliminary classification results of cultivated land were obtained.Combined with multi-temporal normalized vegetation index NDVI and surface water index LSWI curve,early rice and other crops were further distinguished.Through the verification of relevant data,the user accuracy of early rice is 87.50%,the producer accuracy is 85.71%,the overall accuracy is 93.50%,and the Kappa coefficient is 0.837.It proves that this method is feasible to improve the accuracy of information extraction of early rice,and also provides scientific decision-making basis for the prediction of late rice yield.