基于Radarsat和TM图像融合与分类的土壤盐渍化信息遥感监测研究

单一雷达影像数据通常不能提供足够的用以监测干旱地区盐渍化的信息。雷达图像与TM图像的融合可以提高遥感数据的利用率,增强数据的可靠性和信息的互补性,有助于提高分类精度。本文采用了GramSchmidt变换融合法将Radarsat和TM图像进行融合,并将该融合方法与一些常用融合方法(HIS融合、PCA融合、Brovey融合)进行了比较,结果表明该融合方法优于其他方法。最后采用支持向量机分类法(SVM)对Radarsat、TM融合后的图像进行了分类。结果表明:同单独Radarsat影像和TM影像分类结果相比,该融合分类法将分类精度分别提高了近30%和2%。因此该融合分类法更适合于遥感图像盐渍化信息监测。

Remote sensing monitoring of soil salinization based on fusion and classification of Radarsat and TM image

Usually it is hard to achieve enough information for monitoring the salinization information of arid area by using sole Radar image.The fusion of Radar and TM image could improve the utilization of remote sensing information,enhance the reliability and mutual-complementing of information,and be helpful for improving classification accuracy.This paper adopted Gram-Schmidt fusion,carried on fusion of Radarsat and TM images,and then compared this fusion method with some other common methods such as HIS,PCA and Brovey.The result indicated that this method was superior to others.Furthermore,in this paper Support Vector Machine(SVM) classification method was accepted,then Radarsat,TM and the fusion image were classified.It can be inferred from the result that,in comparison with the solely Radarsat image or the TM image,the fusion image had higher classification accuracy.It had as much as 30% and 2% accuracy than those of single Radarsat and TM images respectively.Therefore this fusion and classification method could be more applicable for monitoring of arid area salinization by remote sensing images.