针对泰安市遥感影像地物分类方法比较.docxVIP

摘要 遥感影像的分类识别简单来说就是根据不同地物光谱特征的差异在遥感影像上分类并判读出具体地物，从而将地物的实际分布转换为可被计算机应用的影像数据。 传统方法进行分类是运用某种具体方法基于光谱特征进行地物识别分类识别，由于单一方法的准确性有限，对于不同对象的差异性明显，很难达到较高的精度。本文研究的具体内容是应用泰安市lansat8遥感影像数据，对其进行多种分类方法分类。针对泰安地区将所有地物类别分为五种地物类别。通过精度对比结果发现最小距离法的精度最高，达到64.4934% ，但是对于河流水域分类精度只有33.60%。支持向量机分类方法虽然总体分类精度较低，但是对河流水域的分类精度达到了最高的78.48 %。为了使各种方法进行优势互补，在此进行单一分类器分类后使用多分类器并行的方式分类，分类结果显示多分类器融合后各地物分类精度都有了明显提高。通过不断提高分类精度，可以协调区域内发展，对区域变化进行科学预测。 关键字：遥感影像分类、ENVI；分类精度；多分类器融合 ABSTRACT The classification and recognition of remote sensing images is simply to classify and interpret specific ground objects on remote sensing images according to the differences of spectral characteristics of different ground objects, so as to convert the actual distribution of ground objects into image data that can be applied by computer. The traditional method is to use a specific method to classify and recognize ground objects based on spectral features. Because of the limited accuracy of a single method, it is difficult to achieve high accuracy because of the obvious differences between different objects. The concrete content of this paper is to use lansat8 remote sensing image data of Taian City to classify them by various classification methods. According to Taian area, all the ground objects are divided into five categories. The results of precision comparison show that the minimum distance method has the highest accuracy of 64.4934%, but the classification accuracy of river waters is only 33.60%. Support vector machine classification method although The overall classification accuracy is low, but the classification accuracy of river waters is the highest 78.48%. In order to complement the advantages of various methods, multiple classifiers are used to classify them in parallel after a single classifier. the classification results show that the classification accuracy of each ground object is improved obviously after the fusion of multiple classifiers. By continuously improving the classification accuracy, we can coordinat