论文部分内容阅读
以塔里木盆地北缘绿洲——渭干河-库车河三角洲绿洲为例,借助ENVI遥感软件,利用ETM+数据,探讨了该绿洲土壤盐渍化信息提取的方法。传统的遥感图像分类方法多数在解决问题上存在精度不高、分类效率较低、不确定性强的缺陷,所以,选择好的分类方法对于提取盐渍化信息是至关重要的。近年来,将SVM应用于遥感图像分类已成为新的发展趋势。文章提出了基于纹理特征的支持向量机(Support Vector Machine,SVM)的分类方法,得出以下结论:分别结合3×3,5×5,7×7,9×9,11×11,13×13窗口纹理特征和光谱的SVM分类精度都很高,达到93%以上。并且在验证分类精度时,发现结合光谱和9×9窗口纹理信息的SVM分类的结果更符合实际情况。所以说加入纹理特征后使得光谱信息比较接近的3类地物(重度、中度、轻度盐渍地)的区分性增大,从而使精度提高。因此,基于纹理特征的SVM分类方法更有利于遥感图像分类和盐渍化信息监测,是地物遥感信息提取的有效途径。
Taking the oasis - Weigan River - Kuqa River delta oasis in the northern margin of Tarim Basin as an example, this paper discusses the method of soil salinization information extraction using ENVI remote sensing software and ETM + data. Most traditional remote sensing image classification methods have some disadvantages such as low precision, low classification efficiency and high uncertainty in solving the problem. Therefore, the selection of good classification methods is crucial for extracting salinization information. In recent years, the application of SVM to remote sensing image classification has become a new trend. In this paper, the classification method of Support Vector Machine (SVM) based on texture features is proposed. The following conclusions are drawn: the combination of 3 × 3, 5 × 5, 7 × 7, 9 × 9, 11 × 11, 13 × 13 Window texture features and spectral SVM classification accuracy are high, reaching more than 93%. And in verifying the classification accuracy, it is found that the result of the SVM classification combining the spectrum and the 9 × 9 window texture information is more realistic. Therefore, after the texture features are added, the distinction of the three kinds of ground objects (severe, moderate and mild salted areas) with the closer spectral information increases, thereby improving the accuracy. Therefore, the SVM classification method based on texture features is more conducive to remote sensing image classification and salinization information monitoring, which is an effective way to extract remote sensing information of land objects.