论文部分内容阅读
遥感技术由于具有观测范围广、实时强等特点适合用来研究土壤盐渍化现象。利用遥感手段提取盐渍土信息已经取得了一定的成效。利用面向对象方法,以TM卫星图像数据和野外实地数据为数据源进行提取盐渍地信息。首先,对遥感影像进行预处理,预处理包括几何校正和辐射校正,然后对图像进行图像分割,图像分割使用了分割方法的多尺度分割法、特征选择、面向对象分类和分类图像进行精度评价。对面向对象方法和传统的基于像元分类(最大似然法和最小距离法)结果进行对比分析。结果表明:利用面向对象方法对TM遥感图像进行分类,能有效抑制“椒盐现象”的发生,分类精度比传统的分类方法更高,为盐渍地信息的自动提取提供了广阔的前景。
Because of its wide range of observation and strong real-time, remote sensing technology is suitable for studying soil salinization. The use of remote sensing to extract salty soil information has achieved some success. Using the object-oriented method, TM satellite image data and field field data were used as data sources to extract saline soil information. First of all, preprocessing of remote sensing images is carried out. The preprocessing includes geometric correction and radiation correction. Then the image is segmented. The segmentation method is used for multi-scale segmentation, feature selection, object-oriented classification and classification of images for accuracy evaluation. The object-oriented method and the traditional pixel-based classification (maximum likelihood method and minimum distance method) results for comparative analysis. The results showed that using TM to classify TM remote sensing images can effectively suppress the phenomenon of “salt and pepper”, and the classification accuracy is higher than traditional classification methods, which provides a broad prospect for the automatic extraction of salty ground information.