为了快速准确识别地物、设计野外路线并减少踏勘后对前期解译工作的修改,本文参考Google Earth软件提供的高分辨率遥感图像,利用ETM+解译生成训练样本,然后采用最大似然监督分类算法进行ETM+图像分类.结果表明:与非监督分类和非监督-监督混合分类方法相比,基于Google Earth高分辨率遥感图像的ETM最大似然监督分类方法效果好、精度高,是一种经济、高效的技术手段,可用于初步识别地物分布情况、设计野外路线和勘查点等工作,对野外工作具有一定的指导意义；不同融合方式、不同波段组合的图像分类结果明显不同,该区域ETM+图像R(5)G(4)B(3)波段组合、PCA融合图像的分类总精度最好.%Through referring the high - resolution remote sensing images provided by Google Earth, the training samples were generated by the manual interpretation of the Landsat ETM+ images. The samples were used to conduct ETM+ image classification by using the maximum likelihood supervised classification algorithm. The results showed that; in comparison with the methods of non -supervised classification and unsupervised - supervision mixed classification, the ETM maximum likelihood supervised classification method based on Google Earth high - resolution remote sensing images worked well with high precision, which was an economical and efficient technical means. It could be used to roughly identify the distribution of surface feature, and to design field routes and exploration points, which had a certain guiding significance on field work. The classification results of different fusion methods and different band combinations of images were significantly different. The overall classification accuracy of ETM+ images with R(5)G(4)B(3) band combination and PCA fused image in this region was the best.
展开▼
