Identification of Remote Sensing Image of Adverse Geological Body Based on Classification

摘要

Identification of interested landmark is a hot topic in the field of remote sensing. Setting QuickBird as an example, this paper focuses on the typical adverse geological phenomenon, such as desert, saltmarsh, gobi, lakes, etc., in Yuli Rob Village of Xinjiang Province. Three classification methods, i.e., extreme learning machine, SVM algorithm, and K-means algorithm, are used for classification and recognition of remote sensing images. The image recognition rate and accuracy are analyzed. Experimental results and comparison analysis indicate that the extreme learning machine algorithm, SVM algorithm and K-means algorithm in general are not significant. The SVM algorithm for image continuity provides better results. The extreme learning machine obtains classification results, and it is easy to fall into local optimum.