Research on feature extraction from remote sensing image

Abstract

Remote sensing image as a digital product, one of its important significance lies in quickly extracting information and applying the information to practice. However, in the process of extracting information, one of the basic problems is to conduct feature extraction. In this paper, the point and line features of remote sensing image are extracted, and the comparison of a variety of feature extraction methods is also completed. On one hand, the point feature is extracted by Movarec operator, Forstner operator, SUSAN operator and Harris operator, etc; on the other hand, the line feature is extracted by Sobel operator, LOG operator and Canny operator, etc. As well as, a comparative analysis among the distribution of feature extraction, the amount of feature points and the time of extraction is completed. Finally, a matching is carried on for corner point extraction methods, ERDAS software is used to assess the accuracy of extraction methods and achieves good results.