Skip to Main Content
The availability of high-resolution (HR) remote sensing multispectral imagery brings opportunities and challenges for land cover classification. The methodology of multiscale segmentation is wildly accepted for feature extraction and classification in HR image. However, the relationship among chosen scale parameters, selected features, and classification accuracy is less considered. A classification approach combining the hierarchy segment algorithm and SVM is presented in this paper. Firstly, a family of nested image partitions with ascending region areas is constructed by iteratively merging procedure; Then, multiscale morphological features are extracted in every segmentation level; Finally, the classification accuracy in different scales are compared and analyzed. The experiments shown that a more conservative scale parameter benefits land cover classification algorithm and different land objects has different optimal scale for classification.