An optional scale segmentation method based on scale comparison in object-oriented remote sensing classification

Multi-scale image segmentation technique is a key technology of object-oriented remote sensing image information extraction, among which the segmentation scale is particularly important. In accordance with object-oriented analysis methods, this paper presents the optimal scale selection method of scale comparison and conducts the experiments. Experiments show that this method can quickly and accurately obtain the target features in the optimal segmentation scale, which effectively improved the credibility of the segmentation results.