针对高分辨率遥感影像几何结构和纹理信息更加明显的特点,提出了一种多规则的面向对象提取方法。改进了基于边缘检测的影像分割方法,建立影像的高、低尺度空间,分别在不同尺度空间对影像进行边缘检测的分割,二者的结果相互约束调整合并,很好地抑制了过分割现象。利用分割得到的影像像对,设计特定的分类规则集,对采石场用地信息进行提取技术应用。总体分类精度达到83.1645%。
For the characteristics of geometric structure and texture information in high resolution remote sensing image, a new method of multi-rules of object-oriented is proposed. The image classification method based on edge detection is improved, and the image is segmented by the high and low scale space, and the results of the two are combined with each other. By using image pairs, the multi-rules are designed to classify the information of the quarry, and the overall classification accuracy is 83.1645%.