摘要： 针对图割算法实时性差的问题，提出一种基于四叉树与图割的遥感图像分割方法。用四叉树分割方法将遥感图像分割成多个区域，根据区域的邻接性合并相似小区域，结合区域间的灰度和位置信息，利用Normalized Cut方法在区域间进行划分，完成图像的分割。实验结果证明该方法能有效消除四叉树方法的过分割现象。

关键词: 图像分割, 四叉树, 区域合并, 图割

Abstract: Aiming at the problem of poor real-time ability of graph cut, this paper suggests a remote sensing image segmentation approach based on quarter-tree and graph cut. The approach uses quarter-tree to segment the remote sensing image into a large number of small partitions, with the adjacency of each partition, smaller and similar partitions are merged into some areas. By integrating gray feature and spatial location of each partition, normalized cutting approach is used to segment the image among partition from global view, by which the final segmented image can be generated. Experimental results show that the over-segmentation can be eliminated effectively.

Key words: image segmentation, quarter-tree, region merger, graph cut

中图分类号: 

• TP391.41