论文部分内容阅读
目的针对目前区域分割算法获取的区域边界与真实地物边界不一致问题,利用高分辨率遥感影像地物内具有均质性和地物间边缘信息突出的特点,提出一种融合边界信息的高分辨率遥感影像分割优化算法。方法首先采用Canny算法对遥感影像进行边缘提取并进行边缘连接处理,产生闭合边界;然后将边界与初始分割结果进行融合处理,获得新的分割结果;最后在闭合边界约束下,基于灰度相似性准则对新的分割结果进行区域合并,获得优化后的最终分割结果。结果采用本文提出的分割优化算法对Mean Shift算法和eCognition软件获得的分割结果进行优化处理,优化后的分割结果与初始分割结果相比正确分割率(RR)平均提高了4%,验证了本文算法的有效性。结论该优化算法适用性广,可优化基于区域、基于边界和基于聚类等多种分割方法,同时该算法既能保持高分辨率遥感影像分割的区域完整性,又能保持地物边缘细节特征,提高了分割精度。
Aim at the inconsistency between the boundary of region and the boundary of real object obtained by the current segmentation algorithm, this paper proposes a high-resolution fusion boundary information based on the features of high-resolution remote sensing image with homogeneity and edge information highlighting. Rate Remote Sensing Image Segmentation Optimization Algorithm. Firstly, the edge of the remote sensing image is extracted by using the Canny algorithm and the edge connection is processed to generate the closed boundary. Secondly, the boundary and the initial segmentation result are fused to get the new segmentation result. Finally, based on the gray similarity The criterion is to merge the new segmentation result regionally and obtain the optimized final segmentation result. Results The segmentation algorithm proposed in this paper was used to optimize the segmentation results obtained by Mean Shift and eCognition. The optimized segmentation result showed an average improvement of 4% compared with the initial segmentation results, which showed that the algorithm proposed in this paper Effectiveness. Conclusion This algorithm has wide applicability and can optimize multiple segmentation methods such as region-based, boundary-based and cluster-based clustering. At the same time, this algorithm not only maintains the regional integrity of high-resolution remote sensing image segmentation but also preserves the edge detail features , Improve the segmentation accuracy.