基于边缘与区域结合的光学遥感图像舰船分割方法
Method for Ship Segmentation in Optical Remote Sensing Images Based on Edge and Region Information
吴诗婳 1贺梦 2盛景恺3
作者信息
- 1. 中国电子科技集团公司第二十八研究所 南京 210023
- 2. 解放军95806部队 北京 100076
- 3. 解放军93145部队 南京 210000
- 折叠
摘要
针对光学遥感图像中舰船目标分割精度低等问题,提出了一种基于边缘与区域结合的图像中舰船目标分割方法.首先,利用导向滤波改进Canny 算法对待分割舰船遥感图像进行预处理,以保持边缘完整结构,得到舰船目标粗分割结果;然后,提出基于布谷鸟搜索优化的倒数灰度熵阈值选取方法进行舰船目标细分割;最后,引入形态学方法对分割结果进行修正,确保舰船目标的完整性.大量试验结果表明,该方法表现出较优的分割性能,为舰船检测识别领域提供了更好的分割方案.
Abstract
Aiming at problem of low segmentation accuracy for ship target in optical remote sensing images,a ship target segmentation method in images based on edge and region information is pro-posed.Firstly,the traditional Canny algorithm is improved with guided filtering,and the ship's re-mote sensing image waiting for segmentation is preprocessed to maintain the edge integrity of ship's structure.Thus,a coarse segmentation result is obtained.Then,threshold selection method for recip-rocal gray entropy based on optimized cuckoo search is presented to carry out fine ship target segmen-tation.Finally,mathematical morphology method is introduced,and the segmentation result is cor-rected.Thus,the complement of ship target is ensured.A large number of experimental results show that the method shows excellent segmentation performance,and it can provide a better segmentation scheme in domain of ship detection and recognition.
关键词
舰船分割/导向滤波/倒数灰度熵/阈值选取/布谷鸟搜索优化Key words
ship segmentation/guided filtering/reciprocal gray entropy/threshold selection/opti-mized cuckoo search(CS)引用本文复制引用
出版年
2024
NETL
NSTL
万方数据