基于数学形态学的海岛水边线提取模型

针对遥感影像数据大面积覆盖的特性,利用数学形态学提出了一种多目标海岛水边线的同步提取模型,模型包括:利用K-means算法的海岛区域粗分割;基于灰度膨胀算法的海岛初始轮廓曲线确立;利用水平集算法的海岛水边线优化。以福州海域的多目标海岛水边线提取为例,将本文模型与K-means模型、二值化高斯滤波水平集模型(SBGFRLS)和基于改进水平集的海岛边界快速分割模型(ILSM-IBRS)进行了比较。结果表明,本文模型弥补了K-means模型的过分割现象,其计算耗时较SBGFRLS和ILSM-IBRS模型分别节省了96.16%和86.96%;其迭代次数分别减少了97.07%和90.59%。由此可见,本文模型解决了现有提取模型效率低、过分割等现象,实现了多目标海岛水边线的同步提取,为海岛普查等提供了一种快速提取方法。

Extraction Model for Island Waterlines Using Mathematical Morphology

For the large-area coverage of remote sensing image,a multi-island waterlines synchronous extraction model is proposed by mathematical morphology.The model includes tree aspects: (1)Coarse segmentation of islands are extracted based on K-means algorithm; (2)Initial curves of islands are optimized by the gray-scale dilations algorithm;and (3)Island waterline is optimized based on the level set algorithm.Taking the island waterline extraction of Fuzhou sea area as an example,the improved extraction model is compared with K-means model,selective binary and Gaussian filtering regularized level set model(SBGFRLS)and improved level set model for island boundary rapid segmentation(ILSM-IBRS).The result shows that,the segmentation accuracy of the improved extraction model proposed makes up for the oversegmentation of the K-means model.Compared with the SBGFRLS and ILSM-IBRS models,its calculation time is saved by 96.16% and 86.96% respectively; the number of iterations is reduced by 97.07% and 90.59% respectively.In short,the proposed model overcomes the low efficiency and over-segmentation of the traditional segmentation methods,realizes the synchronous extraction of multi-island waterlines,and presents a rapid method for island census.