基于均值漂移的海岛(礁)岸线快速提取

为了解决海岛(礁)岸线提取的自动化问题,提高海岛(礁)岸线提取的速度与精度,提出了基于均值漂移的海岛(礁)岸线快速提取方法。首先在保证影像整体轮廓的同时构建低分辨率影像,从而缩短算法的运行时间;然后利用均值漂移技术进行高斯低通滤波处理,提高算法的抗噪性;最后通过区域增长法自动选择种子点,结合区域合并差异性度量准则实现对遥感影像的形状和颜色的识别,从而完成对海岸线的自动提取。使用该方法对海岛(礁)影像进行了海岛(礁)岸线提取实验,实验结果表明,该算法能够实现所有岛礁岸线的自动提取,相比于通过人工方式,提取时间从729s缩短为3s,极大地缩短岸线提取的时间,同时能较好地满足成图效果。

Rapid abstraction of island or reef shoreline based on mean shift method

To improve the automation degree of the island (reef) coastline extraction and increase its accuracy and speed, a rapid extraction process was proposed based on the mean shift technique. Firstly, a low-resolution image was built while ensuring the overall outline of the image, thereby reducing the running time of the algorithm. Then, Gaussian low-pass filtering was performed based on the mean shift technique to improve the noise immunity of the algorithm. Finally, seeds were automatically selected by region growing method, and the shape and color of remote sensing images were recognized by using regional diversity measure merger guidelines, thus completing the automatic extraction of shoreline. Experimental results of island(reef) shoreline extraction show that the algorithm can automatically extract the all reef coastline, and that the extraction time is decreased from 729 s to 3 s compared with the traditional artificial extracting way.