基于多光谱影像的潮滩区域边界提取方法

Boundary Extraction Method of Tidal Flat Region Based on Multi-spectral Image

当前潮滩区域边界提取方法多采用特征选择算法,此方法易受非边界点的影响,无法有效区分区域边界与岸边地物,导致提取准确度不高。为此,该文提出基于多光谱影像的潮滩区域边界提取方法,该方法利用多光谱成像仪采集潮滩影像,根据影像的波段分布获取潮滩区域的点云数据;再对点云数据进行网格化处理,识别其中的边界点;利用滤波原理去除非边界点,通过计算区域格网的高程范围,实现潮滩区域边界提取。实验结果显示,该方法对不同潮滩区域边界的提取准确率均在95%以上,且提取时间均在5s内,有效实现了潮滩区域的边界提取,提高了准确度。

The current tidal flat region boundary extraction method mostly adopts feature selection algorithm,which is vulnerable to non-boundary points,and can not effectively distinguish the region boundary and shore features,resulting in low extraction accuracy.Therefore,a boundary extraction method for tidal flats based on multi-spectral images is proposed,which uses a multi-spectral imager to collect tidal flat images and obtains point cloud data in the tidal flat area based on the band distribution of the images.The point cloud data is gridded to identify the boundary point.The filtering principle is used to remove non boundary point,and the tidal flat regional boundary is extracted by calculating the elevation range of the regional grid.The experimental results show that the proposed method has an extraction accuracy of over 95% for different tidal flat area boundaries,and the extraction time is within 5 seconds,which effectively realizes the boudaries extraction of the tidal flat area and improves the accuracy.