基于最大类间方差法和数学形态学的遥感图像潮沟提取方法

发布时间：2018-04-26 17:14

本文选题：潮沟 + 九段沙　；参考：《上海海洋大学学报》2017年01期

【摘要】：潮沟作为潮滩的主要地貌类型之一,以长江口九段沙为研究对象,利用2015年2月15日获取的Landsat 8分辨率为15 m的全色波段遥感数据为数据源,选取了3条发育不同的潮沟。首先利用顶帽变换来消除直接利用最大类间方差法对图像亮度背景不均匀不能准确分割的问题,然后通过最大类间方差法找到一个最佳的阈值使潮沟和背景之间方差最大,得到二值化图像;接着通过形态膨胀对断裂的潮沟进行连接,并用形态去除方法剔除非目标;最后对潮沟进行骨架化提取和去除短枝处理,得到完整的潮沟信息骨架图。采用视觉分析和定量分析对提取的潮沟信息进行精度评价。结果表明,最大类间方差法和数学形态学的结合对潮沟信息提取有较好的效果,平均准确度达到93.0%,遗漏误差和冗余误差分别为7.0%和0.5%。
[Abstract]:Tidal trench is one of the main landform types of tidal flat. Taking Jiuduansha in the Changjiang Estuary as the research object, using the panchromatic remote sensing data of Landsat 8 with a resolution of 15 m obtained on February 15, 2015 as the data source, three tidal trenches with different development are selected. First, the top cap transform is used to eliminate the problem that the maximum inter-class variance method is directly used to segment inaccurately the brightness background of the image, and then the maximum inter-cluster variance method is used to find an optimal threshold to make the variance between the tidal trench and the background maximum. The binary image is obtained; then the tidal trench is connected by morphological expansion, and the non-target is eliminated by the method of morphological removal. Finally, the complete information skeleton map of the tidal channel is obtained by skeleton extraction and short branch removal. Visual analysis and quantitative analysis were used to evaluate the accuracy of the extracted tidal channel information. The results showed that the maximum inter-class variance method combined with mathematical morphology had a good effect on tidal channel information extraction, with an average accuracy of 93.0, omission error of 7.0% and redundancy error of 0.5%.
【作者单位】：上海海洋大学海洋科学学院;远洋渔业协同创新中心;
【基金】：国土资源部公益性行业科研专项(201211009)
【分类号】：TP751

【相似文献】