一种基于白帽变换的细小水体信息提取方法
A novel method to extract narrow water with white top-hat transform

提出一种基于形态学白帽变换的线性特征增强处理技术，从而有效提取遥感影像中的细小水体信息. 通过白帽变换对改进的归一化差异水体指数(MNDWI)影像进行线性特征增强处理，获取形态学细小水体指数(MNWI)特征影像，并采用经验阈值和自动阈值相结合的方式实现水体样本信息双阈值分割提取，解决细小水体与线状道路容易混淆的问题. 利用多幅Landsat影像测试本文方法，结果表明，本文方法能较好地提取细小水体信息，较好地保持细小河道形态的连续性，细小水体提取精度达到80%以上.
We proposed a novel methods to extract narrow water from remote sensing image effectively by enhancing line features with a morphological white top-hat transform(WHT) technique. This method obtain line feature enhanced morphological narrow water index (MNWI) feature image by doing a WTH transform on modified normalized difference water index (MNDWI) image，in order to further solve the problems of narrow water easy confused with linear road，we also proposed to adopt a dual-threshold segmentation combined with an experience threshold and an automatic threshold to segment the small and narrow water body from the background information. The effectiveness of the proposed method was validated with multiple Landsat images，and the results show that our method can not only extract narrow and small water body，but also keep the continuity of narrow water effectively，and the extraction accuracy of narrow water can be generally over 80%