LAUSR

A novel water extraction method is proposed to improve the ability of delineation of the narrow water for the modified normalized difference water index (MNDWI) derived from remotely sensed images. This method introduces a morphological white top-hat transforming operation on the MNDWI to construct a morphological narrow water index (MNWI). The MNWI can effectively enhance the local contrast of linear objects, such that the narrow water can be easily separated from mountain shadows, bare patches and other land covers in the MNDWI image. Furthermore, a dual-threshold segmentation method was also used to extract the potential water bodies by combining an empirical threshold segmentation on the MNDWI and an automatic threshold segmentation on the MNWI. The effectiveness of the proposed method was validated with three experimental datasets, clipped from two different Landsat images, and the results demonstrate that the narrow water bodies can be effectively extracted with the overall accuracy being over 90%. Visually, most of the narrow streams or rivers keep continuous shapes in space and their water boundaries are also precisely delineated.