Robust water body extraction from landsat imagery by using gradual assignment of water index and DSM

Water resource is crucial to the existence of every life form and also valuable to our daily life. Among its many advantages, there exist those in environmental, agricultural, industrial, and household activities as well as in climate monitoring. Nonetheless, water is also a causal factor in several major natural disasters. In order to effectively make an educated planning, remote sensing technology which are able to offer immediate and accurate means of determining water resources, are generally adopted. This paper presents detailed analyses and treatments on an adaptive water body extraction from remotely sensed images based on various water indices (NDWI, NDWI2, MNDWI and NDPI). In our framework, relaxation labeling was incorporated in order to ensure reliable and robust classification while suppressing spurious artefacts, inherited from the imaging modality. The subsequent assessment suggested that NDWI2 index yielded the most accurate classification and hence water extraction. The technique was then compared against labour intensive yet accurate manual tracing with promising consistency.