Research and implementation of surface water information extraction method based on high-resolution remote sensing images

Remote sensing technology provides a versatile means to detect surface water information without constraints of time and space. However, challenges in surface water extraction persist, including spectral interference and noise interference. In response, this study introduces a fusion marker-controlled watershed algorithm and an improved normalized water index method to extract surface water bodies. The segmentation of remote sensing images is achieved through morphological opening and closing operations, followed by the application of the improved normalized water index method to reduce spectral interference and minimize extracted noise. Comparative analysis with four classical water extraction methods, namely the single-band threshold method, normalized water index method, improved normalized water index method, and object-oriented method, was conducted. The study concludes with a consistency test using Kappa coefficients and accuracy rates, providing qualitative and quantitative insights into the various extraction methods. The results demonstrate that the method presented in this paper achieves higher extraction accuracy and holds significant potential for water extraction research.