Research On Water Information Extraction Method Based On Landsat 8 Remote Sensing Image

In the frequent occurrence of water-related disasters such as droughts and floods,extracting water information has become a key factor in reducing disaster losses.While on-site investigation of water resources is necessary,it is time-consuming and requires a lot of effort.In comparison,remote sensing images can quickly and comprehensively obtain water resource information.Therefore,using remote sensing images for analyzing water information has become a necessary means.In this paper,we conducted a study on different water information extraction methods using Landsat 8OLI remote sensing images,with the aim of achieving more accurate and efficient water information extraction.The main contributions of this paper are as follows:（1）We employ a new water index and digital image processing technology to extract water bodies accurately from Landsat 8 OLI images.Firstly,we preprocess Landsat 8 OLI images with image cropping,radiometric calibration,atmospheric correction and image sharpening.Subsequently,the humidity component of the Kautlr-Thomas（KT）transformation,the B component of the LBV（general radiance level,visible-infrared radiation balance,radiance variation vector）transformation,and the AWEI_nsh（Automated Water Extraction Index）are respectively used as the red,green,and blue channels of an RGB image.The new RGB image is then subjected to HIS（Hue,Intensity,Saturation）transformation,and the resulting I component is used as the new water index,named WBAWI.Then,we perform linear feature enhancement and improved local adaptive threshold segmentation method to extract small water bodies.Meanwhile,we employ morphological enhancement and improved local adaptive threshold segmentation method to extract large water bodies.Finally,we combine small and large water bodies to get complete water bodies.Compared with other traditional methods,our method has apparent advantages in water extraction,particularly in the extraction of small water bodies.（2）As the performance of threshold-based methods for water extraction heavily relies on parameter selection,we chose to use the more widely-used approach of superpixel segmentation as a substitute for threshold-based segmentation.Simple Linear Iterative Clustering（SLIC）is a simple and effective superpixel segmentation method.However,since it has some limitations in water body segmentation of remote sensing images,we have improved it and proposed a new superpixel segmentation method,called F-SLIC.Firstly,we preprocess Landsat 8 OLI images with image cropping,radiometric calibration,atmospheric correction and image sharpening,and the new water index（WBAWI）was calculated on preprocessed Landsat 8 OLI images to highlight the characteristics of water bodies.Subsequently,we performed linear feature enhancement on the WBAWI image,and calculate the local ternary pattern（LTP）texture feature of the enhanced image,which is a novel texture feature called E＿LTP.Then incorporate the multi-feature fusion of E＿LTP texture feature,spatial feature,and gray feature into the similarity measure of SLIC algorithm clustering to obtain the F-SLIC method.We performed the F-SLIC superpixel segmentation method on the WBAWI image,and finally merge the superpixels of water bodies to obtain the complete water bodies contour.The experimental results showed that compared to the SLIC method,the F-SLIC method improved the boundary recall rate by about 3%and reduced the under-segmentation error rate by about 0.2%,especially achieving significant improvement in small water body segmentation.Figure[20]table[9]reference[62]...