Automatic extraction method and system for sub-pixel level water body from medium-low resolution remote sensing image

Abstract

The invention discloses an automatic extraction method and system for a sub-pixel level water body from a medium-low resolution remote sensing image, and the method comprises the steps: carrying out the preprocessing on the given medium-low resolution remote sensing image, and generating a surface reflectance image; extracting a normalized difference water body index from the surface reflectance image and stretching the normalized difference water body index to construct a stretched water body index image histogram; describing a stretched water body index image histogram by using a Gaussian mixture model, and carrying out estimating to obtain water body and land parameters; determining a water and land separation threshold value according to the water body and land parameters; extracting awater and land mixed pixel according to the water body and land parameters and the water and land separation threshold; and carrying out water abundance calculation on the land-water mixed pixel to obtain a sub-pixel-level water extraction result. According to the method, the water-land separation threshold value can be rapidly and accurately extracted and is not influenced by human factors; andmeanwhile, the problem of missing extraction or over-extraction of small water bodies on the medium-low resolution remote sensing image is solved effectively.

Description

technical field [0001] The invention belongs to the technical field of remote sensing images, and in particular relates to a method and system for automatically extracting sub-pixel-level water bodies from medium and low resolution remote sensing images. Background technique [0002] Water body extraction plays an important role in water resource investigation, flood disaster monitoring and early warning, and water environmental ecological assessment. Using satellite remote sensing image data to quickly and accurately extract water body information has become an important means of water resource investigation and macroscopic monitoring. [0003] At present, there are many remote sensing image data sources and water body extraction methods for water body extraction, such as supervised / unsupervised classification methods and band ratio methods based on remote sensing images with different spatial resolutions. Because high-resolution images are not suitable for large-area wate...

AI Technical Summary

Problems solved by technology

Since high-resolution images are not suitable for large-scale water body extraction, low- and medium-resolution images have become the main data source for water body extraction, but there is a problem of mixed pixels in low- and medium-resolution images

[0004] For medium and low resolution remote sensing images, the water body extraction using the water body index method is mainly characterized by the following four aspects: (1) The threshold segmentation method is usually used, and the selection of the threshold is affected by factors such as image type, acquisition time, and geographical location It is relatively large and has a high degree of manual participation, which is not conducive to the development of automatic water body extraction; (2) Most of the water body extraction is based on the pixel level, but less involved in the sub-pixel level, which is not conducive to the accurate extraction of small water bodies; (3) The effect of extracting simple areas with land and water boundaries is good, but it is rarely involved in the extraction of water bodies such as aquaculture areas with complex boundaries; (4) The water body index is directly calculated using the DN value or apparent reflectance, which will directly affect the water body Extraction accuracy

[0005] Under such circumstances, the traditional method based on the water body index and the method of extracting water bodies through manual participation to determine the threshold value have presented new challenges, and it is easy to mis-extract small water body targets in low- and medium-resolution images, and it is difficult to adapt to water and land boundaries. Water Extraction in Complex Situations

Images