A remote sensing image water body extraction method and system based on deep learning

Abstract

The invention discloses a remote sensing image water body extraction method and system based on deep learning. By utilizing China's high score-No.1 / No.2 (GF-1 / GF-2) and resource-No.3 (ZY-3) satelliteimages and two kinds of deep learning convolution neural networks (CNNs) of U-net and Densenet, a novel CNNs model for extracting water bodies is provided. By labeling water bodies in a large number of high-resolution remote sensing images, obtaining water body labeling results in the images to serve as a training set; Carrying out a training test on the high-resolution remote sensing image by using the established CNNs model to obtain a binary image of the water body; And identifying the pixel value identified in the binary image to obtain the water body target. According to the method, the water body information in the high-resolution remote sensing image can be quickly extracted in a high-precision and automatic manner, the intelligence, automation level and extraction precision of water body extraction are improved, and the method can be widely applied to various fields related to water body extraction.

Description

technical field [0001] The invention relates to the technical field of remote sensing image processing, in particular to a method and system for extracting water bodies from remote sensing images based on deep learning. Background technique [0002] With the development of my country's aerospace industry, more and more ground object information is obtained by earth observation, and the image resolution obtained by satellite images and aerial images is also getting higher and higher. Extracting water bodies on the Earth's surface has always been a fundamental task in remote sensing image analysis. As a key component of the hydrological cycle, the extent of water is critical for monitoring extreme events and understanding the mechanisms of climate change. [0003] Regionally, changes in water have major impacts on ecosystems and human life. Accurate perception of water level status is critical for successful quantitative assessment of spatiotemporal monitoring, change detect...

AI Technical Summary

Problems solved by technology

[0004] Existing remote sensing methods for water body extraction mainly rely on complex spectral analysis limited by experience and the optimal threshold value obtained through extensive training, and the optimal threshold value varies greatly for different images, and the overall level of automation and intelligence is not high. Meet the requirements of high spatial resolution image automation, fast processing of high-precision images, and accurate detection of water bodies

Images