Method and system for extracting emergency water source in drought state

Abstract

The invention discloses a method and a system for extracting an emergency water source in a drought state. The method comprises the following steps: acquiring a panchromatic remote sensing training image and a panchromatic remote sensing test image; constructing an implicit Dirichlet distribution topic model, wherein the implicit Dirichlet distribution topic model is a topic model with Dirichlet prior parameters to be determined; inputting the panchromatic remote sensing training image into an implicit Dirichlet distribution topic model, estimating Dirichlet prior parameters by adopting a fixed-point iteration method, and determining the Dirichlet prior parameters in the implicit Dirichlet distribution topic model; determining the theme model with the determined Dirichlet prior parametersas an emergency water source extraction model; and inputting the panchromatic remote sensing test image into the emergency water source area extraction model to obtain a water source area in the panchromatic remote sensing test image. According to the invention, the extraction accuracy of the emergency water source can be improved, the consumption of great manpower cost can be avoided, and the extraction is convenient.

Description

technical field [0001] The invention relates to the technical field of water source extraction, in particular to an emergency water source extraction method and system in a drought state. Background technique [0002] The spatio-temporal distribution of my country's water resources is affected by precipitation, and the precipitation gradually decreases from the southeast coast to the northwest inland in space, resulting in more water resources in the east and less in the west, and more in the south and less in the north. In recent years, the northwestern region has experienced successive droughts, and resource water shortages in the northwestern region have become increasingly serious. Among them, the precipitation in the arid region of Northwest China is mainly provided by the westerly circulation and the South Asian monsoon. The agitated advance and retreat of the two airflows of the westerly wind and the monsoon determine the main characteristics of the precipitation dist...

AI Technical Summary

Problems solved by technology

[0004] However, since the reflectance of many buildings and water bodies are the same in certain wavelength bands, showing the phenomenon of "different objects with the same spectrum", the traditional method based only on optical remote sensing images will produce a large extraction error, reaching Less than high water identification accuracy

The defect of the object-oriented analysis technology of "segmentation first and then recognition" is that the scale division is difficult to grasp, so the recognition efficiency is low, and the recognition often uses the artificial visual interpretation method. Although it can obtain high interpretation accuracy, it is time-consuming. It is laborious, consumes a lot of labor costs, and it is difficult to meet the real-time requirements

The main disadvantages of using machine learning supervised classification mode are: the selection of training samples must depend on the segmentation results, and it is very inconvenient to collect training samples in the field before image analysis; training samples need to be given at the object level, so the supervision information at the pixel level will not be directly exploitable

Images