CN-119555679-B - Online water quality analysis method and system based on multispectral image analysis model

CN119555679BCN 119555679 BCN119555679 BCN 119555679BCN-119555679-B

Abstract

The invention discloses an online water quality analysis method and system based on a multispectral image analysis model, which belong to the technical field of water quality analysis and comprise the following steps: the multi-spectral image analysis model comprising an analysis model, a classification model, a water plankton coverage inversion model, a water transparency inversion model, a water eutrophication inversion model and a key water quality parameter analysis model is constructed by collecting and preprocessing multi-spectral images of a target water body. And extracting water quality characteristics by using an analytical model, extracting water surface lines by using a classification model, inputting current water quality data and the water surface lines into an inversion model to perform inversion of plankton coverage, water transparency and water surface eutrophication and key water quality parameter analysis, and finally outputting comprehensive water quality assessment by fusing analysis results. The online water quality analysis method provided by the invention can realize efficient and real-time water quality monitoring and analysis, and provides scientific basis for water environment management.

Inventors

DUAN JINRONG
MA XIAOFEI
SHEN NANNAN
HE YANHUI
DONG JIAHUI

Assignees

中国水产科学研究院淡水渔业研究中心

Dates

Publication Date: 20260512
Application Date: 20241210

Claims (4)

1. An online water quality analysis method based on a multispectral image analysis model is characterized by comprising the following steps: Collecting a multispectral image of a target water body, and preprocessing the multispectral image; Constructing a multispectral image analysis model, wherein the multispectral image analysis model comprises an analysis model, a classification model, a water plankton coverage inversion model, a water transparency inversion model, a water eutrophication inversion model and a water quality parameter analysis model; Extracting characteristics of the preprocessed multispectral image by utilizing the analytical model to obtain current water quality data and water body data of a target water body; Inputting the water body data into a classification model, and extracting the water surface line of the target water body; Inputting the current water quality data and the water surface line into the water plankton coverage inversion model, the water transparency inversion model and the water surface eutrophication inversion model, and respectively carrying out water plankton coverage inversion, water transparency inversion and water surface eutrophication inversion to obtain a water plankton coverage inversion result, a water transparency inversion result and a water surface eutrophication inversion result; Carrying out fusion treatment on the water plankton coverage inversion result, the water transparency inversion result and the water surface eutrophication inversion result by utilizing the water quality parameter analysis model, and outputting a final water quality analysis result; performing feature extraction on the preprocessed multispectral image by using the analytical model to obtain current water quality data and water body data of the target water body, wherein the method comprises the following steps: Inputting the preprocessed multispectral image into a first feature extraction module, and extracting local spatial information and global spectral information to obtain local spatial features and global spectral features; Inputting the multispectral image into a second feature extraction module to extract local spectral information and global space information, and obtaining central region features and adjacent edge region features; the local spatial features, the global spectral features, the central region features and the adjacent edge region features are input into a prediction classification module to classify the multispectral image, and class labels of the multispectral image are output; acquiring current water quality data and water body data of a target water body according to the class labels; Inputting the water body data into a classification model, extracting the water surface line of the target water body, comprising: extracting the characteristics of the water body data to obtain grid characteristics of the water body; extracting the characteristics of the water body grid characteristics to obtain at least two partial water body grid characteristics; Feature fusion is carried out on the at least two local water body grid features to obtain global water body grid features; based on classification and regression processing of the global water grid characteristics, obtaining key points of the water surface line; fitting the key points to obtain a water surface system of the target water body; Inputting the current water quality data and the water surface line into the water plankton coverage inversion model to invert the water plankton coverage, comprising: Acquiring historical water quality monitoring data of a plurality of historical time periods of a target water body; generating a self-variable data set and a dependent variable data set according to the historical water quality detection data; Extracting a first self-variable component from the set of self-variable data and extracting a first dependent variable component from the set of dependent variable data; Adopting a partial least square regression algorithm to perform autoregressive processing on the first autorefraction component and the first factor component, and determining model precision according to an autoregressive result; If the model precision does not meet the preset precision condition, extracting a second self-variation component from the self-variation data set, and extracting a second factor component from the dependent variable data set for iterative processing until the model meets the preset iteration stopping condition, so as to obtain the water plankton coverage inversion model; Inputting the current water quality data and the water surface line into the water plankton coverage inversion model, and inverting the water plankton coverage to obtain a water plankton coverage inversion result of a target water body; Inputting the current water quality data and the water surface line into the water surface eutrophication inversion model to perform water surface eutrophication inversion, comprising: Constructing a plurality of water surface eutrophication inversion models based on a plurality of machine learning algorithms, and organically combining the plurality of water surface eutrophication inversion models by utilizing an entropy weight method to obtain an optimal water surface eutrophication inversion model; inputting the current water quality data and the water surface line into the optimal water surface eutrophication inversion model, and performing water surface eutrophication inversion to obtain a current water surface eutrophication inversion result of a target water body; And carrying out fusion processing on the water plankton coverage inversion result, the water transparency inversion result and the water surface eutrophication inversion result by using the water quality parameter analysis model, and outputting a final water quality analysis result, wherein the method comprises the following steps of: Respectively carrying out correlation analysis on a water plankton coverage inversion result, a water transparency inversion result and a water eutrophication inversion result to obtain a correlation analysis result; Adjusting parameters of a water quality analysis model based on the correlation analysis result; And inputting the water plankton coverage inversion result, the water transparency inversion result and the water eutrophication inversion result into an adjusted water quality analysis model to obtain a final water quality analysis result.
2. The method for online water quality analysis based on a multispectral image analysis model of claim 1, wherein the preprocessing of the multispectral image comprises geometric correction and radiation correction, multispectral image fusion, cloud and shadow removal, noise filtering and image enhancement.
3. The online water quality analysis method based on the multispectral image analysis model of claim 2, wherein the multispectral image fusion comprises: Respectively carrying out image registration on the first image and the second image; Extracting features of the first image and the second image after image registration to obtain first features; classifying the first features to obtain a classification result; and carrying out decision-level fusion based on the classification result to obtain a fused multispectral image.
4. An online water quality analysis system based on a multispectral image analysis model, which is characterized by being used for realizing the online water quality analysis method based on the multispectral image analysis model as set forth in any one of claims 1-3, comprising: the acquisition module is used for acquiring multispectral images of the target water body; The transfer station is used for preprocessing the multispectral image; the analysis module is used for carrying out water quality analysis based on the preprocessed multispectral image and comprises the following steps: the analysis unit is used for extracting characteristics of the preprocessed multispectral image by utilizing the analysis model to acquire current water quality data and water body data of the target water body; the water surface line extraction unit is used for inputting the water body data into a classification model and extracting the water surface line of the target water body; The inversion unit is used for inputting the current water quality data and the water surface line into the water plankton coverage inversion model, the water transparency inversion model and the water surface eutrophication inversion model, and respectively carrying out water plankton coverage inversion, water transparency inversion and water surface eutrophication inversion to obtain a water plankton coverage inversion result, a water transparency inversion result and a water surface eutrophication inversion result; and the water quality analysis unit is used for carrying out fusion treatment on the water plankton coverage inversion result, the water transparency inversion result and the water surface eutrophication inversion result by utilizing the water quality parameter analysis model and outputting a final water quality analysis result.

Description

Online water quality analysis method and system based on multispectral image analysis model Technical Field The invention relates to the technical field of water quality analysis, in particular to an online water quality analysis method and system based on a multispectral image analysis model. Background With the rapid development of industrialization and city, the water quality of the surface of inland rivers, lakes, and reservoirs is deteriorated, and the environmental problems of cyanobacteria bloom, water eutrophication, and the like are largely outbreaked, so that the ecological safety of water areas, the safety of drinking water of residents and the social and economic development are directly affected. The water resource and environment monitoring is an important precondition for planning, managing and protecting water and water biological resources, and the rapid, efficient and real-time monitoring of the water quality of the water resource has important significance for water resource assessment, water pollution control and water safety guarantee under the background of global climate warming and human interference aggravation. Traditional water quality monitoring is high in measurement accuracy but time-consuming and labor-consuming by collecting samples from the wild and analyzing the samples in a laboratory, and the measurement data only can represent local sample point information, so that the water quality condition of the whole space of the water body is difficult to reflect rapidly, and the real-time dynamic water quality monitoring requirement cannot be met. In recent years, development of unmanned aerial vehicle remote sensing technology provides new possibilities for water quality monitoring. The multispectral image technology is used as an advanced remote sensing means, and can capture the spectral characteristics of the water body in different wave bands, wherein the characteristics are closely related to various chemical and physical parameters in the water body. By analyzing these spectral data, the water quality condition of the water body can be deduced. However, the application of the current multispectral image analysis technology in water quality analysis has some limitations. First, it is difficult for a single spectral analysis model to address different types of water quality problems, such as water plankton coverage, water transparency inversion, and water surface eutrophication, simultaneously. Secondly, the existing water quality monitoring method is often only capable of obtaining limited water quality parameters, and complex conditions of the water body are difficult to comprehensively reflect. Therefore, how to provide an online water quality analysis method capable of comprehensively reflecting the complex condition of the water body is a problem to be solved by those skilled in the art. Disclosure of Invention In view of the above, the invention provides an online water quality analysis method and system based on a multispectral image analysis model, which can realize real-time and accurate analysis of water quality of a water body by collecting and analyzing multispectral images of a target water body. In order to achieve the above purpose, the present invention adopts the following technical scheme: an online water quality analysis method based on a multispectral image analysis model comprises the following steps: Collecting a multispectral image of a target water body, and preprocessing the multispectral image; Constructing a multispectral image analysis model, wherein the multispectral image analysis model comprises an analysis model, a classification model, a water plankton coverage inversion model, a water transparency inversion model, a water eutrophication inversion model and a water quality parameter analysis model; Extracting characteristics of the preprocessed multispectral image by utilizing the analytical model to obtain current water quality data and water body data of a target water body; Inputting the water body data into a classification model, and extracting the water surface line of the target water body; Inputting the current water quality data and the water surface line into the water plankton coverage inversion model, the water transparency inversion model and the water surface eutrophication inversion model, and respectively carrying out water plankton coverage inversion, water transparency inversion and water surface eutrophication inversion to obtain a water plankton coverage inversion result, a water transparency inversion result and a water surface eutrophication inversion result; And carrying out fusion treatment on the water plankton coverage inversion result, the water transparency inversion result and the water surface eutrophication inversion result by using the water quality parameter analysis model, and outputting a final water quality analysis result. Preferably, preprocessing the multispectral image comprises geometric correction and