A river flow inversion method based on multi-source optical remote sensing images

Abstract

The embodiment of the application discloses a river flow inversion method based on multi-source optical remote sensing images, relates to the technical field of hydrological monitoring, and comprises the following steps: acquiring multi-source remote sensing images containing a target research basin, processing to obtain a plurality of reference image pairs, fusing to-be-fused images and matched reference image pairs to obtain high-resolution fused images; a monitoring area is divided from the target research basin, and a monitoring pixel is screened based on an indication index value and a flow correlation coefficient; a function relationship between an indication index mean value of each date of each type of indication index and river flow is determined to construct a river flow inversion model; and the river flow of the corresponding date of the to-be-predicted remote sensing image is predicted through the river flow inversion model. According to the application, the flow inversion can be carried out by using the disclosed optical remote sensing images without relying on river surface features and hydraulic parameters, missing runoff data can be supplemented, and even hydrological stations can be replaced, so that the time and labor cost of flow observation are reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of hydrological monitoring, and in particular to a river flow inversion method based on multi-source optical remote sensing images. BACKGROUND

[0002] Flow refers to the amount of water passing through a certain section of a river per unit time, and runoff data is of great significance for flood prevention, drought resistance, and comprehensive management of water resources. At present, runoff observation data is mainly obtained through in-situ monitoring. This method generally uses fixed hydrological stations to monitor target areas. In some remote mountainous areas and other places with harsh geographical and climatic conditions, it is difficult to continuously deploy and maintain these fixed hydrological stations. In addition, as the region develops and changes, the deployment location of many hydrological stations will change, resulting in a technical interruption in runoff data collection, and the collected data may not fully cover all spatial ranges of the original target area.

[0003] At present, in order to overcome the above-mentioned problem of spatial coverage missing, remote sensing technology can be used to realize flow monitoring and inversion. Some related technologies can extract elevation information from radar remote sensing image data and calculate the flow per unit time by combining river section parameters. Some related technologies can also use river surface features to calculate the water surface width and then complete flow estimation. However, these related technologies need to rely on prior information of river surface features (such as river width, water area, etc.) or river hydraulic parameters (such as section shape, underwater topography, etc.). In actual application, it is difficult to obtain such data, and the accuracy of the data is low. For example, in remote and complex river sections, due to geographical conditions, traditional surveying methods cannot efficiently obtain detailed data such as river section shape and underwater topography. For dynamic change scenarios, due to factors such as river erosion by floodwater and sediment deposition, the river surface features and hydraulic parameters will change over time, and the timeliness of the collected river surface features and hydraulic parameters is insufficient. As a result, the disturbance and loss of these basic data will greatly reduce the flow inversion accuracy of related technologies, and even make it impossible to achieve effective monitoring.

[0004] Therefore, there is currently a lack of a method that can efficiently and accurately invert river flow. SUMMARY

[0005] The embodiments of the present application provide a river flow inversion method based on multi-source optical remote sensing images to solve the defects of the related art. The technical solution is as follows:

[0006] In a first aspect, the embodiments of the present application provide a river flow inversion method based on multi-source optical remote sensing images, comprising:

[0007] Obtaining multi-source remote sensing images containing a target research basin, and processing a plurality of reference image pairs based on the multi-source remote sensing images;

[0008] Selecting a plurality of low-resolution to-be-fused images from the multi-source remote sensing images, determining a reference image pair matched with each to-be-fused image, and fusing each to-be-fused image and the matched reference image pair to obtain a high-resolution fused image;

[0009] Dividing a monitoring area from the target research basin based on river flow of all dates, and screening monitoring pixels of each type of indicator based on a correlation coefficient between a value of each type of indicator of each pixel in the monitoring area and river flow;

[0010] Obtaining an indicator mean value of all monitoring pixels of each type of indicator on each date, and constructing a river flow inversion model based on a functional relationship between the indicator mean value and river flow of the corresponding date;

[0011] Inputting a to-be-predicted remote sensing image into the river flow inversion model to predict river flow of a date corresponding to the to-be-predicted remote sensing image.

[0012] In an optional implementation of the first aspect, the obtained multi-source remote sensing images include a plurality of low-resolution remote sensing images and high-resolution remote sensing images containing the target research basin;

[0013] The processing of the plurality of reference image pairs based on the multi-source remote sensing images includes:

[0014] Determining a date of each remote sensing image, and selecting a low-resolution remote sensing image and a high-resolution remote sensing image with the same date from the multi-source remote sensing images;

[0015] Constructing a reference image pair based on one low-resolution remote sensing image and one high-resolution remote sensing image with the same date, respectively.

[0016] In an optional implementation of the first aspect, the selecting of the plurality of low-resolution to-be-fused images from the multi-source remote sensing images and the determining of the reference image pair matched with each to-be-fused image include:

[0017] Selecting a plurality of low-resolution remote sensing images from the multi-source remote sensing images, and taking each low-resolution remote sensing image as a to-be-fused image;

[0018] Determining a first principal component of each to-be-fused image based on principal component analysis, and determining a first principal component of a low-resolution remote sensing image in all reference image pairs based on principal component analysis;

[0019] Calculate the Euclidean distance between the first principal component of the image to be fused and the first principal component corresponding to each reference image pair, and select the two reference image pairs with the smallest Euclidean distance as the reference image pairs matched with the image to be fused.

[0020] In an optional implementation of the first aspect, the river flow based on all dates is divided into a monitoring area from the target research basin, including:

[0021] Obtain the river flow of each date;

[0022] Determine the maximum flow date corresponding to the maximum river flow, and determine the high-resolution image or high-resolution fused image corresponding to the maximum flow date, from which the water body region and the non-water body region are segmented;

[0023] In the case where there is no high-resolution image or high-resolution fused image corresponding to the maximum flow date, determine the river flow in all high-resolution images and high-resolution fused images, select the high-resolution image or high-resolution fused image corresponding to the maximum river flow in all high-resolution images and high-resolution fused images, and segment the water body region and the non-water body region therefrom; expand the edge of the segmented water body region outward by a preset number of pixels, and construct a monitoring area for inverting the river flow based on the expanded pixels and the water body region.

[0024] In an optional implementation of the first aspect, the correlation coefficient between each pixel of each type of indicator in the monitoring area and the river flow is used to filter the monitoring pixels of each type of indicator, including:

[0025] For each type of indicator, obtain the indicator value of each pixel in the monitoring area at each date, obtain the river flow at each date, and calculate the correlation coefficient between the indicator value of each pixel and the flow at the corresponding date;

[0026] Determine the pixels with a correlation coefficient greater than the correlation coefficient threshold of the corresponding type of indicator, and filter the monitoring pixels of each type of indicator.

[0027] In an optional implementation of the first aspect, the correlation coefficient threshold is determined based on the following steps:

[0028] For each type of indicator, obtain the absolute value of the correlation coefficient value of each pixel at each date, sort the absolute values based on their sizes, and select the correlation coefficient value at a preset sorting position as the correlation coefficient threshold of the corresponding type of indicator based on the sorting result.

[0029] In an optional implementation of the first aspect, the various types of indication indexes respectively include a normalized water index, a new water index, a water turbidity index, a bare soil index, a vegetation index, a normalized difference moisture index, and reflectivity values of various bands in the remote sensing image.

[0030] In a second aspect, the embodiments of the present application further provide a river flow inversion device based on multi-source optical remote sensing images, comprising:

[0031] An image acquisition module is configured to acquire multi-source remote sensing images containing a target research basin, and process a plurality of reference image pairs based on the multi-source remote sensing images.

[0032] The image acquisition module is further configured to select a plurality of low-resolution to-be-fused images from the multi-source remote sensing images, determine reference image pairs matched with the to-be-fused images, and fuse each to-be-fused image and the matched reference image pair to obtain a high-resolution fused image.

[0033] A monitoring pixel acquisition module is configured to divide a monitoring area from the target research basin based on river flows of all dates, and filter monitoring pixels of each type of indication index based on correlation coefficients between various types of indication index values of each pixel in the monitoring area and river flows.

[0034] The monitoring pixel acquisition module is further configured to acquire an indication index mean value of all monitoring pixels of each type of indication index on each date, and construct a river flow inversion model based on a functional relationship between the indication index mean value and the river flow of the corresponding date.

[0035] A flow prediction module is configured to input a to-be-predicted remote sensing image into the river flow inversion model, and predict a river flow of a corresponding date of the to-be-predicted remote sensing image.

[0036] In a third aspect, the embodiments of the present application further provide an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method provided by the first aspect or any one of the implementation manners of the first aspect.

[0037] In a fourth aspect, the present application further provides a non-transitory computer readable storage medium having a computer program stored thereon, wherein the computer program is executable by a processor to implement the method provided by the first aspect or any one of the implementation manners of the first aspect.

[0038] The technical solutions provided by some embodiments of the present application have at least the following beneficial effects:

[0039] The river flow inversion method based on multi-source optical remote sensing images provided by the embodiment of the application can improve the fusion accuracy of images based on the principal component analysis method to optimally select a reference image pair matched with the to-be-fused image, can obtain a high-resolution fused image by fusing the to-be-fused image and the matched reference image pair, can fully utilize the ground feature and band information of remote sensing images of different sources, and thus can improve the flow inversion accuracy.

[0040] Based on the river flow and the correlation coefficient, the monitoring pixels of each type of indication index can be screened, and thus the sensitivity of the monitoring pixels to flow fluctuations can be ensured, and noise affecting the pixel indication index values other than the flow factor can be excluded, and the robustness of flow inversion can be improved.

[0041] The river flow inversion model constructed based on the functional relationship between the indication index mean value and the river flow of the corresponding date can fully utilize the public optical remote sensing images for flow inversion without relying on prior information such as river surface features and river hydraulic parameters, and can complement missing runoff data or even replace hydrological stations, so as to reduce the time cost and labor cost of flow observation. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the application or related art, the drawings needed to be used in the embodiments or related art description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0043] Figure 1 is a flowchart of a river flow inversion method based on multi-source optical remote sensing images provided by the embodiment of the application;

[0044] Figure 2 is a schematic diagram of the distribution of monitoring pixels of a river flow inversion method based on multi-source optical remote sensing images provided by the embodiment of the application;

[0045] Figure 3 is a structural schematic diagram of a river flow inversion device based on multi-source optical remote sensing images provided by the embodiment of the application;

[0046] Figure 4 is a structural schematic diagram of an electronic device provided by the embodiment of the application. DETAILED DESCRIPTION

[0047] In order to make the purposes, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be described clearly and completely below in conjunction with the accompanying drawings in the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0048] The terms "comprising" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or modules is not limited to the listed steps or modules, but can optionally also include steps or modules that are not listed, or can optionally also include other steps or modules inherent to the process, method, product or device.

[0049] It should be noted that the terms "first" and "second" involved in the present application are only to distinguish similar objects, and do not represent a specific order of the objects. Understandably, "first" and "second" can be interchanged in a specific order or sequence as allowed. It should be understood that the objects distinguished by "first" and "second" can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those described or illustrated herein.

[0050] The present application will be described in detail below in conjunction with specific embodiments.

[0051] Next, in conjunction with Figure 1 , a river flow retrieval method based on multi-source optical remote sensing images is introduced. For details, please refer to Figure 1 , Figure 1 A flowchart of a river flow retrieval method based on multi-source optical remote sensing images is shown. As Figure 1 shown, the method comprises the following steps:

[0052] S101, obtaining multi-source remote sensing images containing a target research basin, and processing a plurality of reference image pairs based on the multi-source remote sensing images;

[0053] S102, selecting a plurality of low-resolution to-be-fused images from the multi-source remote sensing images, determining a reference image pair matched with the to-be-fused images, and fusing each to-be-fused image and the matched reference image pair to obtain a high-resolution fused image;

[0054] S103, dividing the target research basin based on river flows of all dates to obtain a monitoring area, and respectively screening monitoring pixels of each type of indication index based on a correlation coefficient of each type of indication index value and river flow of each pixel in the monitoring area.

[0055] S104, obtaining an average of the indication index of all monitoring pixels of each type of indication index on each date, and constructing a river flow inversion model based on a functional relationship between the average of the indication index and the river flow of the corresponding date;

[0056] S105, inputting the remote sensing image to be predicted into the river flow inversion model to predict the river flow of the corresponding date of the remote sensing image to be predicted.

[0057] In some embodiments, in S101, the obtained multi-source remote sensing images include a plurality of low-resolution remote sensing images and high-resolution remote sensing images containing a target research basin. After obtaining the multi-source remote sensing images, the remote sensing images can be screened according to a determined research date range. The remote sensing images can be screened according to the cloud cover, and cloud-free images are selected to exclude images seriously disturbed by clouds. The screened images can be subjected to pretreatment such as mosaic splicing, terrain correction, and radiation correction.

[0058] Specifically, the image source of a scale matched according to the average river width of the target research basin can also be selected. The river section needs to cover the river channel with historical runoff data and without tributary confluence, so as to train the model according to the existing flow data. After the model is trained, the flow of a specified date can be predicted based on the model, without relying on hydrological data.

[0059] Specifically, the obtained multi-source remote sensing images can include a red light band, a green light band, a blue light band, a near-infrared band, and two short-wave near-infrared bands, without limitation in the embodiments of the present application.

[0060] In some embodiments, the processing of the multi-source remote sensing images to obtain a plurality of reference image pairs includes:

[0061] determining the date of each remote sensing image, and selecting low-resolution remote sensing images and high-resolution remote sensing images with the same date from the multi-source remote sensing images;

[0062] constructing a reference image pair based on one low-resolution remote sensing image and one high-resolution remote sensing image with the same date, respectively.

[0063] If the number of low-resolution remote sensing images and high-resolution remote sensing images with the same date is less than the required number, a high-resolution remote sensing image with a similar hydrological regime (similar flow and similar shallow area exposure area) and a similar shooting date of the low-resolution remote sensing image can be selected as a substitute, so as to ensure the number of obtained reference image pairs. The similar shooting date can be understood as that the time difference between the shooting date of the substitute remote sensing image and the shooting date of the low-resolution remote sensing image is less than a given time threshold, for example, within twice the satellite return period, without limitation in the embodiments of the present application.

[0064] In some embodiments, in S102, a plurality of low-resolution to-be-fused images are selected from the multi-source remote sensing images, and a reference image pair matched with the to-be-fused images is determined, specifically including the steps of:

[0065] A plurality of low-resolution remote sensing images are selected from the multi-source remote sensing images, and each low-resolution remote sensing image is taken as a to-be-fused image;

[0066] A first principal component (PC1) of each to-be-fused image is determined based on a principal component analysis (PCA), and a first principal component of a low-resolution remote sensing image in all reference image pairs is determined based on the principal component analysis.

[0067] It can be understood that a remote sensing image contains remote sensing data of multiple bands, and the principal component analysis on the remote sensing image can be understood as the principal component analysis on the multiple bands of the remote sensing image. The first variable obtained is the first principal component. The principal component can be understood as a linear combination of each band in the remote sensing image. The first principal component can express the main spatial and spectral characteristics of the remote sensing image. The principal component analysis method and the first principal component are the same as the related content in the related technology, and details are not repeated here.

[0068] The Euclidean distance between the first principal component of the to-be-fused image and the first principal component corresponding to each reference image pair is calculated, and two reference image pairs with the smallest Euclidean distance are selected as the reference image pair matched with the to-be-fused image.

[0069] The two low-resolution images with the closest principal components can be selected by the Euclidean distance.

[0070] Specifically, in S102, the fusion step can be performed by an enhanced spatial and temporal adaptive reflectance fusion model (ESTARFM), and each to-be-fused image and the matched reference image pair are fused to obtain a high-resolution fused image. The obtained high-resolution fused image fuses the reflectance information of the five remote sensing images and contains the reflectance of each pixel in each band.

[0071] In some embodiments, in S103, a monitoring area is divided from the target research basin based on the river flow of all dates, specifically including the steps of:

[0072] Obtain the river flow of each date;

[0073] determining a date corresponding to the maximum river flow value;

[0074] Specifically, the river flow value corresponding to the date can be obtained directly from the hydrological data, or calculated according to the information recorded in the high-resolution image or high-resolution fused image. The date corresponding to the maximum river flow value is determined.

[0075] Further, if there is a high-resolution image or high-resolution fused image corresponding to the same date as the maximum flow date, the water body region and the non-water body region are segmented from the high-resolution image corresponding to the maximum flow date.

[0076] In the absence of a high-resolution image or high-resolution fused image corresponding to the maximum flow date, the river flow values in all high-resolution images and high-resolution fused images are determined, and the high-resolution image or high-resolution fused image corresponding to the maximum river flow value in all high-resolution images and high-resolution fused images is selected to segment the water body region and the non-water body region, i.e., the water body region and the non-water body region are segmented based on the high-resolution image or high-resolution fused image corresponding to the maximum value.

[0077] Specifically, the normalized water body index (NDWI) and the advanced normalized water body index (ANDWI) corresponding to the date can be extracted from the high-resolution fused image. Based on the normalized water body index NDWI and the advanced normalized water body index ANDWI, the water body area can be determined. In addition, the water body area can also be obtained by segmenting the river channel water body and the non-water body region according to the maximum inter-class variance method. As a preferred embodiment, the union region of the water body area calculated according to the indicative index and the water body area calculated according to the maximum inter-class variance method can be obtained to obtain the maximum water body area. Thus, the segmentation of the water body region and the non-water body region is completed.

[0078] Finally, the edge of the segmented water body region is extended outward by a predetermined number of pixels, and a monitoring region for retrieving the river flow is constructed based on the extended pixels and the water body region.

[0079] Specifically, a water mask can be made by extending a predetermined range of buffer zones outward from the left and right banks of the river channel based on the maximum water body area. Finally, the water mask is preliminarily determined as the monitoring region for the remote sensing image flow retrieval. Exemplarily, the extension width of the left and right banks can be set to 50 m.

[0080] Exemplarily, the various types of indication indexes of the embodiments of the present application respectively include a normalized water body index, a new water body index, a turbidity water index (TWI), a bare soil index (BSI), a vegetation index (NDVI), and a normalized difference moisture index (NDMI). Each type of indication index can be calculated according to the reflectivity of the corresponding wave band, and the reflectivity can be extracted from the wave band information recorded in the remote sensing image.

[0081] In addition, the remote sensing image also includes information of multiple wave bands, and the reflectivity values of the multiple wave bands can also be used as indication indexes, for example, the reflectivity value of the near-infrared wave band NIR and the reflectivity value of the red light wave band R. The specific ranges of each wave band depend on the application scenarios and the parameters of the satellite-mounted sensors, and the embodiments of the present application do not limit this.

[0082] Among them, the normalized water body index, the new water body index, the reflectivity value of the near-infrared wave band, can be used to represent the water surface change, the reflectivity value of the red light wave band, and the turbidity water index can be used to represent the water quality, and the bare soil index, the vegetation index, and the normalized difference moisture index can be used to represent the changes on both sides of the river bank and the beach.

[0083] In some embodiments, in S103, the monitoring pixels of each type of indication index are respectively filtered based on the correlation coefficients between the values of each type of indication index of each pixel in the monitoring area and the river flow, and specifically include:

[0084] For each type of indication index, the indication index value of each pixel in the monitoring area on each date is obtained, the river flow on each date is obtained, and the correlation coefficient between the indication index value of each pixel and the flow on the corresponding date is calculated.

[0085] The pixels with a correlation coefficient greater than the correlation coefficient threshold of the corresponding type of indication index are determined, and the monitoring pixels of each type of indication index are filtered.

[0086] Specifically, the high-resolution remote sensing images in the multi-source remote sensing images obtained in the foregoing step S101 and the high-resolution fusion images obtained in S102 can be taken as a high-resolution remote sensing image set, each high-resolution remote sensing image has a corresponding date, all pixels corresponding to the location range of the monitoring area in all high-resolution remote sensing images can be extracted, and the correlation coefficients between the values of various types of indication indexes and the river flow are calculated to filter the monitoring pixels of each type of indication index.

[0087] Specifically, the correlation coefficient is specifically a pearson correlation coefficient, which is a statistical index for measuring the strength and direction of linear relationship between two variables, and is suitable for evaluating the linear dependence between two continuous variables.

[0088] Specifically, when compared with the correlation coefficient threshold, the absolute value of each correlation coefficient can be taken, and the correlation coefficient threshold of each type of indication index is independent of each other and can be set to different values according to actual conditions. If the threshold is set too small, the monitoring range can be too large, and the sensitivity of the monitoring pixel to the flow fluctuation can be reduced. If the threshold is set too large, the monitoring range is too small, and the reflectivity robustness of the monitoring pixel is reduced, and there can be noise affecting the reflectivity of the pixel in addition to the flow factor.

[0089] For example, for each type of indication index, the indication index value of each pixel in each high-resolution image located in the monitoring area can be extracted, and the correlation coefficient between the indication index value and the corresponding date flow can be calculated. As long as the correlation coefficient calculated based on the data of any date of the pixel is greater than the correlation coefficient threshold of the type of indication index, the pixel can be retained, which is recorded as a monitoring pixel corresponding to the type of indication index. The same step of judgment is performed for each pixel, so that all monitoring pixels corresponding to the type of indication index can be obtained, and all monitoring pixels of each type of indication index can be determined.

[0090] Understandably, the monitoring pixels determined in S103 can be understood as the location of the monitoring pixels on the remote sensing image and the coordinate range of the geographical position.

[0091] In some embodiments, the correlation coefficient threshold is determined based on the following steps:

[0092] For each type of indication index, the absolute value of the correlation coefficient value of each pixel on each date is obtained, and the absolute values are sorted based on the size. Based on the sorting result, the correlation coefficient value at the preset sorting position is selected as the correlation coefficient threshold of the corresponding type of indication index.

[0093] For example, the correlation coefficient value corresponding to the top 5% of the absolute value can be set, and the pixels exceeding the threshold are determined as monitoring pixels. For example, the distribution of monitoring pixels of different indication indexes is as shown in Figure 2 Figure 2 a is the distribution of water surface index monitoring pixels, corresponding to normalized water index NDWI, new water index ANDWI, and near-infrared band reflectivity NIR, Figure 2 b is the distribution of water quality index monitoring pixels, corresponding to water turbidity index TWI and red band reflectivity R, Figure 2 c is the distribution of beach and riverbank index monitoring pixels, corresponding to bare soil index BSI, vegetation index NDVI, and normalized humidity index NDMI.​

[0094] It should be noted that the high-resolution remote sensing image in the multi-source remote sensing image obtained in the foregoing step S101 and the high-resolution fused image obtained in S102 can be taken as a high-resolution remote sensing image set, and S103 can extract the monitoring pixels of each type of indicative index based on each high-resolution image in the high-resolution remote sensing image set. Since the target research basin is determined in the embodiment of the present application, all the remote sensing images obtained should contain the target research basin, and all the remote sensing images can be aligned according to the geographic coordinates. Thus, the geographic coordinates corresponding to the monitoring pixels obtained in S103 can be used to apply the monitoring pixels to each remote sensing image, and the indicative index and / or band reflectivity value at the position corresponding to the monitoring pixels can be extracted.

[0095] Specifically, in S104, for each type of indicative index, the indicative index value of each monitoring pixel of the type of indicative index on a date can be obtained, the mean value of the indicative index values of all the monitoring pixels is taken as the daily monitoring value of the type of indicative index corresponding to the date, the mean value of the indicative index values of all the monitoring pixels on each date is calculated, and the mean values are arranged in chronological order of the dates to obtain the indicative index mean value time series of the type of indicative index. In this way, the indicative index mean value time series of each type of indicative index can be obtained.

[0096] Meanwhile, the river flow on each date can be obtained, and the river flow time series can be obtained by arranging the river flow in chronological order of the dates.

[0097] Specifically, for the indicative index mean value time series of each type of indicative index and the river flow time series, a multiple linear regression method can be used for fitting to obtain the fitting coefficient of each type of indicative index, so as to construct the functional relationship between the multiple types of indicative index and the river flow, and obtain the river flow inversion model.

[0098] In some embodiments, the river flow inversion model can also be constructed based on a random forest algorithm. The random forest algorithm is a non-supervised learning method, and its generalization ability is more stable than that of a single decision tree. Compared with multiple linear regression, the random forest can more accurately depict the nonlinear relationship between each index and the flow. Each type of indicative index can be taken as a characteristic variable, and the corresponding date flow data can be input into the model. By setting the parameter ranges of the number of trees (n estimators), the minimum number of samples required for node splitting (min samples split), the minimum number of samples of tree nodes (min sample leaf), and the maximum number of features for division (max features), the grid search method (Grid Search) is used to minimize the mean square error (MSE) as the optimization objective, so as to determine the best parameter combination of the river flow inversion model.

[0099] In some embodiments, a sample set can also be constructed based on the indicator mean time series of a plurality of indicators and the river flow time series, with the indicator mean of each indicator on each date as the sample input, and the river flow of the corresponding date as the sample label, to train the constructed neural network model, construct a loss function according to the comparison result of the flow value predicted by the model and the sample label, output the parameters of the converged neural network model, and obtain the river flow inversion model.

[0100] In some embodiments, in S105, the inputted remote sensing image to be predicted can be obtained, the date of the remote sensing image to be predicted can be easily obtained, the monitoring pixels of each indicator determined in S103 can be obtained, the indicator values on the monitoring pixels of the remote sensing image to be predicted can be collected, the indicator values on the monitoring pixels of each type of indicator can be obtained respectively, and the indicator values on the monitoring pixels of each type of indicator can be inputted into the river flow inversion model, so as to obtain the river flow of the corresponding date predicted by the model for the remote sensing image to be predicted. It can be understood that the inputted remote sensing image to be predicted at least needs to cover the monitoring area divided in S103.

[0101] In some embodiments, a plurality of methods such as multiple linear regression method, random forest algorithm, neural network model, etc. can be used to construct the river flow inversion model respectively, in the actual prediction process, the corresponding river flow can be predicted based on each river flow inversion model respectively, and finally the average value of the river flow predicted by each river flow inversion model is taken as the final output to predict the river flow of the corresponding date of the remote sensing image to be predicted.

[0102] The following is an apparatus embodiment of the present application, which can be used to execute the method embodiment of the present application. For details not disclosed in the apparatus embodiment of the present application, please refer to the method embodiment of the present application.

[0103] Next, please refer to Figure 3 The structure schematic diagram of the river flow inversion device based on multi-source optical remote sensing image provided by an exemplary embodiment of the present application. The device can be realized as all or part of the terminal by software, hardware or combination of the two, and can also be integrated as an independent module on the server. The river flow inversion device based on multi-source optical remote sensing image in the embodiment of the present application can be applied to the terminal or cloud, and the device 30 comprises an image preprocessing module 301, a monitoring pixel acquisition module 302 and a flow inversion module 303, wherein:

[0104] The image preprocessing module 301 is used to acquire multi-source remote sensing images containing a target research basin, and a plurality of reference image pairs are processed based on the multi-source remote sensing images;

[0105] The image preprocessing module 301 is further configured to select a plurality of low-resolution to-be-fused images from the multi-source remote sensing images, determine a reference image pair matched with each to-be-fused image, and fuse each to-be-fused image and the matched reference image pair to obtain a high-resolution fused image.

[0106] The monitoring pixel acquisition module 302 is configured to divide the target research basin into a monitoring area based on river flow of all dates, and filter monitoring pixels of each type of indicator based on a correlation coefficient between each type of indicator value of each pixel in the monitoring area and river flow.

[0107] The monitoring pixel acquisition module 302 is further configured to obtain an indicator mean value of all monitoring pixels of each type of indicator on each date, and construct a river flow inversion model based on a functional relationship between the indicator mean value and river flow of the corresponding date.

[0108] The flow inversion module 303 is configured to input a to-be-predicted remote sensing image into the river flow inversion model to predict river flow of a corresponding date of the to-be-predicted remote sensing image.

[0109] It should be noted that the apparatus 30 provided in the above embodiments is only used as an example to divide the above functions into different functional modules when the river flow inversion method based on multi-source optical remote sensing images is executed, and in actual applications, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus provided in the above embodiments and the river flow inversion method based on multi-source optical remote sensing images belong to the same concept, and the implementation process is detailed in the method embodiments, which will not be described here.

[0110] The embodiments of the present application further provide an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the method of any one of the above embodiments when executing the program.

[0111] Please refer to Figure 4 The structure block diagram of an electronic device provided in the embodiments of the present application is shown in FIG. 4.

[0112] As shown in Figure 4 The electronic device 400 includes a processor 401 and a memory 402.

[0113] In the embodiments of the present application, the processor 401 is the control center of the computer system, and can be a processor of a physical machine or a processor of a virtual machine. The processor 401 can include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 401 can be implemented in at least one of a hardware form of a DSP (Digital Signal Processing), an FPGA (Field-Programmable Gate Array), and a PLA (Programmable Logic Array).

[0114] The processor 401 can also include a main processor and a coprocessor. The main processor is a processor for processing data in a wake-up state, also known as a CPU (Central Processing Unit). The coprocessor is a low-power processor for processing data in a standby state.

[0115] The memory 402 can include one or more computer-readable storage media, which can be non-transitory. The memory 402 can also include a high-speed random access memory, and a nonvolatile memory such as one or more disk storage devices, flash storage devices. In some embodiments of the present application, the non-transitory computer-readable storage medium in the memory 402 is used to store at least one instruction for being executed by the processor 401 to implement the method in the embodiments of the present application.

[0116] In some embodiments, the electronic device 400 further includes a peripheral device interface 403 and at least one peripheral device 404. The processor 401, the memory 402, and the peripheral device interface 403 can be connected through a bus or a signal line. Each peripheral device 404 can be connected to the peripheral device interface 403 through a bus, a signal line, or a circuit board. Specifically, the peripheral device interface 403 can be used to connect at least one peripheral device related to I / O (Input / Output) to the processor 401 and the memory 402.

[0117] In some embodiments of the present application, the processor 401, the memory 402, and the peripheral device interface 403 are integrated on the same chip or circuit board; in some other embodiments of the present application, any one or two of the processor 401, the memory 402, and the peripheral device interface 403 can be implemented on a separate chip or circuit board. The embodiments of the present application do not make specific limitations in this regard.

[0118] The electronic device structure block diagram shown in the embodiments of the present application does not constitute a limitation on the electronic device 400, and the electronic device 400 can include more or fewer components than shown, or combine certain components, or use a different arrangement of components.

[0119] The embodiments of the present application also provide a computer readable storage medium, which stores a computer program, and the program is executed by a processor to implement the steps of the method of any of the preceding embodiments. The computer readable storage medium can include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a micro drive, and a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic card or an optical card, a nanosystem (including a molecular memory IC), or any type of medium or device suitable for storing instructions and / or data.

[0120] From the above description of the embodiments, those skilled in the art can clearly understand that the embodiments can be implemented by means of software plus necessary universal hardware platforms, and of course can also be implemented by hardware. Based on such an understanding, the above technical solutions, essentially or in other words, the part that contributes to the related art, can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some part of the embodiments.

[0121] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for river flow inversion based on multi-source optical remote sensing imagery, characterized in that, include: Acquire multi-source remote sensing images containing the target watershed, and process the multi-source remote sensing images to obtain multiple reference image pairs; Multiple low-resolution images to be fused are selected from the multi-source remote sensing images. A reference image pair matching the images to be fused is determined. Each image to be fused and the matching reference image pair are fused to obtain a high-resolution fused image. The monitoring area is divided from the target study watershed based on the river flow for all dates. The monitoring cells for each type of indicator are selected based on the correlation coefficient between the values ​​of various indicator values ​​of each cell in the monitoring area and the river flow. The mean value of the indicator for each type of indicator is obtained for all monitoring cells on each date. Based on the functional relationship between the mean value of the indicator and the river flow on the corresponding date, a river flow inversion model is constructed. The river flow inversion model is used to input the remote sensing image to be predicted, and the river flow corresponding to the date of the remote sensing image to be predicted is predicted. The process of selecting the monitored pixels includes: For each type of indicator, the indicator value of each pixel in the monitoring area is obtained on each date, the river flow on each date is obtained, and the correlation coefficient between the indicator value of each pixel and the flow on the corresponding date is calculated. Pixels with correlation coefficients greater than the correlation coefficient threshold of the corresponding indicator type are identified, and monitoring pixels for each type of indicator are obtained by screening.

2. The method for river flow inversion based on multi-source optical remote sensing imagery according to claim 1, characterized in that, The acquired multi-source remote sensing images include multiple low-resolution and high-resolution remote sensing images containing the target watershed. The process of processing the multi-source remote sensing images to obtain multiple reference image pairs includes: Determine the date of each remote sensing image, and select low-resolution and high-resolution remote sensing images with the same date from the multi-source remote sensing images; A reference image pair is constructed based on a low-resolution remote sensing image and a high-resolution remote sensing image with the same date.

3. The river flow inversion method based on multi-source optical remote sensing imagery according to claim 2, characterized in that, The step of selecting multiple low-resolution images to be fused from the multi-source remote sensing images and determining a reference image pair that matches the images to be fused includes: Multiple low-resolution remote sensing images are selected from the multi-source remote sensing images, and each low-resolution remote sensing image is used as an image to be fused. The first principal component of each image to be fused is determined based on principal component analysis, and the first principal component of all reference image pairs of medium and low resolution remote sensing images is determined based on principal component analysis. Calculate the Euclidean distance between the first principal component of the image to be fused and the first principal component of each reference image pair, and select the two reference image pairs with the smallest Euclidean distance as the reference image pairs that match the image to be fused.

4. The method for river flow inversion based on multi-source optical remote sensing imagery according to claim 2, characterized in that, The monitoring area, derived from the river flow data across all dates, is divided from the target study watershed and includes: Get the river flow rate for each date; Determine the date of maximum flow corresponding to the maximum river flow, determine the high-resolution image or high-resolution fused image corresponding to the date of maximum flow, and segment the water body area and non-water body area from it; In the absence of high-resolution images or high-resolution fused images corresponding to the date of the maximum flow, the river flow in all high-resolution images and high-resolution fused images is determined. The high-resolution image or high-resolution fused image corresponding to the maximum river flow in all high-resolution images and high-resolution fused images is selected, and the water body area and non-water body area are segmented from it. A preset number of pixels are extended outward from the edge of the segmented water body area, and a monitoring area for inverting the river flow is constructed based on the extended pixels and the water body area.

5. The method for river flow inversion based on multi-source optical remote sensing imagery according to claim 1, characterized in that, The correlation coefficient threshold is determined based on the following steps: For each type of indicator, obtain the absolute value of the correlation coefficient of each cell on each date, sort them based on the size of the absolute value, and select the correlation coefficient value of the preset sorting position as the correlation coefficient threshold of the corresponding type of indicator based on the sorting result.

6. A method for river flow inversion based on multi-source optical remote sensing imagery according to any one of claims 1-5, characterized in that, The various indicators include the normalized water index, the new water index, the water turbidity index, the bare soil index, the vegetation index, the normalized humidity index, and the reflectance values ​​of each band in the remote sensing image.

7. An apparatus for river flow inversion based on multi-source optical remote sensing imagery as described in any one of claims 1-6, characterized in that, The device includes: The image preprocessing module is used to acquire multi-source remote sensing images containing the target watershed and to process multiple reference image pairs based on the multi-source remote sensing images. The image preprocessing module is also used to select multiple low-resolution images to be fused from the multi-source remote sensing images, determine a reference image pair that matches the images to be fused, and fuse each image to be fused and the matching reference image pair to obtain a high-resolution fused image. The monitoring pixel acquisition module is used to divide the monitoring area from the target study watershed based on the river flow of all dates, and to filter the monitoring pixels of each type of indicator based on the correlation coefficient between the various indicator values ​​of each pixel in the monitoring area and the river flow. The monitoring pixel acquisition module is also used to acquire the average value of the indicator for all monitoring pixels of each type of indicator on each date, and to construct a river flow inversion model based on the functional relationship between the average value of the indicator and the river flow on the corresponding date. The flow inversion module is used to input the remote sensing image to be predicted into the river flow inversion model to predict the river flow for the corresponding date of the remote sensing image to be predicted.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 6.

Images