Water depth inversion method and device for inland navigable river

Abstract

The invention provides a water depth inversion method and device for an inland navigable river, and belongs to the technical field of remote sensing surveying and mapping, and the method comprises the steps: constructing a water depth sample library of the inland navigable river; the water depth sample library comprises a plurality of water depth sub-sample libraries with different geographic positions and water regimens; obtaining a target geographic position and a target water regimen of the to-be-inverted region, and determining a target water depth sub-sample library in the plurality of water depth sub-sample libraries based on the target geographic position and the target water regimen; constructing a water depth inversion model based on a random forest based on the target water depth sub-sample library; obtaining a to-be-inverted remote sensing image of the to-be-inverted region, and determining a water depth inversion result of the to-be-inverted region based on the to-be-inverted remote sensing image and the water depth inversion model; wherein the water depth inversion result comprises the predicted water depth of each point in the to-be-inverted region. According to the invention, multiple regions and multiple time phases of the sample are realized, so that the precision of the water depth inversion result determined based on the constructed water depth inversion model is improved.

Description

Technical Field

[0001] The invention relates to the technical field of remote sensing mapping, and in particular to a method and device for inverting the water depth of an inland navigable river. Background Art

[0002] Water depth measurement helps to better understand the underwater terrain and is of great significance for inland waterway navigation, water conservancy, and water-related engineering construction. The echo sounding method has strong penetration and high accuracy in water bodies, but it is greatly affected by factors such as weather environment and geographical location, and requires a lot of manpower and material resources. The measurement frequency of some navigable river sections is low, and the observation time interval is long, which is not conducive to timely detection of sudden situations such as sudden siltation of the river channel, and will cause safety hazards to ship navigation. Therefore, it is very important to quickly obtain the changing trend of the water depth of navigable rivers. Remote sensing water depth inversion has become an important means of water depth detection because of its advantages such as wide data sources, no geographical constraints, short cycle, and low comprehensive cost. The current remote sensing water depth inversion models can be divided into three categories: physical models, statistical models, and semi-empirical and semi-theoretical models. The physical model uses the radiation transfer equation to establish the relationship between radiation brightness and water depth. However, this method requires the acquisition of many water body optical parameters, and the acquisition process is too complicated. The semi-empirical plate theoretical model first obtains some parameters through statistical methods, and then uses the theoretical model to assist in inverting the water depth, but the accuracy needs to be improved. The statistical model based on machine learning is the most widely used. It uses the statistical relationship between remote sensing data and the water depth values ​​of known points to invert the water depth values ​​of corresponding locations. The model construction is relatively simple, and the inversion results are also highly reliable.

[0003] However, existing water depth inversion methods are mainly applied to nearshore waters, large lakes, and estuaries, with less attention paid to inland navigable rivers. On the other hand, existing studies mainly construct machine learning models through measured data of a single phase and a single region, and the constructed machine learning models are used for water depth inversion in other regions. It has the following technical problems: water body reflectivity data is easily affected by factors such as water turbidity and underwater topography. The underwater topography and sand content of inland navigable rivers vary greatly, and the water level varies greatly with the seasons. The model trained with samples of a single phase and a single region has poor migration capabilities and cannot fit the entire area well. That is, when the trained machine learning model performs water depth inversion for multiple regions, the inversion results are less accurate.

[0004] Therefore, there is an urgent need to provide a method and device for inland navigable river depth inversion to achieve accurate water depth inversion of inland navigable rivers in different regions and phases. Summary of the invention

[0005] In view of this, it is necessary to provide a method and device for inland navigable river depth inversion to solve the technical problems existing in the prior art, such as poor model transferability and low inversion accuracy due to incomplete sample information and single model.

[0006] On the one hand, in order to solve the above technical problems, the present invention provides a method for inverting the water depth of an inland navigable river, comprising:

[0007] Constructing a water depth sample library for inland navigable rivers; the water depth sample library includes a plurality of water depth sub-sample libraries with different geographical locations and water conditions;

[0008] Acquire a target geographical location and a target water condition of the area to be inverted, and determine a target water depth sub-sample library among the multiple water depth sub-sample libraries based on the target geographical location and the target water condition;

[0009] Building a random forest-based water depth inversion model based on the target water depth subsample library;

[0010] Acquire a remote sensing image of the area to be inverted, and determine a water depth inversion result of the area to be inverted based on the remote sensing image and the water depth inversion model;

[0011] The water depth inversion result includes the predicted water depth of each point in the area to be inverted.

[0012] In a possible implementation, the step of constructing a water depth sample library of an inland navigable river includes:

[0013] Acquire historical measured water depth data, historical water level observation data, and historical remote sensing images corresponding to the historical measured water depth data;

[0014] Determine water body correction data based on the historical water level observation data, and modify the historical measured water depth data based on the water body correction data to obtain corrected historical water depth data, wherein the corrected historical water depth data and the historical remote sensing images constitute the water depth sample library;

[0015] The water depth sample library is divided based on the geographical location and the water conditions to obtain the multiple water depth sub-sample libraries.

[0016] In a possible implementation, the corrected historical water depth data is:

[0017] D RS =D M +ΔH

[0018] ΔH=H RS -H M

[0019] Where D RSis the corrected historical water depth data; ΔH is the water body correction data; H M The historical water level observation data is consistent with the historical measured water depth data date; H RS is the historical water level observation value consistent with the date of the historical remote sensing image; D M It is the historical measured water depth data.

[0020] In a possible implementation manner, before dividing the water depth sample library based on the geographical location and the water conditions, the method further includes:

[0021] Determining overlapping corrected historical water depth data in the corrected historical water depth data that overlaps with the ship in the historical remote sensing image;

[0022] The overlapping corrected historical water depth data is deleted from the water depth sample library.

[0023] In a possible implementation, the step of constructing a random forest-based water depth inversion model based on the target water depth subsample library includes:

[0024] Dividing the target water depth subsample library into training samples and test samples;

[0025] Constructing an initial water depth inversion model based on the training samples;

[0026] The model performance of the initial water depth inversion model is verified based on the test set, and after the verification passes, the initial water depth inversion model is the water depth inversion model.

[0027] In a possible implementation, the water depth inversion model is used to characterize the mapping relationship between multiple image features of the remote sensing image and the predicted water depth; the multiple image features include the red band reflectivity, green band reflectivity, blue band reflectivity, near-infrared band reflectivity of the remote sensing image, the ratio of the red band reflectivity to the green band reflectivity, the ratio of the red band reflectivity to the blue band reflectivity, the ratio of the red band reflectivity to the near-infrared band reflectivity, the ratio of the green band reflectivity to the blue band reflectivity, the ratio of the green band reflectivity to the near-infrared band reflectivity, and the ratio of the blue band reflectivity to the near-infrared band reflectivity.

[0028] In a possible implementation, determining the water depth inversion result of the area to be inverted based on the remote sensing image to be inverted and the water depth inversion model includes:

[0029] Extracting a plurality of image features to be inverted of the remote sensing image to be inverted;

[0030] The multiple image features to be inverted are input into the deep water inversion model to obtain the water depth inversion result.

[0031] In a possible implementation, the method further includes:

[0032] Determine water index characteristics for separating water bodies from non-water bodies;

[0033] Determine the segmentation threshold of the water body and the non-water body based on the maximum inter-class variance method and the water body index feature;

[0034] determining a water body boundary of the water body based on the segmentation threshold;

[0035] The water depth inversion result is clipped based on the water body boundary to obtain a water depth distribution map.

[0036] In a possible implementation, determining a water body index feature for separating a water body from a non-water body includes:

[0037] Acquiring the image quality level of the remote sensing image to be inverted; the image quality level includes a high level and a low level;

[0038] When the image quality level is a high level, the water index feature is a normalized water index;

[0039] When the image quality level is a low level, the water index feature is a dual-polarization water index.

[0040] On the other hand, the present invention also provides a water depth inversion device for an inland navigable river, comprising:

[0041] A water depth sample library construction unit is used to construct a water depth sample library for inland navigable rivers; the water depth sample library includes a plurality of water depth sub-sample libraries with different geographical locations and water conditions;

[0042] A target water depth sub-sample library determining unit is used to obtain a target geographical location and a target water condition of the area to be inverted, and determine a target water depth sub-sample library among the multiple water depth sub-sample libraries based on the target geographical location and the target water condition;

[0043] A water depth inversion model construction unit, used to construct a water depth inversion model based on a random forest based on the target water depth subsample library;

[0044] A water depth inversion unit, used for acquiring the remote sensing image to be inverted of the area to be inverted, and determining the water depth inversion result of the area to be inverted based on the remote sensing image to be inverted and the water depth inversion model;

[0045] The water depth inversion result includes the predicted water depth of each point in the area to be inverted.

[0046] The beneficial effects of the present invention are as follows: the water depth inversion method for inland navigable rivers provided by the present invention constructs multiple water depth sub-sample libraries including different geographical locations and water conditions. When performing water depth inversion on the remote sensing image to be inverted in the inversion area, the target water depth sub-sample library in the multiple water depth sub-sample libraries is determined based on the target geographical location and target water conditions of the area to be inverted, and the influence of region and time on water depth inversion is considered, so as to achieve the adaptability of the constructed water depth inversion model to the area to be inverted, and improve the accuracy of the inversion result based on the water depth inversion model. In addition, the constructed water depth sample library includes water depth sub-sample libraries with different geographical locations and different water depths, and multiple water depth inversion models can be obtained based on the adaptability of different water depth sub-sample libraries, thereby improving the migration ability of the water depth inversion model.

[0047] Furthermore, the present invention can obtain the predicted water depth by inverting the remote sensing image to be inverted, thereby realizing the rapid acquisition of the water depth of the area to be inverted in the absence of the measured water depth, avoiding sudden situations such as sudden siltation of the river channel, and ensuring the navigation safety of ships in inland navigable rivers. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0049] Figure 1 A schematic flow chart of an embodiment of a method for inverting the water depth of an inland navigable river provided by the present invention;

[0050] Figure 2 For the present invention Figure 1 A schematic flow chart of an embodiment of step S101;

[0051] Figure 3 For the present invention Figure 1 A schematic flow chart of an embodiment of step S103;

[0052] Figure 4 A schematic diagram of an embodiment of the evaluation index results of the water depth inversion model provided by the present invention;

[0053] Figure 5 For the present invention Figure 1 A schematic flow chart of an embodiment of step S104;

[0054] Figure 6 A schematic diagram of a flow chart of an embodiment of determining a water depth distribution map provided by the present invention;

[0055] Figure 7 For the present invention Figure 6 A schematic flow chart of an embodiment of step S601;

[0056] Figure 8 A schematic diagram of an embodiment of a water depth distribution map determined by the present invention;

[0057] Fig. 9 A schematic diagram of the structure of an embodiment of a water depth inversion device for an inland navigable river provided by the present invention;

[0058] Fig.10 A schematic structural diagram of an embodiment of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0059] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0060] It should be understood that the schematic drawings are not drawn to scale. The flowchart used in the present invention shows the operations implemented according to some embodiments of the present invention. It should be understood that the operations of the flowchart can be implemented out of order, and the steps without logical context can be reversed in order or implemented simultaneously. In addition, those skilled in the art can add one or more other operations to the flowchart under the guidance of the content of the present invention, and can also remove one or more operations from the flowchart. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0061] Reference to an "embodiment" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiment may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0062] The present invention provides a method and device for inverting the water depth of an inland navigable river, which are respectively described below.

[0063] Figure 1 A schematic flow chart of an embodiment of the method for inverting the water depth of an inland navigable river provided by the present invention is shown in FIG. Figure 1 As shown in the figure, the depth inversion methods for inland navigable rivers include:

[0064] S101. Construct a water depth sample library for inland navigable rivers; the water depth sample library includes multiple water depth sub-sample libraries with different geographical locations and water conditions;

[0065] S102, obtaining a target geographical location and a target water condition of the area to be inverted, and determining a target water depth sub-sample library among a plurality of water depth sub-sample libraries based on the target geographical location and the target water condition;

[0066] S103, constructing a water depth inversion model based on random forest based on the target water depth subsample library;

[0067] S104, obtaining a remote sensing image of the area to be inverted, and determining a water depth inversion result of the area to be inverted based on the remote sensing image and the water depth inversion model;

[0068] The water depth inversion result includes the predicted water depth of each point in the area to be inverted.

[0069] It should be noted that the geographical location division standard is consistent with the waterway division standard of inland waterways, and is divided according to the geographical environment and hydrological characteristics of the river; the water regime division standard refers to the inland waterway division standard for flood season, dry season and medium water season, and is divided according to the water level and flow rate of the river.

[0070] In a specific embodiment of the present invention, the target geographical location of the area to be inverted is the Hukou waterway in the main stream of the Yangtze River, and the target water condition is the dry season.

[0071] Compared with the prior art, the method for inland navigable river depth inversion provided by the embodiment of the present invention constructs multiple depth sub-sample libraries including different geographical locations and water conditions. When performing depth inversion on the remote sensing image to be inverted in the inversion area, the target depth sub-sample library in the multiple depth sub-sample libraries is determined based on the target geographical location and target water conditions of the area to be inverted, and the influence of region and time on the depth inversion is considered. The adaptability of the constructed depth inversion model and the area to be inverted is achieved, and the accuracy of the inversion result based on the depth inversion model is improved. In addition, the constructed depth sample library includes depth sub-sample libraries with different geographical locations and different water depths, and multiple depth inversion models can be obtained based on the adaptability of different depth sub-sample libraries, thereby improving the migration ability of the depth inversion model.

[0072] Furthermore, the embodiments of the present invention can obtain the predicted water depth by inverting the remote sensing image to be inverted, thereby realizing the rapid acquisition of the water depth of the area to be inverted in the absence of the measured water depth, avoiding sudden situations such as sudden siltation of the river channel, and ensuring the navigation safety of ships in inland navigable rivers.

[0073] Since the accuracy of the water depth label value in the water depth sample library is crucial to the prediction accuracy of the water depth inversion model, in order to ensure the accuracy of the constructed water depth sample library, in some embodiments of the present invention, Figure 2 As shown, step S101 includes:

[0074] S201, obtaining historical measured water depth data, historical water level observation data, and historical remote sensing images corresponding to the historical measured water depth data;

[0075] S202, determining water body correction data based on historical water level observation data, and modifying historical measured water depth data based on the water body correction data to obtain corrected historical water depth data, and the corrected historical water depth data and historical remote sensing images constitute a water depth sample library;

[0076] S203: Divide the water depth sample library based on the geographical location and water conditions to obtain multiple water depth sub-sample libraries.

[0077] Since the historical measured data and the historical remote sensing images are collected at different time points, and the water depth will change dynamically with time, the embodiment of the present invention modifies the historical measured data based on the historical water level observation data to obtain the corrected historical water depth data corresponding to the historical remote sensing image time, thereby ensuring the accurate correspondence between the historical remote sensing images and the corrected historical water depth data, and further ensuring the accuracy of the water depth prediction model constructed using the water depth sub-sample library, thereby improving the accuracy and precision of the inversion results.

[0078] The historical measured water depth data is obtained as follows: the measurement method is field measurement using a shipborne single-beam depth sounder. The equipment models are HD-MAX, HY1601, and HY1603. The coordinate system used is CGCS2000. The depth preprocessing software supporting the measuring equipment is used to remove measurement anomalies.

[0079] The historical water level observation data is obtained by recording the water gauge observation values ​​using the water gauge as the measurement method.

[0080] The historical remote sensing images are obtained by selecting Sentinel-2 L2.

[0081] In a specific embodiment of the present invention, the historical water depth data is corrected as follows:

[0082] D RS =D M +ΔH

[0083] ΔH=H RS -H M

[0084] Where D RS is the corrected historical water depth data; ΔH is the water body correction data; H MThe historical water level observation data is consistent with the historical measured water depth data date; H RS is the historical water level observation value consistent with the date of the historical remote sensing image; D M It is the historical measured water depth data.

[0085] Since there are ships passing through inland navigable rivers, in order to avoid adverse effects on samples in the water depth sample library when the ships overlap with the water depth measurement points, in some embodiments of the present invention, before step S203, the following is further included:

[0086] Determine the overlapped corrected historical water depth data in the corrected historical water depth data with the ships in the historical remote sensing image;

[0087] The overlapping corrected historical water depth data are deleted from the water depth sample library.

[0088] The embodiment of the present invention avoids the pollution of the water depth sample library by the existence of the ship by deleting the overlapping corrected historical water depth data overlapping with the ship, thereby ensuring the accuracy of the water depth sample library.

[0089] In some embodiments of the present invention, Figure 3 As shown, step S103 includes:

[0090] S301, dividing the target water depth sub-sample library into training samples and test samples;

[0091] S302, constructing an initial water depth inversion model based on training samples;

[0092] S303: verifying the model performance of the initial water depth inversion model based on the test set. After passing the verification, the initial water depth inversion model is the water depth inversion model.

[0093] First of all, it should be noted that the water depth inversion model in the embodiment of the present invention constructs a mapping relationship between multiple image features of remote sensing images and predicted water depth. Therefore, each sample data in the target water depth subsample library includes multiple image features of multiple historical remote sensing images and corresponding water depth labels.

[0094] That is, it is necessary to extract features from historical remote sensing images to obtain multiple image features.

[0095] It should also be noted that in order to ensure the accuracy of the extracted multiple image features, the historical remote sensing images need to be preprocessed before feature extraction. The preprocessing methods include but are not limited to radiation correction and atmospheric correction.

[0096] In a specific embodiment of the present invention, the ratio of the training set to the test set in step S301 is 7:3.

[0097] Before step S301, it is also necessary to perform standardization processing on multiple image features to convert the image features into a standard normal distribution, and the conversion function is:

[0098]

[0099] In the formula, x * is the converted image feature, μ is the mean of all sample data, and σ is the standard deviation of all sample data.

[0100] Specifically, step S302 includes: using the Bagging algorithm and the Bootstrap resampling technique to randomly select multiple sub-training sets from the training samples with replacement. During the sampling process, the same sample may be selected multiple times. The size of the sub-training set is usually the same as the size of the original data set. Each sub-training set is used to build a decision tree.

[0101] When training a single decision tree, the steps can be divided into three steps:

[0102] The first step is to select the splitting features and splitting points. The entire training subset (the subset generated by Bootstrap sampling) is regarded as the data of the root node. For each node split, a part of the features is randomly selected. For each candidate feature, all possible splitting points are tried. The current node is taken as the root node, and it is divided into two left and right child nodes according to the current features and splitting points. The variance reduction is used to evaluate the splitting effect. The variance of the current node and the total weighted variance of the left and right nodes after the current node is split are calculated, and the variance reduction is calculated. The specific calculation process is as follows:

[0103] ΔVar=Var(t)-Var split

[0104]

[0105] N=N L +N R

[0106] Where ΔVar is the variance reduction, Var(t) is the node variance before splitting, and Var split is the weighted variance after splitting, N is the number of samples of the current node, y i is the target variable of the i-th sample, is the mean value of the target variable of the current node, N L and N R are the number of samples of the left and right child nodes after the current node splits, Var(t L ) and Var(t R ) are the variances of the left and right child nodes respectively, and the calculation method is similar to that of the root node. The features and split points that reduce the variance the most are selected as the optimal split features and nodes.

[0107] The second is data division. According to the selected optimal splitting node, the data of the current node is divided into two sub-nodes.

[0108] The third is to recursively construct child nodes. Repeat the following steps for each child node: (1) Calculate the mean of the node (used as the final prediction value); (2) Check the stopping condition. If the minimum number of samples, maximum depth, impurity and other parameters of the current node meet the splitting stopping condition, terminate the split and the decision tree is completed; (3) If the current node does not meet the splitting stopping condition, repeat steps 1 and 2.

[0109] Repeat the above operation on each decision tree to build a forest and obtain the initial water depth inversion model.

[0110] In a specific embodiment of the present invention, step S303 is: using a cross-validation method to optimize the model parameters of the maximum depth of the decision tree and the minimum number of node samples in the initial water depth inversion model. The image features of the test sample are input into the optimized initial water depth inversion model. For the newly input measured water depth sample, each decision tree will generate an inverted water depth value, and the final water depth inversion result is obtained by averaging the inversion results of multiple decision trees.

[0111] The evaluation indicators selected by the embodiment of the present invention include root mean square error (RMSE) and determination coefficient (R 2 ), the least squares method was used to fit the inversion model, RMSE and R 2 The calculation method is as follows:

[0112]

[0113] In the formula, y i is the measured water depth of the i-th test point; P i is the inverted water depth of the i-th test point; is the average measured water depth of the test point, and n is the number of test samples. RMSE reflects the deviation between the inverted water depth value and the measured value. The smaller the RMSE, the better the inversion effect. 2 It is the Pearson correlation coefficient between the measured water depth and the inverted water depth. The larger the value, the better the inversion effect.

[0114] In a specific embodiment of the present invention, the evaluation results of the two evaluation indicators are as follows: Figure 4 As shown by Figure 4 It can be seen that: R obtained by the method proposed in the embodiment of the present invention 2 It reaches 0.94, and the RMSE is 1.19m, which shows that the embodiment of the present invention can better invert the water depth of inland navigable rivers in the absence of measured data, realize rapid monitoring of river depth change trends, and ensure the safety of ship navigation.

[0115] In a specific embodiment of the present invention, the multiple image features include the red band reflectivity, green band reflectivity, blue band reflectivity, near infrared band reflectivity, the ratio of the red band reflectivity to the green band reflectivity, the ratio of the red band reflectivity to the blue band reflectivity, the ratio of the red band reflectivity to the near infrared band reflectivity, the ratio of the green band reflectivity to the blue band reflectivity, the ratio of the green band reflectivity to the near infrared band reflectivity, and the ratio of the blue band reflectivity to the near infrared band reflectivity of the remote sensing image.

[0116] The embodiment of the present invention improves the diversity and comprehensiveness of image features by taking the red band reflectivity, green band reflectivity, blue band reflectivity, near-infrared band reflectivity and their pairwise ratios as image features, thereby improving the precision and accuracy of the water depth inversion model constructed based on multiple image features, and thereby improving the precision and accuracy of the inversion results of the inversion area to be inverted.

[0117] In some embodiments of the present invention, Figure 5 As shown, step S104 includes:

[0118] S501, extracting multiple image features to be inverted of the remote sensing image to be inverted;

[0119] S502: Input multiple image features to be inverted into a deep water inversion model to obtain a water depth inversion result.

[0120] Similarly, in order to improve the accuracy of the extracted features of the image to be inverted, before step S501, the process further includes: preprocessing the remote sensing image to be inverted, and the preprocessing includes but is not limited to radiation correction and atmospheric correction.

[0121] Since the entire remote sensing image is inverted in step S104, the inversion result is the entire remote sensing image rectangular area, which includes not only water bodies but also non-water bodies, while the water depth inversion only focuses on the water body area. Therefore, in some embodiments of the present invention, Figure 6 As shown in Figure 2, the depth inversion method for inland navigable rivers also includes:

[0122] S601, determining a water body index feature for separating water bodies and non-water bodies;

[0123] S602, determining the segmentation threshold of water body and non-water body based on the maximum between-class variance method (OTSU) and water body index characteristics;

[0124] S603, determining the water body boundary of the water body based on the segmentation threshold;

[0125] S604: Clip the water depth inversion result based on the water body boundary to obtain a water depth distribution map.

[0126] The embodiment of the present invention clips the water depth inversion result by determining the segmentation threshold, so as to obtain a water depth distribution map including only the water body area. The water depth distribution map can be used to understand the changes of the water body area with factors such as seasons, providing a reference for subsequent analysis.

[0127] The specific solution process of step S602 is as follows:

[0128] T=argmaxσ 2 (k)

[0129] σ 2 (k) = ω0(μ-μ0) 2 +ω1(μ-μ1) 2

[0130] In the formula, the segmentation threshold is T, pixels less than T are water bodies, the number accounts for ω0 of the total image, and the average grayscale is μ0; pixels greater than T are non-water bodies, the number accounts for ω1, and the average grayscale is μ1, σ 2 (k) calculates the maximum variance between the two categories, σ 2 (k) T corresponding to the maximum value is the segmentation threshold, which is automatically obtained by the OTSU algorithm.

[0131] Specifically, step S603 includes: performing denoising, raster-to-vector conversion and other operations on the determined water body to obtain the water body boundary. Specifically, the density of ships in inland navigable rivers is large, which is shown as small spots in the water body extraction result. The noise points in the water body extraction result are removed based on the 8-adjacent pixel connectivity, and the post-processed water body raster data is converted into vector data to obtain the water body boundary.

[0132] Since the water body index feature is used as the basis for determining the segmentation threshold, the accuracy of the water body index feature is crucial to the accurate determination of the segmentation threshold. To achieve this goal, in some embodiments of the present invention, Figure 7 As shown, step S601 includes:

[0133] S701, obtaining the image quality level of the remote sensing image to be inverted; the image quality level includes a high level and a low level;

[0134] S702, when the image quality level is high, the water index feature is the Normalized Difference Water Index (NDWI);

[0135] S703: When the image quality level is low, the water index feature is a dual-polarized water index (SDWI).

[0136] The embodiment of the present invention can ensure the accuracy of the selected water index feature by selecting different water index features corresponding to the acquired image quality level. Specifically, when there is no cloud coverage in the area outside the river channel in the remote sensing image to be inverted, the NDWI that can be directly obtained is selected, and when there is a lot of cloud coverage in the area outside the river channel in the remote sensing image to be inverted, the SDWI corresponding to the radar remote sensing image with stronger penetration is selected, ensuring that the selected water index feature can accurately determine the segmentation threshold.

[0137] Among them, NDWI is:

[0138]

[0139] Where Green and NIR are the reflectance of the green band and near infrared band of the multispectral remote sensing image, respectively.

[0140] SDWI is:

[0141] SDWI=ln(10·VV·VH)-8

[0142] Where VV and VH are the VV and VH backscattering intensities of Sentinel-1, respectively.

[0143] In a specific embodiment of the present invention, the water depth distribution map obtained is as follows: Figure 8 As shown by Figure 8 It can be seen that the water depth inversion method for inland navigable rivers proposed in the embodiment of the present invention can effectively obtain a water depth distribution map.

[0144] In order to better implement the water depth inversion method for an inland navigable river in the embodiment of the present invention, based on the water depth inversion method for an inland navigable river, the embodiment of the present invention also provides a water depth inversion device for an inland navigable river, such as Fig. 9 As shown, the water depth inversion device 900 for an inland navigable river includes:

[0145] A water depth sample library construction unit 901 is used to construct a water depth sample library for inland navigable rivers; the water depth sample library includes multiple water depth sub-sample libraries with different geographical locations and water conditions;

[0146] A target water depth sub-sample library determining unit 902 is used to obtain a target geographical location and a target water condition of the area to be inverted, and determine a target water depth sub-sample library among multiple water depth sub-sample libraries based on the target geographical location and the target water condition;

[0147] A water depth inversion model building unit 903 is used to build a water depth inversion model based on a random forest based on a target water depth subsample library;

[0148] The water depth inversion unit 904 is used to obtain the remote sensing image to be inverted of the area to be inverted, and determine the water depth inversion result of the area to be inverted based on the remote sensing image to be inverted and the water depth inversion model;

[0149] The water depth inversion result includes the predicted water depth of each point in the area to be inverted.

[0150] The depth inversion device 900 for an inland navigable river provided in the above embodiment can implement the technical solution described in the above embodiment of the depth inversion method for an inland navigable river. The specific implementation principles of the above modules or units can refer to the corresponding contents in the above embodiment of the depth inversion method for an inland navigable river, which will not be repeated here.

[0151] like Fig.10 As shown, the present invention also provides an electronic device 1000. The electronic device 1000 includes a processor 1001, a memory 1002 and a display 1003. Fig.10 Only some components of the electronic device 1000 are shown, but it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0152] In some embodiments, the processor 1001 may be a central processing unit (CPU), a microprocessor or other data processing chip, used to run program codes or process data stored in the memory 1002, such as the water depth inversion method for inland navigable rivers in the present invention.

[0153] In some embodiments of the present invention, processor 1001 may be a single server or a server group. The server group may be centralized or distributed. In some embodiments, processor 1001 may be local or remote. In some embodiments, processor 1001 may be implemented in a cloud platform. In one embodiment, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an internal cloud, a multi-cloud, etc., or any combination thereof.

[0154] In some embodiments, the memory 1002 may be an internal storage unit of the electronic device 1000, such as a hard disk or memory of the electronic device 1000. In other embodiments, the memory 1002 may also be an external storage device of the electronic device 1000, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device 1000.

[0155] Furthermore, the memory 1002 may include both an internal storage unit of the electronic device 1000 and an external storage device. The memory 1002 is used to store application software installed in the electronic device 1000 and various data.

[0156] In some embodiments, the display 1003 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, etc. The display 1003 is used to display information of the electronic device 1000 and to display a visual user interface. The components 1001-1003 of the electronic device 1000 communicate with each other via a device bus.

[0157] In some embodiments of the present invention, when the processor 1001 executes the water depth inversion program of the inland navigable river in the memory 1002, the following steps may be implemented:

[0158] Construct a water depth sample library for inland navigable rivers; the water depth sample library includes multiple water depth sub-sample libraries with different geographical locations and water conditions;

[0159] Obtaining a target geographical location and a target water condition of the area to be inverted, and determining a target water depth sub-sample library among multiple water depth sub-sample libraries based on the target geographical location and the target water condition;

[0160] A random forest-based water depth inversion model is constructed based on the target water depth subsample library;

[0161] Acquire a remote sensing image of the area to be inverted, and determine a water depth inversion result of the area to be inverted based on the remote sensing image and the water depth inversion model;

[0162] The water depth inversion result includes the predicted water depth of each point in the area to be inverted.

[0163] It should be understood that: when the processor 1001 executes the water depth inversion program of the inland navigable river in the memory 1002, in addition to the above functions, other functions can also be implemented. For details, please refer to the description of the corresponding method embodiment above.

[0164] Correspondingly, an embodiment of the present invention also provides a computer-readable storage medium, which is used to store computer-readable programs or instructions. When the program or instructions are executed by a processor, the steps or functions of the depth inversion method for inland navigable rivers provided in the above-mentioned method embodiments can be implemented.

[0165] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by instructing related hardware (such as a processor, a controller, etc.) through a computer program, and the computer program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0166] The above is a detailed introduction to a method and device for depth inversion of an inland navigable river provided by the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for technical personnel in this field, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A method for inverting the water depth of an inland navigable river, characterized in that: include: Constructing a water depth sample library for inland navigable rivers; the water depth sample library includes a plurality of water depth sub-sample libraries with different geographical locations and water conditions; Acquire a target geographical location and a target water condition of the area to be inverted, and determine a target water depth sub-sample library among the multiple water depth sub-sample libraries based on the target geographical location and the target water condition; Building a random forest-based water depth inversion model based on the target water depth subsample library; Acquire a remote sensing image of the area to be inverted, and determine a water depth inversion result of the area to be inverted based on the remote sensing image and the water depth inversion model; The water depth inversion result includes the predicted water depth of each point in the area to be inverted.

2. The method for inverting the water depth of an inland navigable river according to claim 1, characterized in that: The construction of the water depth sample library of inland navigable rivers includes: Acquire historical measured water depth data, historical water level observation data, and historical remote sensing images corresponding to the historical measured water depth data; Determine water body correction data based on the historical water level observation data, and modify the historical measured water depth data based on the water body correction data to obtain corrected historical water depth data, wherein the corrected historical water depth data and the historical remote sensing images constitute the water depth sample library; The water depth sample library is divided based on the geographical location and the water conditions to obtain the multiple water depth sub-sample libraries.

3. The method for inverting the water depth of an inland navigable river according to claim 2, characterized in that: The corrected historical water depth data is: D RS =D M +ΔH ΔH=H RS -H M Where D RS is the corrected historical water depth data; ΔH is the water body correction data; H M The historical water level observation data is consistent with the historical measured water depth data date; H RS is the historical water level observation value consistent with the date of the historical remote sensing image; D M It is the historical measured water depth data.

4. The method for inverting the water depth of an inland navigable river according to claim 2, characterized in that: Before dividing the water depth sample library based on the geographical location and the water conditions, the method further includes: Determining overlapping corrected historical water depth data in the corrected historical water depth data that overlaps with the ship in the historical remote sensing image; The overlapping corrected historical water depth data is deleted from the water depth sample library.

5. The method for inverting the water depth of an inland navigable river according to claim 1, characterized in that: The step of constructing a random forest-based water depth inversion model based on the target water depth subsample library includes: Dividing the target water depth subsample library into training samples and test samples; Constructing an initial water depth inversion model based on the training samples; The model performance of the initial water depth inversion model is verified based on the test set, and after the verification passes, the initial water depth inversion model is the water depth inversion model.

6. The method for inverting the water depth of an inland navigable river according to claim 5, characterized in that: The water depth inversion model is used to characterize the mapping relationship between multiple image features of the remote sensing image and the predicted water depth; the multiple image features include the red band reflectivity, green band reflectivity, blue band reflectivity, near-infrared band reflectivity, the ratio of the red band reflectivity to the green band reflectivity, the ratio of the red band reflectivity to the blue band reflectivity, the ratio of the red band reflectivity to the near-infrared band reflectivity, the ratio of the green band reflectivity to the blue band reflectivity, the ratio of the green band reflectivity to the near-infrared band reflectivity, and the ratio of the blue band reflectivity to the near-infrared band reflectivity.

7. The method for inverting the water depth of an inland navigable river according to claim 6, characterized in that: The step of determining the water depth inversion result of the area to be inverted based on the remote sensing image to be inverted and the water depth inversion model comprises: Extracting a plurality of image features to be inverted of the remote sensing image to be inverted; The multiple image features to be inverted are input into the deep water inversion model to obtain the water depth inversion result.

8. The method for inverting the water depth of an inland navigable river according to claim 1, characterized in that: The method further comprises: Determine water index characteristics for separating water bodies from non-water bodies; Determine the segmentation threshold of the water body and the non-water body based on the maximum inter-class variance method and the water body index feature; determining a water body boundary of the water body based on the segmentation threshold; The water depth inversion result is clipped based on the water body boundary to obtain a water depth distribution map.

9. The method for inverting the water depth of an inland navigable river according to claim 8, characterized in that: The water index characteristics for separating water bodies and non-water bodies are determined, including: Acquiring the image quality level of the remote sensing image to be inverted; the image quality level includes a high level and a low level; When the image quality level is a high level, the water index feature is a normalized water index; When the image quality level is a low level, the water index feature is a dual-polarization water index.

10. A water depth inversion device for an inland navigable river, characterized in that: include: A water depth sample library construction unit is used to construct a water depth sample library for inland navigable rivers; the water depth sample library includes a plurality of water depth sub-sample libraries with different geographical locations and water conditions; A target water depth sub-sample library determining unit is used to obtain a target geographical location and a target water condition of the area to be inverted, and determine a target water depth sub-sample library among the multiple water depth sub-sample libraries based on the target geographical location and the target water condition; A water depth inversion model construction unit, used to construct a water depth inversion model based on a random forest based on the target water depth subsample library; A water depth inversion unit, used for acquiring the remote sensing image to be inverted of the area to be inverted, and determining the water depth inversion result of the area to be inverted based on the remote sensing image to be inverted and the water depth inversion model; The water depth inversion result includes the predicted water depth of each point in the area to be inverted.

Images