A method and system for detecting ship exhaust gas

By constructing a ship exhaust gas detection model and dynamically regulating ship power using remote sensing images and power output values, the problem that the existing technology cannot effectively reduce ship exhaust gas emissions is solved, and the reduction of environmental pollution and controllability of waste gas emissions is achieved.

CN118781108BActive Publication Date: 2024-11-15GUANGDONG OCEAN UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411259176.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-10
Publication Date
2024-11-15
Estimated Expiration
2044-09-10

AI Technical Summary

Technical Problem

Existing ship exhaust gas detection methods cannot reduce the emission of polluted gases from the root, and due to the performance of purification devices and pipeline instruments, it cannot effectively control the exhaust gas emissions, resulting in environmental pollution.

Method used

By obtaining the remote sensing image set of ship exhaust gas and power output values, a waste gas detection model is constructed, and the ship's power output is dynamically regulated to adjust the waste gas emissions, ensuring that the emissions meet the standard values, and reducing environmental pollution.

Benefits of technology

It has achieved the control of ship exhaust gas emissions from the root, reduce environmental pollution, dynamically adjust power output to keep exhaust gas emissions within a controllable range, and reduce environmental pollution.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118781108B_ABST
    Figure CN118781108B_ABST
Patent Text Reader

Abstract

The present invention discloses a method and system for detecting ship exhaust gas, the method comprising: obtaining a ship exhaust gas remote sensing image set at a current moment, a current ship exhaust gas emission standard value and a current ship power output value; wherein the ship exhaust gas remote sensing image set comprises: a plurality of ship exhaust gas remote sensing images taken at multiple angles; inputting the ship exhaust gas remote sensing image set at a current moment into an exhaust gas detection model, so that the exhaust gas detection model outputs the ship exhaust gas emission at a current moment; determining a ship power control amount according to the ship exhaust gas emission at a current moment, the current ship exhaust gas emission standard value and the current ship power output value; adjusting the ship power output value at a next moment according to the ship power control amount; by implementing the present invention, the generation of ship exhaust gas emission can be regulated from the root, thereby reducing pollution to the environment.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of exhaust gas detection, and in particular to a method and system for detecting ship exhaust gas. Background Art

[0002] The emission of harmful substances such as sulfur oxides (SOx), nitrogen oxides (NOx), carbon oxides (COx) and particulate matter (PM) contained in ship exhaust gas is a major source of air pollution. With the increasing importance of environmental protection issues, reducing the emission of harmful gases has become an important development direction for ship exhaust treatment in the world today. The existing ship exhaust gas detection method mainly monitors the pipeline exhaust gas by deploying detection instruments on the ship. For example, in the patent CN111948169B, a ship exhaust online monitoring and analysis system, the real-time detection of exhaust gas is mainly achieved by setting up a flue gas collection instrument and a flue gas analysis instrument on the exhaust pipe to detect the gas to be discharged in the ship pipeline. However, the existing method mainly detects the gas to be discharged in the pipeline, and then purifies the gas before the gas is discharged. However, this method cannot reduce the emission of polluted gases from the source, and is limited by the performance of the purification device and pipeline instruments. The most direct way to reduce exhaust emissions is to reduce the power output of the ship and reduce exhaust emissions by dynamically adjusting the power output of the ship. Therefore, there is an urgent need for a method to reduce ship exhaust emissions from the source so that exhaust emissions are always within a controllable range, thereby reducing pollution to the environment. Summary of the invention

[0003] The embodiments of the present invention provide a method and system for detecting ship exhaust gas, which can regulate the generation of ship exhaust gas emissions from the source and reduce pollution to the environment.

[0004] An embodiment of the present invention provides a method for detecting ship exhaust gas, comprising:

[0005] Acquire a ship exhaust remote sensing image set at the current moment, the exhaust emission standard value of the current ship, and the current ship power output value; wherein the ship exhaust remote sensing image set includes: a plurality of ship exhaust remote sensing images taken from multiple angles;

[0006] Inputting the current ship exhaust remote sensing image set into the exhaust detection model so that the exhaust detection model outputs the current ship exhaust emission;

[0007] Determine the ship power control amount according to the ship exhaust gas emission at the current moment, the current ship exhaust gas emission standard value and the current ship power output value;

[0008] Adjusting the power output value of the ship at the next moment according to the ship power control amount;

[0009] Wherein, the construction of the exhaust gas detection model includes:

[0010] Acquire a sample set of ship exhaust remote sensing images; wherein the sample set of ship exhaust remote sensing images includes a subset of ship exhaust remote sensing images at a number of historical moments; each subset of ship exhaust remote sensing images includes a number of ship exhaust remote sensing image samples taken at multiple angles;

[0011] Constructing an initial exhaust gas detection model;

[0012] The initial exhaust gas detection model is iteratively trained using the ship exhaust gas remote sensing image sample set until the initial exhaust gas detection model reaches a preset convergence condition, thereby generating an exhaust gas detection model.

[0013] Further, the initial exhaust gas detection model includes: a classification module, a conversion module, a first detection module, a second detection module, a feature fusion module and a correction module;

[0014] The iterative training of the initial exhaust gas detection model using the ship exhaust gas remote sensing image sample set includes:

[0015] For each round of iterative training, a subset of ship exhaust remote sensing image samples at a selected historical moment is input into a classification module, so that the classification module extracts a first image feature of each ship exhaust remote sensing image sample in the subset of ship exhaust remote sensing images at the selected historical moment; wherein the first image feature includes: image background environment and image background color;

[0016] Determine the image type of each ship exhaust remote sensing image sample according to the first image feature; wherein the image type includes: daytime image and nighttime image;

[0017] Input each ship exhaust remote sensing image sample whose image type is night image into the conversion module, so that the conversion module converts each ship exhaust remote sensing image sample whose image type is night image into a ship exhaust remote sensing image sample whose image type is daytime image, and inputs each converted ship exhaust remote sensing image sample into the first detection module and the second detection module respectively;

[0018] Inputting each ship exhaust remote sensing image sample whose image type is a daytime image into the first detection module and the second detection module respectively;

[0019] The first detection module extracts the second image features of each ship exhaust remote sensing image sample one by one, and inputs the second image features corresponding to each ship exhaust remote sensing image sample into the feature fusion module; wherein the second image features include: exhaust gas group shape, exhaust gas group area color and exhaust gas group concentration;

[0020] The second detection module determines the area where the three-dimensional model of the exhaust gas mass is located and the three-dimensional model of the exhaust gas mass according to the exhaust gas remote sensing image samples of each ship, extracts the second image features corresponding to the area where the three-dimensional model of the exhaust gas mass is located and the second image features corresponding to the three-dimensional model of the exhaust gas mass, and inputs the second image features corresponding to the area where the three-dimensional model of the exhaust gas mass is located and the second image features corresponding to the three-dimensional model of the exhaust gas mass into the feature fusion module;

[0021] The feature fusion module fuses the second image features corresponding to each ship exhaust remote sensing image sample, the second image features corresponding to the area where the exhaust gas group three-dimensional model is located, and the second image features corresponding to the exhaust gas group three-dimensional model to obtain a first fusion feature, and determines the predicted ship exhaust emissions of the ship exhaust remote sensing image sample subset at the selected historical moment according to the first fusion feature, and inputs the predicted ship exhaust emissions into the correction module;

[0022] The correction module determines the prediction error based on the subset of ship exhaust remote sensing image samples at the selected historical moment and the predicted ship exhaust emissions, and adjusts the network parameters of the initial exhaust detection model according to the prediction error.

[0023] Furthermore, the step of determining the region where the exhaust gas cluster three-dimensional model is located and the exhaust gas cluster three-dimensional model according to the exhaust gas remote sensing image samples of each ship includes:

[0024] Selecting a ship exhaust gas remote sensing image sample from each ship exhaust gas remote sensing image sample as a selected ship exhaust gas remote sensing image sample;

[0025] Select the two-dimensional region where the exhaust gas mass is located in the selected ship exhaust gas remote sensing image sample, establish a spatial rectangular coordinate system with the centroid of the two-dimensional region as the origin, and select a three-dimensional region that can fully cover the two-dimensional region where the exhaust gas mass is located on the spatial rectangular coordinate system, and use the three-dimensional region as the region where the initial exhaust gas mass three-dimensional model is located;

[0026] Determining a first relative position relationship between the shooting angle of the selected ship exhaust remote sensing image sample and the origin of the spatial rectangular coordinate system according to the shooting angle of the selected ship exhaust remote sensing image sample;

[0027] The remaining ship exhaust remote sensing image samples except the selected ship exhaust remote sensing image samples are used as the complementary ship exhaust remote sensing image samples;

[0028] For each completed ship exhaust remote sensing image sample, determining a second relative position relationship between the shooting angle of the completed ship exhaust remote sensing image sample and the origin of the spatial rectangular coordinate system according to the shooting angle of the completed ship exhaust remote sensing image sample;

[0029] Correcting the area where the initial exhaust gas mass three-dimensional model is located according to the first relative position relationship, each second relative position relationship and each completed ship exhaust gas remote sensing image sample to determine the area where the exhaust gas mass three-dimensional model is located;

[0030] The three-dimensional model of the exhaust gas mass is determined according to the area where the three-dimensional model of the exhaust gas mass is located.

[0031] Furthermore, the region where the initial exhaust gas cluster three-dimensional model is located is corrected according to the first relative position relationship, each second relative position relationship and each completed ship exhaust gas remote sensing image sample, including:

[0032] For each completed ship exhaust remote sensing image sample, projection is performed in the spatial rectangular coordinate system where the initial exhaust gas group three-dimensional model is located according to the second relative position relationship to obtain a first projection image of the completed ship exhaust remote sensing image sample on the horizontal plane of the spatial rectangular coordinate system, a second projection image on the vertical plane, and a third projection image on the lateral plane;

[0033] The area where the initial exhaust gas mass three-dimensional model is located is corrected according to the first relative position relationship and the first projection image, the second projection image and the third projection image.

[0034] Furthermore, the step of correcting the area where the initial exhaust gas mass three-dimensional model is located according to the first relative position relationship and the first projection image, the second projection image, and the third projection image includes:

[0035] Determine a first corrected three-dimensional model according to the first relative position relationship, the first projection image, and the second projection image;

[0036] Determine a second corrected three-dimensional model according to the first relative position relationship, the first projection image, and the third projection image;

[0037] Determine a third corrected three-dimensional model according to the first relative position relationship, the second projection image and the third projection image;

[0038] Based on the first relative position relationship, superimposing the first corrected three-dimensional model and the second corrected three-dimensional model to obtain a first superimposed corrected three-dimensional model, superimposing the first corrected three-dimensional model and the third corrected three-dimensional model to obtain a second superimposed corrected three-dimensional model, and superimposing the second corrected three-dimensional model and the third corrected three-dimensional model to obtain a third superimposed corrected three-dimensional model;

[0039] Based on the first relative position relationship, fitting the first corrected three-dimensional model, the second corrected three-dimensional model and the third corrected three-dimensional model to obtain a fitted corrected three-dimensional model;

[0040] The fitted corrected three-dimensional model is adjusted by using the first superimposed corrected three-dimensional model, the second superimposed corrected three-dimensional model and the third superimposed corrected three-dimensional model to obtain a three-dimensional model that completes the ship exhaust remote sensing image sample;

[0041] The area where the initial exhaust gas cluster three-dimensional model is located is corrected according to the three-dimensional model of the ship exhaust gas remote sensing image sample.

[0042] Furthermore, the second image features corresponding to the remote sensing image samples of exhaust gas from each ship, the second image features corresponding to the area where the three-dimensional model of the exhaust gas group is located, and the second image features corresponding to the three-dimensional model of the exhaust gas group are fused to obtain the first fused features, including:

[0043] The second image feature corresponding to the area where the three-dimensional model of the exhaust gas group is located is fused with the second image feature corresponding to the three-dimensional model of the exhaust gas group to obtain a second fused feature;

[0044] The second image features corresponding to the remote sensing image samples of exhaust gas from each ship are fused to obtain a third fused feature;

[0045] Calculate the first Euclidean distance between the second fusion feature and the third fusion feature;

[0046] When the first Euclidean distance is not greater than a first preset similarity threshold, obtaining a first fused feature according to the second fused feature and the third fused feature;

[0047] When the first Euclidean distance is greater than a first preset similarity threshold, calculating each second Euclidean distance between the second image feature corresponding to each ship exhaust remote sensing image sample and the second fusion feature;

[0048] The second image features corresponding to the remote sensing image samples of ship exhaust gas whose second Euclidean distances are less than the second preset similarity threshold are selected, and the first fusion feature is determined according to the second image features and the second fusion features corresponding to the remote sensing image samples of ship exhaust gas whose second Euclidean distances are less than the second preset similarity threshold; wherein the first preset similarity threshold is less than the second preset similarity threshold.

[0049] Further, determining the three-dimensional model of the exhaust gas mass according to the area where the three-dimensional model of the exhaust gas mass is located includes:

[0050] Projecting the area where the three-dimensional model of the exhaust gas mass is located in a spatial rectangular coordinate system to obtain a fourth projection image and a first projection distance of the area where the three-dimensional model of the exhaust gas mass is located on a horizontal plane of the spatial rectangular coordinate system, a fifth projection image and a second projection distance on a vertical plane, and a sixth projection image and a third projection distance on a lateral plane;

[0051] Performing edge detection on the fourth projection image, the fifth projection image, and the sixth projection image, and marking a number of edge coordinate points corresponding to each edge in the fourth projection image, the fifth projection image, and the sixth projection image;

[0052] Obtain a three-dimensional point cloud model corresponding to the area where the three-dimensional model of the exhaust gas mass is located according to each edge coordinate point, the first projection distance, the second projection distance, and the third projection distance;

[0053] The three-dimensional model of the exhaust gas mass is determined based on the three-dimensional point cloud model.

[0054] Based on the above method embodiment, the present invention provides a corresponding system embodiment;

[0055] An embodiment of the present invention provides a ship exhaust gas detection system, including: a data acquisition module, an exhaust gas emission detection module, a power control module and a detection model construction module;

[0056] The data acquisition module is used to acquire a set of ship exhaust remote sensing images at the current moment, a current ship exhaust emission standard value, and a current ship power output value; wherein the set of ship exhaust remote sensing images includes: a plurality of ship exhaust remote sensing images taken from multiple angles;

[0057] The exhaust gas emission detection module is used to input the current ship exhaust remote sensing image set into the exhaust gas detection model, so that the exhaust gas detection model outputs the current ship exhaust gas emission;

[0058] The power control module is used to determine the ship power control amount according to the current ship exhaust emission, the current ship exhaust emission standard value and the current ship power output value; and adjust the ship power output value at the next moment according to the ship power control amount;

[0059] The detection model building module is used to obtain a sample set of ship exhaust remote sensing images; wherein the sample set of ship exhaust remote sensing images includes a subset of ship exhaust remote sensing images at several historical moments; each subset of ship exhaust remote sensing images includes a plurality of ship exhaust remote sensing image samples taken at multiple angles; an initial exhaust detection model is constructed; the sample subsets of ship exhaust remote sensing images at each historical moment are used as input, and the initial exhaust detection model is iteratively trained with the sample set of ship exhaust remote sensing images until the initial exhaust detection model reaches a preset convergence condition, thereby generating an exhaust detection model.

[0060] Another embodiment of the present invention provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and when the processor executes the computer program, a method for detecting ship exhaust gas described in the above-mentioned embodiment of the invention is implemented.

[0061] Another embodiment of the present invention provides a storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute a method for detecting ship exhaust gas as described in the above-mentioned embodiment of the invention.

[0062] The following beneficial effects are achieved by implementing the present invention:

[0063] The present invention provides a method and system for detecting ship exhaust gas. The method comprises the following steps: obtaining a ship exhaust remote sensing image set at a current moment, an exhaust emission standard value of a current ship, and a current ship power output value; inputting the ship exhaust remote sensing image set at a current moment into an exhaust detection model, so that the exhaust detection model outputs the ship exhaust emission at a current moment; constructing an exhaust detection model, detecting the exhaust emission of the ship at a current moment based on the ship exhaust remote sensing image set collected at a current moment, and then determining whether the exhaust emission emitted by the ship power output value at a current moment meets the standard value, obtaining the ship power control amount based on the standard value, the current ship power output value, and the actual exhaust emission at a current moment, and then adjusting the ship power output value at a next moment, so that the ship exhaust emission at a next moment changes dynamically, so as to control the generation of exhaust gas from the source and reduce pollution to the environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 It is a flow chart of a method for detecting ship exhaust gas provided in one embodiment of the present invention.

[0065] Figure 2 It is a structural schematic diagram of a ship exhaust gas detection system provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0066] 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 ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0067] like Figure 1 As shown, a method for detecting ship exhaust gas provided by an embodiment of the present invention includes:

[0068] Step S1: obtaining a set of ship exhaust remote sensing images at the current moment, a current ship exhaust emission standard value, and a current ship power output value; wherein the set of ship exhaust remote sensing images includes: a plurality of ship exhaust remote sensing images taken from multiple angles;

[0069] Step S2: inputting the current ship exhaust remote sensing image set into the exhaust detection model, so that the exhaust detection model outputs the current ship exhaust emission;

[0070] Step S3: determining the ship power control amount according to the current ship exhaust emission, the current ship exhaust emission standard value and the current ship power output value;

[0071] Step S4: adjusting the power output value of the ship at the next moment according to the ship power control amount;

[0072] For step S1, a plurality of ship exhaust remote sensing images taken from multiple angles at the current time of the ship's navigation are obtained, and a ship exhaust remote sensing image set is formed by a plurality of ship exhaust remote sensing images taken from multiple angles. This shooting method can be performed by multiple drones around the ship, and the captured images are required to contain a complete exhaust gas group image. In addition, the exhaust emission standard value corresponding to the current ship and the power output value of the current ship's travel are also required to be obtained.

[0073] For step S2, the remote sensing image set of ship exhaust gas at the current moment is input into the trained exhaust gas detection model, so that the exhaust gas detection model outputs the exhaust gas emission of the ship at the current moment.

[0074] The construction of the exhaust gas detection model includes: obtaining a sample set of ship exhaust remote sensing images, which is constructed by a subset of ship exhaust remote sensing images at several historical moments, and each subset of ship exhaust remote sensing images at each historical moment is marked with the ship exhaust emissions at the corresponding historical moment; each subset of ship exhaust remote sensing images is composed of a number of ship exhaust remote sensing image samples taken from multiple angles at the corresponding historical moment. Construct an initial exhaust gas detection model, train the initial exhaust gas detection model through the ship exhaust remote sensing image sample set, and when the initial exhaust gas detection model reaches the maximum number of iterations, or the loss function of the model is minimized, the preset convergence condition of the model is reached, and the exhaust gas detection model is generated at this time.

[0075] For steps S3 and S4, the exhaust gas emission of the ship at the current moment output by the exhaust gas detection model is compared with the exhaust gas emission standard value of the current ship, and the difference between the exhaust gas emission of the ship at the current moment and the exhaust gas emission standard value of the current ship is calculated, and the power control amount of the ship is determined according to the difference and the power output value of the current ship, and then the power output value of the ship at the next moment is adjusted according to the control amount. If the above difference is a positive number, the exhaust gas emission of the current ship is greater than the standard value, and the minimum power control amount reduced to the standard value is determined, and then the power output value of the ship at the next moment is regulated to reduce the exhaust gas emission of the ship at the next moment to the standard value. If the above difference is a negative number or 0, the exhaust gas emission of the current ship is less than or equal to the standard value, and the maximum power control amount increased to the standard value is determined, and then the power output value of the ship at the next moment is regulated, so that the ship can appropriately increase the output power at the next moment, and the increased maximum output power does not exceed the maximum power control amount, so that the exhaust gas emission is within the standard range by dynamically controlling the propagation power output.

[0076] In a preferred embodiment, the initial exhaust gas detection model includes: a classification module, a conversion module, a first detection module, a second detection module, a feature fusion module and a correction module;

[0077] The iterative training of the initial exhaust gas detection model using the ship exhaust gas remote sensing image sample set includes:

[0078] For each round of iterative training, a subset of ship exhaust remote sensing image samples at a selected historical moment is input into a classification module, so that the classification module extracts the first image feature of each ship exhaust remote sensing image sample in the subset of ship exhaust remote sensing images at the selected historical moment; wherein the first image feature includes: image background environment and image background color; the image type of each ship exhaust remote sensing image sample is determined according to the first image feature; wherein the image type includes: daytime image and nighttime image; each ship exhaust remote sensing image sample whose image type is nighttime image is input into a conversion module, so that the conversion module converts each ship exhaust remote sensing image sample whose image type is nighttime image into a ship exhaust remote sensing image sample whose image type is daytime image, and inputs each converted ship exhaust remote sensing image sample into a first detection module and a second detection module respectively; each ship exhaust remote sensing image sample whose image type is daytime image is input into a first detection module and a second detection module respectively; the first detection module extracts the second image feature of each ship exhaust remote sensing image sample one by one, and inputs the second image feature corresponding to each ship exhaust remote sensing image sample into A feature fusion module; wherein the second image feature includes: exhaust gas group shape, exhaust gas group area color and exhaust gas group concentration; the second detection module determines the area where the exhaust gas group three-dimensional model is located and the exhaust gas group three-dimensional model according to each ship exhaust gas remote sensing image sample, extracts the second image feature corresponding to the area where the exhaust gas group three-dimensional model is located and the second image feature corresponding to the exhaust gas group three-dimensional model, and inputs the second image feature corresponding to the area where the exhaust gas group three-dimensional model is located and the second image feature corresponding to the exhaust gas group three-dimensional model into the feature fusion module; the feature fusion module fuses the second image feature corresponding to each ship exhaust gas remote sensing image sample, the second image feature corresponding to the area where the exhaust gas group three-dimensional model is located and the second image feature corresponding to the exhaust gas group three-dimensional model to obtain a first fusion feature, and determines the predicted ship exhaust gas emissions of the ship exhaust gas remote sensing image sample subset at the selected historical moment according to the first fusion feature, and inputs the predicted ship exhaust gas emissions into the correction module; the correction module determines the prediction error according to the ship exhaust gas remote sensing image sample subset at the selected historical moment and the predicted ship exhaust gas emissions, and adjusts the network parameters of the initial exhaust gas detection model according to the prediction error.

[0079] Specifically, in the process of training the initial exhaust gas detection model, for each round of iterative training, a subset of ship exhaust gas remote sensing image samples at a historical moment is selected from the ship exhaust gas remote sensing image sample set, and several ship exhaust gas remote sensing image samples contained in the subset are used as inputs of the initial exhaust gas detection model. The classification module input into the classification module of the exhaust gas detection model is usually a classifier, and the classification module extracts the image background environment and image background color in each ship exhaust gas remote sensing image sample, and determines the image type of the image sample by extracting the image background environment and image background color; if the image background environment contains a night background environment element such as the moon, and the image background color is non-white, it is classified as a night image, and the remaining image samples of non-night images are classified as daytime images; each ship exhaust gas remote sensing image sample whose image type is a night image is input into the conversion module, so that the conversion module converts each ship exhaust gas remote sensing image sample whose image type is a night image into a corresponding ship exhaust gas remote sensing image sample of a daytime image; for each ship exhaust gas remote sensing image sample originally classified as a daytime image, there is no need to convert. Through the conversion module, the exhaust remote sensing image samples of each ship in the ship exhaust remote sensing image sample subset are unified to improve the accuracy of subsequent feature extraction from the image samples and reduce interference factors during feature extraction.

[0080] The converted remote sensing image samples of each ship exhaust are input into the first detection module and the second detection module respectively, and the remote sensing image samples of each ship exhaust of the daytime image are input into the first detection module and the second detection module respectively. In the first detection module, the second image features in each ship exhaust remote sensing image sample are extracted respectively, that is, the shape of the exhaust gas group, the color of the exhaust gas group area and the exhaust gas group concentration in the image sample are extracted; the color of the exhaust gas group area mainly reflects the sulfide content in the exhaust gas, and the shape of the exhaust gas group and the exhaust gas group concentration can determine the exhaust gas emission. By extracting the second image features in each image sample, the model refines and learns the second image features of each ship exhaust remote sensing image sample one by one. In the second detection module, the area where the exhaust gas group three-dimensional model is located and the three-dimensional model of the exhaust gas group are reconstructed first according to the input exhaust gas remote sensing image samples of each ship, so that the exhaust gas remote sensing image samples of each ship are converted from two-dimensional to three-dimensional. After reconstructing it into an overall three-dimensional model, the features of the exhaust gas remote sensing image samples of each ship are reconstructed, and then the overall features are extracted; that is, the second image features corresponding to the area where the three-dimensional model of the exhaust gas group is located and the second image features corresponding to the three-dimensional model of the exhaust gas group are extracted.

[0081] In a preferred embodiment, the method of determining the area where the three-dimensional model of the exhaust gas mass is located and the three-dimensional model of the exhaust gas mass according to each ship exhaust gas remote sensing image sample includes: selecting a ship exhaust gas remote sensing image sample from each ship exhaust gas remote sensing image sample as a selected ship exhaust gas remote sensing image sample; selecting a two-dimensional area where the exhaust gas mass is located in the selected ship exhaust gas remote sensing image sample, establishing a spatial rectangular coordinate system with the centroid of the two-dimensional area as the origin, and selecting a three-dimensional area on the spatial rectangular coordinate system that can fully cover the two-dimensional area where the exhaust gas mass is located, and using the three-dimensional area as the initial area where the three-dimensional model of the exhaust gas mass is located; determining the selected ship exhaust gas remote sensing image sample according to the shooting angle of the selected ship exhaust gas remote sensing image sample. a first relative position relationship between a shooting angle and the origin of a spatial rectangular coordinate system; using the remaining ship exhaust remote sensing image samples except the selected ship exhaust remote sensing image sample as the completed ship exhaust remote sensing image samples; for each completed ship exhaust remote sensing image sample, determining the second relative position relationship between the shooting angle of the completed ship exhaust remote sensing image sample and the origin of the spatial rectangular coordinate system according to the shooting angle of the completed ship exhaust remote sensing image sample; correcting the area where the initial exhaust mass three-dimensional model is located according to the first relative position relationship, each second relative position relationship and each completed ship exhaust remote sensing image sample to determine the area where the exhaust mass three-dimensional model is located; determining the exhaust mass three-dimensional model according to the area where the exhaust mass three-dimensional model is located.

[0082] Specifically, an image sample is selected from the ship exhaust remote sensing image samples included in the ship exhaust remote sensing sample image subset as the selected ship exhaust remote sensing image sample, and the selected image sample is used as the basis for reconstructing the area where the exhaust gas group three-dimensional model is located and the exhaust gas group three-dimensional model. It should be noted that the area where the exhaust gas group three-dimensional model is located in the present invention is usually a cubic area, which completely wraps the exhaust gas group three-dimensional model. Taking the selected image sample as a reference, the two-dimensional area where the exhaust gas group is located in the selected image sample is framed, and the exhaust gas group is usually framed in the form of a rectangle. Whether the rectangle completely wraps the exhaust gas group is determined by detecting whether there are still exhaust gas group features outside the framed area. After the frame selection is completed, the center of mass of the rectangle is selected as the origin of the spatial rectangular coordinate system to establish a spatial rectangular coordinate system. The two-dimensional area that wraps the exhaust gas group is wrapped by the cube established by the spatial rectangular coordinate system, so that the three-dimensional area that can fully cover the two-dimensional area, that is, the above-mentioned cube is used as the area where the initial exhaust gas group three-dimensional model is located. Determine the first relative position relationship between the shooting angle of the selected image sample and the origin of the spatial rectangular coordinate system, and use this relative position relationship as the reference angle. When the subsequent image samples are used to complete the area where the initial exhaust gas group three-dimensional model is located and the initial exhaust gas group three-dimensional model, the angle adjustment needs to be based on this reference angle, so as to improve the accuracy of the completion. Furthermore, the remaining image samples except the selected image sample are used as the completed image samples, and the second relative position relationship between each completed image sample and the origin of the spatial rectangular coordinate system is determined according to the shooting angle of the completed image sample. The second relative position relationship is adjusted based on the first relative position relationship, and then the area where the exhaust gas group three-dimensional model is located and the exhaust gas group three-dimensional model wrapped by it are completed and corrected according to the adjusted completed image sample, so as to obtain the area where the exhaust gas group three-dimensional model is located and the exhaust gas group three-dimensional model wrapped by it.

[0083] In a preferred embodiment, the area where the initial exhaust gas group three-dimensional model is located is corrected according to the first relative position relationship, each second relative position relationship and each completed ship exhaust gas remote sensing image sample, including: for each completed ship exhaust gas remote sensing image sample, according to the second relative position relationship, projecting the initial exhaust gas group three-dimensional model in the spatial rectangular coordinate system where the initial exhaust gas group three-dimensional model is located, to obtain the first projection image of the completed ship exhaust gas remote sensing image sample on the horizontal plane of the spatial rectangular coordinate system, the second projection image on the vertical plane and the third projection image on the side plane; according to the first relative position relationship and the first projection image, the second projection image and the third projection image, the area where the initial exhaust gas group three-dimensional model is located is corrected.

[0084] In another preferred embodiment, the correction of the area where the initial exhaust gas group three-dimensional model is located according to the first relative position relationship and the first projection image, the second projection image and the third projection image includes: determining a first corrected three-dimensional model according to the first relative position relationship, the first projection image and the second projection image; determining a second corrected three-dimensional model according to the first relative position relationship, the first projection image and the third projection image; determining a third corrected three-dimensional model according to the first relative position relationship, the second projection image and the third projection image; based on the first relative position relationship, superimposing the first corrected three-dimensional model and the second corrected three-dimensional model to obtain a first superimposed corrected three-dimensional model, and superimposing the first corrected three-dimensional model The positive three-dimensional model and the third corrected three-dimensional model are superimposed to obtain a second superimposed corrected three-dimensional model, and the second corrected three-dimensional model and the third corrected three-dimensional model are superimposed to obtain a third superimposed corrected three-dimensional model; based on the first relative position relationship, the first corrected three-dimensional model, the second corrected three-dimensional model and the third corrected three-dimensional model are fitted to obtain a fitted corrected three-dimensional model; the fitted corrected three-dimensional model is adjusted with the first superimposed corrected three-dimensional model, the second superimposed corrected three-dimensional model and the third superimposed corrected three-dimensional model to obtain a three-dimensional model that completes the ship exhaust remote sensing image sample; according to the three-dimensional model of the ship exhaust remote sensing image sample, the area where the initial exhaust gas group three-dimensional model is located is corrected.

[0085] Specifically, when each completed image sample completes and corrects the area where the initial exhaust gas group three-dimensional model is located and the exhaust gas group three-dimensional model wrapped by it, the completed image sample is first projected on each plane of the spatial rectangular coordinate system according to the second relative position relationship, specifically projected on the horizontal plane, the vertical plane and the side plane respectively, and the first projection image, the second projection image and the third projection image are obtained in sequence. According to the first relative position relationship, the position relationship of the first projection image and the position of the second projection image are adjusted, and the adjusted first projection image and the second projection image are combined to determine the first corrected three-dimensional model; similarly, the second corrected three-dimensional model and the third corrected three-dimensional model can be obtained. Based on the limitation of the first relative position relationship, while ensuring that the position relationship does not change, the first corrected three-dimensional model and the second corrected three-dimensional model are superimposed to obtain a first superimposed corrected three-dimensional model, the first corrected three-dimensional model and the third corrected three-dimensional model are superimposed to obtain a second superimposed corrected three-dimensional model, and the second corrected three-dimensional model and the third corrected three-dimensional model are superimposed to obtain a third superimposed corrected three-dimensional model. After completing the above processing, the first corrected three-dimensional model, the second corrected three-dimensional model and the third corrected three-dimensional model are fitted to obtain a fitted corrected three-dimensional model, and then the fitted corrected three-dimensional model is further adjusted through each superimposed corrected three-dimensional model to obtain a three-dimensional model corresponding to each image sample, and then each three-dimensional model is used to correct and complete the area where the initial exhaust gas group three-dimensional model is located and the exhaust gas group three-dimensional model wrapped by it; it should be noted that during the correction and completion process, if there is an overlapping area, the data of the overlapping area is fitted. The above repeated superposition and correction process is to make the three-dimensional model corresponding to each image sample closer to the real three-dimensional exhaust gas group of the corresponding part in its image sample, so that the error is reduced in the process of transformation from two-dimensional to three-dimensional, and the accuracy of subsequent correction is improved.

[0086] In a preferred embodiment, determining the three-dimensional model of the exhaust gas mass according to the area where the three-dimensional model of the exhaust gas mass is located includes: projecting the area where the three-dimensional model of the exhaust gas mass is located in a spatial rectangular coordinate system to obtain a fourth projection image and a first projection distance of the area where the three-dimensional model of the exhaust gas mass is located on the horizontal plane of the spatial rectangular coordinate system, a fifth projection image and a second projection distance on the vertical plane, and a sixth projection image and a third projection distance on the side plane; performing edge detection on the fourth projection image, the fifth projection image, and the sixth projection image, and marking a number of edge coordinate points corresponding to each edge in the fourth projection image, the fifth projection image, and the sixth projection image; obtaining a three-dimensional point cloud model corresponding to the area where the three-dimensional model of the exhaust gas mass is located according to each edge coordinate point, the first projection distance, the second projection distance, and the third projection distance; and determining the three-dimensional model of the exhaust gas mass based on the three-dimensional point cloud model.

[0087] Specifically, the area where the three-dimensional model of the exhaust gas mass is located wraps the three-dimensional model of the exhaust gas mass. When determining the three-dimensional model of the exhaust gas mass, the three-dimensional model of the exhaust gas mass is actually separated from the area where the three-dimensional model of the exhaust gas mass is located. At this time, the three-dimensional model can be converted into a two-dimensional projection on each plane by projecting the area where the three-dimensional model of the exhaust gas mass is located in the spatial rectangular coordinate system, and the projection distance on each plane is determined. The fourth projection image, the fifth projection image, and the sixth projection image after projection are edge detected, so as to realize the edge detection of the edge of the area and the edge of the exhaust gas mass shape in each projection image, and determine the coordinate points of each edge under edge detection. Then, the coordinates determined by the two-dimensional projection are converted into three-dimensional coordinates through the projection distance, and the three-dimensional coordinate points in each edge can be determined, and then the three-dimensional point cloud model corresponding to the area where the three-dimensional model of the exhaust gas mass is located can be obtained. The three-dimensional point cloud model corresponding to the exhaust gas mass is obtained by screening the three-dimensional coordinate points that characterize the edge of the exhaust gas mass, and then the three-dimensional model of the exhaust gas mass is determined. Preferably, when screening coordinate points, the continuity of the coordinate points can be used for screening, and the shape of the area is a cube, that is, the coordinate points are continuously arranged on any coordinate axis, and the edges of the exhaust gas group are irregular edges, that is, the coordinate points are irregularly arranged.

[0088] The first detection module inputs the second image features corresponding to the extracted remote sensing image samples of exhaust gas from each ship into the feature fusion module, and the second detection module inputs the second image features corresponding to the area where the three-dimensional model of the exhaust gas group is located and the second image features corresponding to the three-dimensional model of the exhaust gas group into the feature fusion module.

[0089] The feature fusion module fuses all the input second image features to obtain the first fusion feature, which is the fused exhaust gas group shape, exhaust gas group area color, exhaust gas group concentration and other features; based on the first fusion feature, the predicted ship exhaust emissions of the ship exhaust remote sensing image sample subset at the selected historical moment are determined.

[0090] In a preferred embodiment, the first image features corresponding to the exhaust remote sensing image samples of each ship, the first image features corresponding to the area where the exhaust gas group three-dimensional model is located, and the first image features corresponding to the exhaust gas group three-dimensional model are fused to obtain the first fused feature, including: fusing the first image features corresponding to the area where the exhaust gas group three-dimensional model is located and the first image features corresponding to the exhaust gas group three-dimensional model to obtain a second fused feature; fusing the first image features corresponding to the exhaust remote sensing image samples of each ship to obtain a third fused feature; calculating the first Euclidean distance between the second fused feature and the third fused feature; when the first Euclidean distance is not greater than a first preset similarity threshold, the first fusion feature is obtained according to the second fusion feature and the third fusion feature; when the first Euclidean distance is greater than the first preset similarity threshold, the second Euclidean distances between the first image feature corresponding to each ship exhaust remote sensing image sample and the second fusion feature are calculated; the first image features corresponding to each ship exhaust remote sensing image sample whose second Euclidean distances are less than the second preset similarity threshold are selected, and the first fusion feature is determined according to the first image features and the second fusion features corresponding to each ship exhaust remote sensing image sample whose second Euclidean distances are less than the second preset similarity threshold; wherein the first preset similarity threshold is less than the second preset similarity threshold.

[0091] Specifically, the input features of the first detection module are fused to obtain the third fused feature; the input features of the second detection module are fused to obtain the second fused feature. The similarity of the fused features of the first detection module and the second detection module is calculated by the Euclidean distance calculation formula, that is, the first Euclidean distance between the second fused feature and the third fused feature is calculated. If the first Euclidean distance is not greater than the first preset similarity threshold, that is, the similarity is high, then the first fused feature is obtained by taking the average value of the second fused feature and the third fused feature. If the first Euclidean distance is greater than the first preset similarity threshold, that is, the difference in features after fusion is large, the third fused feature is not used at this time, and the similarity between the second image feature corresponding to each ship exhaust remote sensing image sample and the second fused feature is calculated to screen out some features with low similarity. Specifically, the second Euclidean distances between the second image features corresponding to each ship exhaust remote sensing image sample and the second fusion feature are calculated, and the second image features corresponding to each ship exhaust remote sensing image sample whose second Euclidean distances are less than the second preset similarity threshold are selected, that is, several second image features with higher similarity are selected, and then the selected several second image features with higher similarity and the second fusion feature are further fused, so that the second fusion feature has richer feature information and is not affected by features with lower similarity, thereby improving the accuracy of the exhaust emission prediction determined according to the first fusion feature.

[0092] The predicted ship exhaust emissions are input into the correction module, which obtains the actual ship exhaust emissions corresponding to the subset of ship exhaust remote sensing image samples at a certain historical moment, and then determines the prediction error. Based on the prediction error, the network parameters of the initial exhaust gas detection model are adjusted to increase the accuracy of its prediction effect in the next round of iterative training.

[0093] Based on the above method embodiments, the present invention provides corresponding system embodiments.

[0094] like Figure 2 As shown, an embodiment of the present invention provides a ship exhaust gas detection system, including: a data acquisition module, an exhaust gas emission detection module, a power control module and a detection model construction module;

[0095] The data acquisition module is used to acquire a set of ship exhaust remote sensing images at the current moment, a current ship exhaust emission standard value, and a current ship power output value; wherein the set of ship exhaust remote sensing images includes: a plurality of ship exhaust remote sensing images taken from multiple angles;

[0096] The exhaust gas emission detection module is used to input the current ship exhaust remote sensing image set into the exhaust gas detection model, so that the exhaust gas detection model outputs the current ship exhaust gas emission;

[0097] The power control module is used to determine the ship power control amount according to the current ship exhaust emission, the current ship exhaust emission standard value and the current ship power output value; and adjust the ship power output value at the next moment according to the ship power control amount;

[0098] The detection model building module is used to obtain a sample set of ship exhaust remote sensing images; wherein the sample set of ship exhaust remote sensing images includes a subset of ship exhaust remote sensing images at several historical moments; each subset of ship exhaust remote sensing images includes a plurality of ship exhaust remote sensing image samples taken at multiple angles; an initial exhaust detection model is constructed; the sample subsets of ship exhaust remote sensing images at each historical moment are used as input, and the initial exhaust detection model is iteratively trained with the sample set of ship exhaust remote sensing images until the initial exhaust detection model reaches a preset convergence condition, thereby generating an exhaust detection model.

[0099] It should be noted that the system embodiment described above is merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the system embodiment provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0100] Those skilled in the art can clearly understand that for the sake of convenience and brevity, the specific working process of the system described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0101] Based on the above method item embodiments, the present invention provides corresponding terminal device item embodiments.

[0102] An embodiment of the present invention provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, a method for detecting ship exhaust gas as described in any one of the present invention is implemented.

[0103] The terminal device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0104] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the terminal device, and uses various interfaces and lines to connect various parts of the entire terminal device.

[0105] The memory can be used to store the computer program, and the processor realizes various functions of the terminal device by running or executing the computer program stored in the memory and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required for a function, etc.; the data storage area can store data created according to the use of the mobile phone, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, 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), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0106] Based on the above method item embodiments, the present invention provides a corresponding storage medium item embodiment.

[0107] An embodiment of the present invention provides a storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute a ship exhaust gas detection method as described in any one of the present inventions.

[0108] The storage medium is a computer-readable storage medium, and the computer program is stored in the computer-readable storage medium. When the computer program is executed by the processor, the steps of each of the above-mentioned method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0109] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for detecting ship exhaust gas, characterized in that: include: Acquire a ship exhaust remote sensing image set at the current moment, the exhaust emission standard value of the current ship, and the current ship power output value; wherein the ship exhaust remote sensing image set includes: a plurality of ship exhaust remote sensing images taken from multiple angles; Inputting the current ship exhaust remote sensing image set into the exhaust detection model so that the exhaust detection model outputs the current ship exhaust emission; Determine the ship power control amount according to the ship exhaust gas emission at the current moment, the current ship exhaust gas emission standard value and the current ship power output value; Adjusting the power output value of the ship at the next moment according to the ship power control amount; Wherein, the construction of the exhaust gas detection model includes: Acquire a sample set of ship exhaust remote sensing images; wherein the sample set of ship exhaust remote sensing images includes a subset of ship exhaust remote sensing images at a number of historical moments; each subset of ship exhaust remote sensing images includes a number of ship exhaust remote sensing image samples taken at multiple angles; Constructing an initial exhaust gas detection model; Iteratively training the initial exhaust gas detection model with the ship exhaust gas remote sensing image sample set until the initial exhaust gas detection model reaches a preset convergence condition, thereby generating an exhaust gas detection model; The initial exhaust gas detection model includes: a classification module, a conversion module, a first detection module, a second detection module, a feature fusion module and a correction module; The iterative training of the initial exhaust gas detection model using the ship exhaust gas remote sensing image sample set includes: For each round of iterative training, a subset of ship exhaust remote sensing image samples at a selected historical moment is input into a classification module, so that the classification module extracts a first image feature of each ship exhaust remote sensing image sample in the subset of ship exhaust remote sensing images at the selected historical moment; wherein the first image feature includes: image background environment and image background color; Determine the image type of each ship exhaust remote sensing image sample according to the first image feature; wherein the image type includes: daytime image and nighttime image; Input each ship exhaust remote sensing image sample whose image type is night image into the conversion module, so that the conversion module converts each ship exhaust remote sensing image sample whose image type is night image into a ship exhaust remote sensing image sample whose image type is daytime image, and inputs each converted ship exhaust remote sensing image sample into the first detection module and the second detection module respectively; Inputting each ship exhaust remote sensing image sample whose image type is a daytime image into the first detection module and the second detection module respectively; The first detection module extracts the second image features of each ship exhaust remote sensing image sample one by one, and inputs the second image features corresponding to each ship exhaust remote sensing image sample into the feature fusion module; wherein the second image features include: exhaust gas group shape, exhaust gas group area color and exhaust gas group concentration; The second detection module determines the area where the three-dimensional model of the exhaust gas mass is located and the three-dimensional model of the exhaust gas mass according to the exhaust gas remote sensing image samples of each ship, extracts the second image features corresponding to the area where the three-dimensional model of the exhaust gas mass is located and the second image features corresponding to the three-dimensional model of the exhaust gas mass, and inputs the second image features corresponding to the area where the three-dimensional model of the exhaust gas mass is located and the second image features corresponding to the three-dimensional model of the exhaust gas mass into the feature fusion module; The feature fusion module fuses the second image features corresponding to each ship exhaust remote sensing image sample, the second image features corresponding to the area where the exhaust gas group three-dimensional model is located, and the second image features corresponding to the exhaust gas group three-dimensional model to obtain a first fusion feature, and determines the predicted ship exhaust emissions of the ship exhaust remote sensing image sample subset at the selected historical moment according to the first fusion feature, and inputs the predicted ship exhaust emissions into the correction module; The correction module determines the prediction error according to the sample subset of the ship exhaust remote sensing image at the selected historical moment and the predicted ship exhaust emission, and adjusts the network parameters of the initial exhaust detection model according to the prediction error; The method of determining the area where the exhaust gas cluster three-dimensional model is located and the exhaust gas cluster three-dimensional model according to the exhaust gas remote sensing image samples of each ship includes: Selecting a ship exhaust gas remote sensing image sample from each ship exhaust gas remote sensing image sample as a selected ship exhaust gas remote sensing image sample; Select the two-dimensional region where the exhaust gas mass is located in the selected ship exhaust gas remote sensing image sample, establish a spatial rectangular coordinate system with the centroid of the two-dimensional region as the origin, and select a three-dimensional region that can fully cover the two-dimensional region where the exhaust gas mass is located on the spatial rectangular coordinate system, and use the three-dimensional region as the region where the initial exhaust gas mass three-dimensional model is located; Determining a first relative position relationship between the shooting angle of the selected ship exhaust remote sensing image sample and the origin of the spatial rectangular coordinate system according to the shooting angle of the selected ship exhaust remote sensing image sample; The remaining ship exhaust remote sensing image samples except the selected ship exhaust remote sensing image samples are used as the complementary ship exhaust remote sensing image samples; For each completed ship exhaust remote sensing image sample, determining a second relative position relationship between the shooting angle of the completed ship exhaust remote sensing image sample and the origin of the spatial rectangular coordinate system according to the shooting angle of the completed ship exhaust remote sensing image sample; Correcting the area where the initial exhaust gas mass three-dimensional model is located according to the first relative position relationship, each second relative position relationship and each completed ship exhaust gas remote sensing image sample to determine the area where the exhaust gas mass three-dimensional model is located; The three-dimensional model of the exhaust gas mass is determined according to the area where the three-dimensional model of the exhaust gas mass is located.

2. A method for detecting ship exhaust gas according to claim 1, characterized in that: The step of correcting the area where the initial exhaust gas group three-dimensional model is located according to the first relative position relationship, each second relative position relationship and each completed ship exhaust gas remote sensing image sample comprises: For each completed ship exhaust remote sensing image sample, projection is performed in the spatial rectangular coordinate system where the initial exhaust gas group three-dimensional model is located according to the second relative position relationship to obtain a first projection image of the completed ship exhaust remote sensing image sample on the horizontal plane of the spatial rectangular coordinate system, a second projection image on the vertical plane, and a third projection image on the lateral plane; The area where the initial exhaust gas mass three-dimensional model is located is corrected according to the first relative position relationship and the first projection image, the second projection image and the third projection image.

3. A method for detecting ship exhaust gas according to claim 2, characterized in that: The method of correcting the area where the initial exhaust gas group three-dimensional model is located according to the first relative position relationship and the first projection image, the second projection image and the third projection image includes: Determine a first corrected three-dimensional model according to the first relative position relationship, the first projection image, and the second projection image; Determine a second corrected three-dimensional model according to the first relative position relationship, the first projection image, and the third projection image; Determine a third corrected three-dimensional model according to the first relative position relationship, the second projection image and the third projection image; Based on the first relative position relationship, superimposing the first corrected three-dimensional model and the second corrected three-dimensional model to obtain a first superimposed corrected three-dimensional model, superimposing the first corrected three-dimensional model and the third corrected three-dimensional model to obtain a second superimposed corrected three-dimensional model, and superimposing the second corrected three-dimensional model and the third corrected three-dimensional model to obtain a third superimposed corrected three-dimensional model; Based on the first relative position relationship, fitting the first corrected three-dimensional model, the second corrected three-dimensional model and the third corrected three-dimensional model to obtain a fitted corrected three-dimensional model; The fitted corrected three-dimensional model is adjusted by using the first superimposed corrected three-dimensional model, the second superimposed corrected three-dimensional model and the third superimposed corrected three-dimensional model to obtain a three-dimensional model that completes the ship exhaust remote sensing image sample; The area where the initial exhaust gas cluster three-dimensional model is located is corrected according to the three-dimensional model of the ship exhaust gas remote sensing image sample.

4. A method for detecting ship exhaust gas according to claim 1, characterized in that: The second image features corresponding to the remote sensing image samples of exhaust gas from each ship, the second image features corresponding to the area where the three-dimensional model of the exhaust gas group is located, and the second image features corresponding to the three-dimensional model of the exhaust gas group are fused to obtain the first fused features, including: The second image feature corresponding to the area where the three-dimensional model of the exhaust gas group is located is fused with the second image feature corresponding to the three-dimensional model of the exhaust gas group to obtain a second fused feature; The second image features corresponding to the remote sensing image samples of exhaust gas from each ship are fused to obtain a third fused feature; Calculate the first Euclidean distance between the second fusion feature and the third fusion feature; When the first Euclidean distance is not greater than a first preset similarity threshold, obtaining a first fused feature according to the second fused feature and the third fused feature; When the first Euclidean distance is greater than a first preset similarity threshold, calculating each second Euclidean distance between the second image feature corresponding to each ship exhaust remote sensing image sample and the second fusion feature; The second image features corresponding to the remote sensing image samples of ship exhaust gas whose second Euclidean distances are less than the second preset similarity threshold are selected, and the first fusion feature is determined according to the second image features and the second fusion features corresponding to the remote sensing image samples of ship exhaust gas whose second Euclidean distances are less than the second preset similarity threshold; wherein the first preset similarity threshold is less than the second preset similarity threshold.

5. A method for detecting ship exhaust gas according to claim 1, characterized in that: The step of determining the three-dimensional model of the exhaust gas mass according to the area where the three-dimensional model of the exhaust gas mass is located comprises: Projecting the area where the three-dimensional model of the exhaust gas mass is located in a spatial rectangular coordinate system to obtain a fourth projection image and a first projection distance of the area where the three-dimensional model of the exhaust gas mass is located on a horizontal plane of the spatial rectangular coordinate system, a fifth projection image and a second projection distance on a vertical plane, and a sixth projection image and a third projection distance on a lateral plane; Performing edge detection on the fourth projection image, the fifth projection image, and the sixth projection image, and marking a number of edge coordinate points corresponding to each edge in the fourth projection image, the fifth projection image, and the sixth projection image; Obtain a three-dimensional point cloud model corresponding to the area where the three-dimensional model of the exhaust gas mass is located according to each edge coordinate point, the first projection distance, the second projection distance, and the third projection distance; The three-dimensional model of the exhaust gas mass is determined based on the three-dimensional point cloud model.

6. A ship exhaust gas detection system, characterized in that: include: Data acquisition module, exhaust emission detection module, power control module and detection model building module; The data acquisition module is used to acquire a set of ship exhaust remote sensing images at the current moment, a current ship exhaust emission standard value, and a current ship power output value; wherein the set of ship exhaust remote sensing images includes: a plurality of ship exhaust remote sensing images taken from multiple angles; The exhaust gas emission detection module is used to input the current ship exhaust remote sensing image set into the exhaust gas detection model, so that the exhaust gas detection model outputs the current ship exhaust gas emission; The power control module is used to determine the ship power control amount according to the current ship exhaust emission, the current ship exhaust emission standard value and the current ship power output value; and adjust the ship power output value at the next moment according to the ship power control amount; The detection model building module is used to obtain a sample set of ship exhaust remote sensing images; wherein the sample set of ship exhaust remote sensing images includes a subset of ship exhaust remote sensing images at several historical moments; each subset of ship exhaust remote sensing images includes a plurality of ship exhaust remote sensing image samples taken at multiple angles; an initial exhaust detection model is constructed; the sample subsets of ship exhaust remote sensing images at each historical moment are used as input, and the initial exhaust detection model is iteratively trained with the sample set of ship exhaust remote sensing images until the initial exhaust detection model reaches a preset convergence condition, thereby generating an exhaust detection model; The initial exhaust gas detection model includes: a classification module, a conversion module, a first detection module, a second detection module, a feature fusion module and a correction module; The iterative training of the initial exhaust gas detection model using the ship exhaust gas remote sensing image sample set includes: For each round of iterative training, a subset of ship exhaust remote sensing image samples at a selected historical moment is input into a classification module, so that the classification module extracts a first image feature of each ship exhaust remote sensing image sample in the subset of ship exhaust remote sensing images at the selected historical moment; wherein the first image feature includes: image background environment and image background color; Determine the image type of each ship exhaust remote sensing image sample according to the first image feature; wherein the image type includes: daytime image and nighttime image; Input each ship exhaust remote sensing image sample whose image type is night image into the conversion module, so that the conversion module converts each ship exhaust remote sensing image sample whose image type is night image into a ship exhaust remote sensing image sample whose image type is daytime image, and inputs each converted ship exhaust remote sensing image sample into the first detection module and the second detection module respectively; Inputting each ship exhaust remote sensing image sample whose image type is a daytime image into the first detection module and the second detection module respectively; The first detection module extracts the second image features of each ship exhaust remote sensing image sample one by one, and inputs the second image features corresponding to each ship exhaust remote sensing image sample into the feature fusion module; wherein the second image features include: exhaust gas group shape, exhaust gas group area color and exhaust gas group concentration; The second detection module determines the area where the three-dimensional model of the exhaust gas mass is located and the three-dimensional model of the exhaust gas mass according to the exhaust gas remote sensing image samples of each ship, extracts the second image features corresponding to the area where the three-dimensional model of the exhaust gas mass is located and the second image features corresponding to the three-dimensional model of the exhaust gas mass, and inputs the second image features corresponding to the area where the three-dimensional model of the exhaust gas mass is located and the second image features corresponding to the three-dimensional model of the exhaust gas mass into the feature fusion module; The feature fusion module fuses the second image features corresponding to each ship exhaust remote sensing image sample, the second image features corresponding to the area where the exhaust gas group three-dimensional model is located, and the second image features corresponding to the exhaust gas group three-dimensional model to obtain a first fusion feature, and determines the predicted ship exhaust emissions of the ship exhaust remote sensing image sample subset at the selected historical moment according to the first fusion feature, and inputs the predicted ship exhaust emissions into the correction module; The correction module determines the prediction error according to the sample subset of the ship exhaust remote sensing image at the selected historical moment and the predicted ship exhaust emission, and adjusts the network parameters of the initial exhaust detection model according to the prediction error; The method of determining the area where the exhaust gas cluster three-dimensional model is located and the exhaust gas cluster three-dimensional model according to the exhaust gas remote sensing image samples of each ship includes: Selecting a ship exhaust gas remote sensing image sample from each ship exhaust gas remote sensing image sample as a selected ship exhaust gas remote sensing image sample; Select the two-dimensional region where the exhaust gas mass is located in the selected ship exhaust gas remote sensing image sample, establish a spatial rectangular coordinate system with the centroid of the two-dimensional region as the origin, and select a three-dimensional region that can fully cover the two-dimensional region where the exhaust gas mass is located on the spatial rectangular coordinate system, and use the three-dimensional region as the region where the initial exhaust gas mass three-dimensional model is located; Determining a first relative position relationship between the shooting angle of the selected ship exhaust remote sensing image sample and the origin of the spatial rectangular coordinate system according to the shooting angle of the selected ship exhaust remote sensing image sample; The remaining ship exhaust remote sensing image samples except the selected ship exhaust remote sensing image samples are used as the complementary ship exhaust remote sensing image samples; For each completed ship exhaust remote sensing image sample, determining a second relative position relationship between the shooting angle of the completed ship exhaust remote sensing image sample and the origin of the spatial rectangular coordinate system according to the shooting angle of the completed ship exhaust remote sensing image sample; Correcting the area where the initial exhaust gas mass three-dimensional model is located according to the first relative position relationship, each second relative position relationship and each completed ship exhaust gas remote sensing image sample to determine the area where the exhaust gas mass three-dimensional model is located; The three-dimensional model of the exhaust gas mass is determined according to the area where the three-dimensional model of the exhaust gas mass is located.

7. A terminal device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, a method for detecting ship exhaust gas as described in any one of claims 1 to 5 is implemented.

8. A storage medium, characterized in that: The storage medium includes a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute a ship exhaust gas detection method as described in any one of claims 1 to 5.