KR20200063302A - System and method for inspection of ship painting condition using image analysis - Google Patents

Abstract

The present invention relates to a system and a method for inspecting a painting condition of a vessel by using image analysis. According to an embodiment of the present invention, the system for inspecting the painting condition of a vessel by using image analysis includes: a camera obtaining an image by photographing the exterior of a vessel; a coating thickness measurer measuring the thickness of a coating on the surface of the vessel; an inspection part determining the painting condition of the vessel by analyzing the image obtained by the camera; and a display part displaying an analysis result of the analysis part. The inspection part determines the painting condition of the vessel through deep learning technology using a deep neural network (DNN). According to the present invention, the system is capable of accurately and efficiently inspecting the painting condition of the vessel through image analysis using deep learning, and providing a guide about painting work to be performed now in accordance with an inspection result about the painting condition of the vessel, thereby enabling efficient painting work.

Description

System and method for inspecting ship's painting condition using image analysis{SYSTEM AND METHOD FOR INSPECTION OF SHIP PAINTING CONDITION USING IMAGE ANALYSIS}

The present invention relates to a ship painting condition inspection system and method using image analysis, and more specifically, to analyze the image obtained by photographing the exterior of a ship using deep-learning technology to determine the ship's painting status. The present invention relates to a ship painting condition inspection system and method using image analysis that provides a guide to the user according to the judgment result.

In general, in the drying stage of the ship, a painting process is performed on the surface of the ship to prevent corrosion of the hull due to contact with air or water. The painting process on the surface of the ship is a very important process as the life of the ship varies depending on its performance, and accordingly, a technique capable of inspecting the condition or quality of the painting surface of the ship is required.

Conventionally, in inspecting such a painting state or quality, a method that depends on an inspector's sense, such as the hand's sense or the naked eye, has been used, or a device capable of measuring the depth of the paint has been used. However, since the inspection method according to the prior art is performed manually by the inspector, there is a high possibility of error, and the standards are different for each inspector, which makes it difficult to control the quality or quality of the coating. There is a problem that it is very inefficient because the surface must be inspected individually.

According to the problems of the prior art, there is a need for a technique capable of efficiently and accurately inspecting a ship's painting condition to manage the ship.

Republic of Korea Patent Publication No. 10-2014-0140828 (2014.12.10)

The present invention has been devised to solve the above-mentioned problems, and an object thereof is to provide a system and method for accurately and efficiently inspecting a ship's painting state through image analysis using deep learning.

In addition, the present invention is to provide an efficient coating operation by providing a guide for the current coating operation to be performed according to the inspection result of the coating condition of the ship.

The technical problems of the present invention are not limited to the technical problems mentioned above, and other technical problems not mentioned will be clearly understood by a person skilled in the art from the following description.

Vessel painting condition inspection system using image analysis according to an embodiment of the present invention for achieving the above-described technical problem, the camera to acquire the image by photographing the exterior of the ship, measuring the thickness of the coating film painted on the surface of the ship It includes a film thickness meter, an inspection unit that analyzes the image acquired by the camera to determine the painting condition of the ship, and a display unit that displays the analysis result of the analysis unit, wherein the inspection unit uses a deep neural network (DNN) Through the deep learning technology, it is possible to determine the painting condition of the ship.

According to an embodiment of the present disclosure, the inspection unit may determine a painting state including one or more of whether the painting is completed, whether the painting surface is defective, or the painting quality.

According to one embodiment, the inspection unit, a storage unit for collecting and storing the thickness of the coating film or the image of the ship's painting surface according to the painting state learning unit for learning the image pattern according to the painting state of the ship using the image stored in the storage unit And it may include an analysis unit for determining the coating state of the vessel obtained from the camera or film thickness meter according to the learning result.

According to an embodiment, the learning unit may learn an image of a vessel painting surface according to each stage of top, middle, and bottom.

According to an embodiment of the present disclosure, the learning unit may learn a defect image of a ship's painting surface.

The ship painting condition inspection system using the image analysis according to an embodiment may further include a guide unit providing a guide for painting to the user according to the analysis result of the inspection unit.

According to an embodiment of the present disclosure, when it is determined that painting is incomplete as a result of analysis of the analysis unit, the guide unit may provide a guide for completing the painting operation.

According to one embodiment, the guide unit may provide a guide for removing a defect when it is determined that the analysis portion of the analysis unit is defective on the surface of the ship.

On the other hand, the ship coating state inspection method using the image analysis according to an embodiment of the present invention for achieving the above-described technical problem, the camera photographing the exterior of the ship, the film thickness gauge is coated on the surface of the ship coating film Measuring the thickness of the, including the step of determining the painting state by analyzing the image acquired by the camera, the display unit, and displaying the analysis result, the step of determining the painting state, the deep neural network Through the deep learning technology using (DNN: Deep Neural Network), it is possible to determine the painting condition of the ship.

According to one embodiment, the step of determining the painting state is a step of collecting and storing an image of the ship's painting surface or a coating film according to the painting state, and learning the image pattern according to the painting state of the ship using the stored image And determining a coating state of the ship obtained from the camera or the film thickness meter according to the learning result.

The ship painting state inspection method using the image analysis according to an embodiment may further include providing a guide for a painting operation to a user according to the analysis result.

According to an embodiment, the step of providing the guide may provide a guide for completing the painting operation when it is determined that the painting is incomplete as a result of the analysis.

According to an embodiment, the step of providing the guide may provide a guide for removing a defect when it is determined that the ship's painted surface is defective as a result of the analysis.

According to the present invention, it is possible to provide a system and method for accurately and efficiently inspecting a ship's painting state through image analysis using deep learning.

In addition, according to the present invention, it is possible to provide an efficient painting operation by providing a guide for the painting operation to be performed according to the inspection result of the painting condition of the ship.

The effects of the present invention are not limited to the above-mentioned effects, and other effects not mentioned will be clearly understood by those skilled in the art from the following description.

1 is a view showing the configuration of a ship painting condition inspection system using image analysis according to an embodiment of the present invention.
2 is a diagram showing the configuration of a deep neural network (DNN) according to an embodiment of the present invention.
3 is a view showing the configuration of a learning unit according to an embodiment of the present invention.
4 is a view for explaining the operation of each component of the ship painting condition inspection system shown in FIG.
5 is a flowchart illustrating a ship painting condition inspection method using image analysis according to an embodiment of the present invention.
6 is a flowchart illustrating a process in which the inspection unit determines a painting state according to an embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. Advantages and features of the present invention, and methods for achieving them will be clarified with reference to embodiments described below in detail together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various different forms, and only the embodiments allow the publication of the present invention to be complete, and general knowledge in the technical field to which the present invention pertains. It is provided to fully inform the holder of the scope of the invention, and the invention is only defined by the scope of the claims. The same reference numerals refer to the same components throughout the specification.

Unless otherwise defined, all terms (including technical and scientific terms) used in the present specification may be used as meanings commonly understood by those skilled in the art to which the present invention pertains. In addition, terms defined in the commonly used dictionary are not ideally or excessively interpreted unless specifically defined. The terminology used herein is for describing the embodiments and is not intended to limit the present invention. In the present specification, the singular form also includes the plural form unless otherwise specified in the phrase.

In the drawings, parts not related to the description are omitted in order to clearly describe the present invention, and the same reference numerals are attached to the same or similar elements throughout the specification. And, when a part is said to “include” a certain component, this means that other components may be further included instead of excluding other components, unless otherwise stated. In addition, "part" described in the specification means a unit or block that performs a specific function.

1 is a view showing the configuration of a ship painting condition inspection system 10 using image analysis according to an embodiment of the present invention.

Referring to FIG. 1, the ship painting condition inspection system 10 using image analysis according to an embodiment of the present invention includes a camera 100, a film thickness meter 200, an inspection unit 300, a display unit 400, and a guide It may include a unit 500 and the control unit 600.

The camera 100 may acquire an image including the painted surface of the corresponding vessel by photographing the exterior of the vessel to determine the painted condition.

In the present specification, the image refers to visual information displayed and displayed on a 2D or 3D screen, and includes both a photographic image representing a temporary shape of the target object or a video image of the target object for a certain period of time. can do. In addition, the camera 100 may include all optical devices capable of acquiring a captured image of an object, and there is no limitation on the type and installation method.

The coating film thickness meter 200 may measure the thickness of the coating film painted on the surface of the ship. The film thickness meter 200 may measure the thickness of the coating film for a point on the surface of the ship, or measure the thickness of the coating film for multiple points arranged in the lateral or longitudinal direction to determine the maximum, minimum, or average value. Can be measured.

The inspection unit 300 may determine the coating state of the ship by analyzing the image obtained by the camera 100 or the thickness of the coating film measured by the film thickness meter 200.

According to an embodiment, when the inspection unit 300 analyzes an image acquired by the camera 100, the camera 100 acquires through deep learning using a deep neural network (DNN). By analyzing the image, it is possible to determine the painting condition of the ship included in the image.

Deep neural network (DNN) is one of models of deep learning that classifies input data based on pre-trained data, and builds multiple layers in one or more computers to build multiple layers of data. Refers to a system or network that performs judgment.

2 is a diagram illustrating the configuration of a deep neural network (DNN) according to an embodiment of the present invention.

Referring to FIG. 2, a deep neural network (DNN) may include an input layer, one or more hidden layers, and an output layer.

Data for learning is input to the input layer, and the result value calculated through the hidden layer and the output layer is compared with the ground truth, and the weight Update the value in reverse. After all the training is over, you can enter the information that requires prediction to get the result value.

According to one embodiment, the hidden layer (hidden layer) may include a convolution layer (convolution layer), a pooling layer (pooling layer) and a fully connected layer (fully connected layer).

The convolution layer may extract a feature map and perform a convolution operation on the image input to the input layer.

The pooling layer is connected to the convolution layer to perform subsampling on the output of the convolution layer.

The fully connected layer can be connected to the pooling layer to learn the output of the subsampled pooling layer and learn according to the category to be output to the output layer.

The deep neural network (DNN) according to an embodiment of the present invention may be a convolutional neural network (CNN), a recurrent neural network (RNN), or an artificial neural network having a similar structure.

3 is a view showing the configuration of the inspection unit 300 according to an embodiment of the present invention.

Referring to FIG. 3, the inspection unit 300 according to an embodiment of the present invention may include a storage unit 310, a learning unit 320, and an analysis unit 330.

The storage unit 310 may collect and store images of the ship's painted surface and thickness values of the coating film according to the painted state.

The image stored in the storage unit 310 is used for deep learning using the learning unit 320 to be described later, and may include images according to various painting conditions, and images directly photographed by the camera 100 or It may include an image received from an external server.

According to an embodiment, the storage unit 310 may collect and store images of a ship's painting surface according to various painting conditions according to whether the painting is completed, whether the painting surface is defective, and the painting quality including the thickness of the painting. That is, the storage unit 310 stores the image of each ship in the painting operation, the image of the defect on the painted surface, and the image of the ship's painted surface, including the image according to the coating quality, including the thickness of the coating film, and learning to be described later. The unit 320 may use this as input data to learn an image pattern according to the painting state.

For example, in the step of painting the surface of the ship, the storage unit 310 collects and stores an image of the ship's painting surface according to each stage of primer, middle, and top, or collects and stores various defect images generated on the ship's painting surface. Can be. In addition, the storage unit 310 may collect and store a value for the average thickness of the coating film on the coating surface according to each of the above-described coating steps of coating.

The learning unit 320 may learn an image pattern according to a ship's painting state using an image stored in the storage unit 310 using a deep neural network (DNN).

More specifically, the learning unit 320 inputs an image stored in the storage unit 310 in an input layer on a deep neural network (DNN), and passes through a hidden layer in an output layer. By comparing the output value with the actual painting state, the weight can be updated to learn the image pattern according to the painting state of the ship.

According to one embodiment, the learning unit 320 may perform learning on the input data using any one of supervised-learning or reinforcement-learning. Supervised-learning and reinforcement-learning are a type of machine learning. After inputting the input data and target data to be outputted together, learning is performed, and then prediction is performed on the completed neural network. This is a technique to output the appropriate result by inputting the desired data.

As described above, the learning unit 320 receives data of an actual painting state with respect to an image input for learning, and each line constituting a deep neural network (DNN) to output data of the actual painting state in response to the input data. By updating the weight of, you can learn the image pattern according to the painting condition of the ship.

For example, the user can pre-classify the steps of the painting process according to a predetermined criterion, and update the weights so that when each image is input, data corresponding to the painting process progress steps classified by the actual user is output.

As another example, the user may classify the painting quality according to the painting state in advance according to a certain criterion, and update the weight so that when each image is input, data corresponding to the painting quality classified by the actual user is output.

However, this is for illustration only, and the learning unit 230 may perform learning on the painting state of the ship arbitrarily classified by the user using the same method.

The analysis unit 330 may use the learning result of the learning unit 320 to determine the paint state of the ship photographed by the camera 100 or the film thickness meter 200 to measure the thickness of the coating film.

According to an embodiment of the present disclosure, the analysis unit 330 may determine whether the ship has been painted, whether there is a defect on the painted surface, the quality of the coating, and the coating condition according to the thickness of the coating film.

More specifically, the analysis unit 330 may determine whether the painting operation of the ship included in the image obtained from the camera 100 is completed or, if not, which stage of the operation is currently in progress. For example, the analysis unit 330 determines whether the vessel in the image is currently in the finished work, the intermediate work is completed, or the finished work is completed, and the current painting work progress is determined. You can do it. However, the classification according to each of the lower, middle, and upper stages is for illustration only and does not limit the embodiment of the present invention, and can be classified into various stages according to the user's setting.

In addition, the analysis unit 330 may determine whether there is a defect on the painted surface of the ship included in the image obtained from the camera 100. In addition, when the defect exists on the painted surface of the ship in the image, the analysis unit 330 may analyze information including the type, location, and size of the defect.

In addition, the analysis unit 330 may determine the painting quality of the ship included in the image obtained from the camera 100. The painting quality can be arbitrarily classified by the user according to a certain criterion, and the analysis unit 330 is included in the image previously acquired by the learning unit 320 and the image obtained based on the painting quality classification for the image. It is possible to judge the painting quality of a ship.

In addition, the analysis unit 330 may use the thickness of the vessel coating film measured by the coating film thickness meter 200 to determine the current state of the coating of the ship, including the current coating step or defects. For example, when the thickness of the coating film measured by the coating film thickness meter 200 for a plurality of points on the surface of the ship is uniform, the analysis unit 330 is said to be in the process of painting corresponding to the thickness of the uniform coating film. I can judge. Alternatively, if there is a portion of the coating film thickness meter 200 having a non-uniform thickness of the coating film among a plurality of points on the surface of the ship, it may be determined that there is a defect or the coating has not been completed.

Returning to the description of FIG. 1 again, the rest of the configuration of the painting condition inspection system 10 will be described.

The display 400 may display an image obtained by the camera 100 and an analysis result of the inspection unit 300 for the image.

The guide unit 500 may provide a guide for painting to the user according to the analysis result of the inspection unit 300. That is, the guide unit 500 is determined by the analysis of the inspection unit 300 as if the painting of the ship included in the image obtained by the camera 100 is not completed, or when it is determined that a defect exists. If there is currently a painting operation to be performed, a guide can be provided so that the user can perform it according to an appropriate method.

According to an embodiment, the guide unit 500 may provide a guide for completing the painting operation when it is determined that the painting is incomplete as a result of the analysis of the inspection unit 300. That is, after determining the current painting state of the ship through the analysis of the inspection unit 300, if the painting has not yet been completed, the guide unit 500 provides the user with information about the next task to be performed to complete it. can do.

According to an embodiment, the guide unit 500 may provide a guide for removing defects when it is determined that the ship's painted surface is defective as the result of the analysis of the inspection unit 300. In this case, the guide unit 500 may provide information on the type, location, and size of the defect and information on a work process to be performed to remove the defect.

According to an embodiment, the guide provided by the guide unit 500 may be displayed to the user by the display unit 400.

The controller 600 may control the overall operation of the ship painting condition inspection system 10 using image analysis.

4 is a diagram schematically showing the operation of each component of the ship painting condition inspection system 10 using the image analysis described above.

Referring to FIG. 4, the storage unit 310 collects and stores a plurality of images including a vessel painting surface, and the learning unit 320 uses this to learn image patterns according to the painting state.

As such, after learning using the learning unit 320 is completed, the camera 100 acquires an image by photographing a ship to be inspected and transmits it to the analysis unit 330.

The analysis unit 330 determines the painting state of the ship in the image acquired by the camera 100 based on the learning result of the learning unit 320, and the guide unit 500 analyzes the analysis result of the analysis unit 330 According to this, a user can provide a guide for painting.

The display unit 400 may display the image of the ship acquired by the camera 100, the analysis result of the analysis unit 330, and guide information of the guide unit 500 to the user.

Each component of the ship painting condition inspection system 10 using the above-described image analysis can be controlled by the control unit 600.

Meanwhile, a ship painting condition inspection method using the ship painting condition inspection system 10 using the above-described image analysis may be provided.

5 is a view showing a ship painting state inspection method using image analysis according to an embodiment of the present invention.

Referring to Figure 5, the ship painting state inspection method using the image analysis according to an embodiment of the present invention, the camera (100) photographing the appearance of the ship (S510), the film thickness meter 200 of the ship Measuring the thickness of the coating film coated on the surface (S520), the inspection unit 300 analyzes the thickness of the image obtained by the camera 100 and the thickness of the coating film measured by the film thickness meter 200 to determine the coating state. Step (S530) and the display unit 400 may include the step of displaying the analysis result (S540).

According to an embodiment, the step of determining the painting state may determine the painting state of the ship through deep learning technology using a deep neural network (DNN).

With regard to the deep neural network (DNN), the detailed description of the coating state inspection system 10 has been described above, so a detailed description thereof will be omitted.

6 is a flowchart illustrating a process in which the inspection unit 300 determines a painting state using an image acquired by the camera 100 according to an embodiment of the present invention.

According to one embodiment, the step of determining the painting state (S520) is collecting and storing an image of the ship's painting surface or the thickness of the coating film according to the painting state (S610), and painting the ship using the stored image It may include the step of learning the image pattern according to the state (S620) and the step of determining the coating state of the ship obtained from the camera 100 or the film thickness meter 200 according to the learning result (S630).

Meanwhile, a method for inspecting a painting state using image analysis according to an embodiment may further include providing a guide for a painting operation to a user according to the analysis result.

According to an embodiment, the step of providing the guide may provide a guide for completing the painting operation when it is determined that the painting is incomplete as a result of the analysis. That is, after the step (S520) of determining the painting state, if the painting result has not yet been completed, the user may be provided with information on the next task to be performed to complete the painting.

According to an embodiment, the step of providing the guide may provide a guide for removing a defect when it is determined that the ship's painted surface is defective as a result of the analysis. That is, after the step of determining the painting state (S520), if it is determined that there is a defect on the surface of the ship coating as a result of determination, a guide for removing the defect may be provided. In this case, the guide unit 500 may provide information on the type, location, and size of the defect and information on a work process to be performed to remove the defect.

The system and method for inspecting the painted state of a ship using image analysis according to an embodiment of the present invention has been described above. According to the present invention, it is possible to accurately and efficiently inspect the painting state of a ship through image analysis using deep learning, and a guide to the painting task to be performed according to the inspection result of the painting state of the ship. By providing, it is possible to make an efficient painting operation.

In addition, each step of the ship painting condition inspection method using image analysis may be performed by each component of the ship painting condition inspection system 10, and the above-described image even though not described in detail in the description of the method Needless to say, a method for inspecting a ship's paint condition using image analysis according to various embodiments using functions performed by components of the ship's paint condition inspection system 10 using analysis may be provided.

The embodiments of the present invention described above are disclosed for purposes of illustration, and the present invention is not limited thereto. In addition, a person having ordinary knowledge in the technical field of the present invention can make various modifications and changes within the spirit and scope of the present invention, and such modifications and changes should be regarded as belonging to the scope of the present invention.

10: Ship painting condition inspection system
100: camera 200: film thickness meter
300: inspection unit 310: storage unit
320: learning unit 330: analysis unit
400: display unit 500: guide unit
600: control unit

Claims (13)

A camera that acquires an image by photographing the exterior of a ship;
A film thickness meter for measuring the thickness of a coating film painted on the surface of a ship;
An inspection unit that analyzes the thickness of the coating film measured by the image or the film thickness meter acquired by the camera to determine the coating state of the ship; And
A display unit that displays an analysis result of the analysis unit;
Including,
The inspection unit is a ship painting condition inspection system using image analysis to determine the painting condition of the ship through deep learning technology using a deep neural network (DNN).
According to claim 1,
The inspection unit,
Ship painting condition inspection system using image analysis characterized in that it determines the painting condition including any one or more of whether the painting is completed, whether the painting surface is defective, or the quality of the painting.
According to claim 1,
The inspection unit,
A storage unit for collecting and storing the image of the ship's painting surface or the thickness of the coating film according to the painting state;
A learning unit for learning an image pattern according to a ship painting state using an image stored in the storage unit; And
An analysis unit for determining a painting state of the ship obtained from the camera or the film thickness meter according to the learning result;
Ship painting condition inspection system using image analysis, characterized in that it comprises a.
According to claim 1,
The learning unit,
Ship painting condition inspection system using image analysis, characterized by learning the image of the ship's painting surface according to each stage of top, middle, and bottom.
According to claim 1,
The learning unit,
Ship painting condition inspection system using image analysis characterized by learning the defect image of the ship's painting surface.
According to claim 1,
A guide unit providing a guide for painting work to a user according to the analysis result of the inspection unit;
Ship painting condition inspection system using image analysis, characterized in that it further comprises.
The method of claim 6,
The guide portion,
Ship analysis status inspection system using image analysis characterized in that it provides a guide for completing the painting operation when it is determined that the analysis result of the analysis unit is incomplete.
The method of claim 6,
The guide portion,
Ship analysis condition inspection system using image analysis characterized in that it provides a guide for removing defects when it is determined that the analysis part of the analysis part is defective on the painted surface of the ship.
A camera photographing the exterior of the ship;
Measuring the thickness of the coating film painted on the surface of the ship by a film thickness meter;
Determining a coating state by analyzing an image obtained by the camera or a thickness of the coating film measured by the coating thickness meter; And
Displaying a result of the analysis by a display unit;
Including,
The step of determining the painting state,
Vessel painting condition inspection method using image analysis to determine the painting condition of the ship through deep learning technology using deep neural network (DNN).
The method of claim 9,
The step of determining the painting state,
Collecting and storing the image of the ship's painting surface or the thickness of the coating film according to the painting state;
Learning an image pattern according to a ship painting state using the stored image; And
Determining a painting state of a ship obtained from the camera or a film thickness meter according to the learning result;
Ship painting state inspection method using the image analysis, characterized in that it comprises a.
The method of claim 9,
Providing a guide for the painting work to the user according to the analysis results;
Ship painting state inspection method using the image analysis, characterized in that it further comprises.
The method of claim 11,
The step of providing the guide,
A ship inspection condition inspection method using image analysis, characterized in that a guide for completing the painting operation is provided when it is determined that the painting is incomplete.
The method of claim 11,
The step of providing the guide,
When it is determined that the ship's painted surface is defective as a result of the analysis, a method for inspecting the painted state of a ship using image analysis is characterized by providing a guide for removing defects.

Images