An AI intelligent 3D spraying control system for urban art device surface treatment

Abstract

The application provides an AI intelligent 3D spraying control system for urban art device surface treatment, and belongs to the technical field of control. Through a trained neural network model, only the track of part of layers can be generated through images in the process of multi-layer image spraying construction, and the tracks of the remaining layers are generated by using the neural network model, so that the calculation power when directly generating the track according to the image is reduced, the total calculation time is improved, and the spraying work efficiency is further improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of control, in particular to an AI intelligent 3D spraying control system for surface treatment of urban art installations. BACKGROUND

[0002] The 3D device for surface treatment of urban art installations can accurately control the spraying process, ensuring uniform and beautiful coating. This advanced technology not only improves spraying efficiency, but also greatly improves the visual effect of art installations. Through precise programming and real-time monitoring, the 3D spraying control system can adjust the spraying angle, speed and thickness according to different artistic needs, thus achieving perfect reproduction of complex patterns and textures.

[0003] In the actual use of the 3D spraying device, multiple spraying trajectories corresponding to multiple layer images need to be generated according to multiple layer images of the original spraying image. However, generating trajectories according to multiple layer images consumes a large amount of computing power, resulting in a longer time consumed in trajectory planning and lower overall efficiency. SUMMARY

[0004] The present application provides an AI intelligent 3D spraying control system for surface treatment of urban art installations to improve the above problems.

[0005] To achieve the above purpose, the present application adopts the following technical solutions:

[0006] In a first aspect, the present application provides an AI intelligent 3D spraying control system for surface treatment of urban art installations, which includes a controller and an execution terminal, and is configured to:

[0007] The controller acquires spraying data, which is image data that needs to be sprayed, and includes n layers and n layer images corresponding to the n layers, i.e., the 1st layer image, the 2nd layer image,..., and the nth layer image, where n is a natural number greater than or equal to 3;

[0008] The controller acquires the m-jth layer image and the m+jth layer image, and generates the m-jth layer trajectory and the m+jth layer trajectory according to the m-jth layer image and the m+jth layer image, where m and j are natural numbers not equal to 0, n > m, and m > j;

[0009] The controller acquires a trained neural network model, and inputs the m-jth layer trajectory and the m+jth layer trajectory into the trained neural network model, which is used to output a predicted layer trajectory of the intermediate layer of the two layer images according to the input two layer trajectories;

[0010] The controller obtains the m-th layer trajectory according to the output result of the neural network training model, and controls the execution terminal to perform spraying construction on the m-th layer image according to the m-th layer trajectory.

[0011] In combination with the first aspect, in some embodiments, the system is configured to:

[0012] The controller obtains the m-th layer image and the m+j-th layer image, and generates the m-th layer trajectory and the m+j-th layer trajectory according to the m-th layer image and the m+j-th layer image, wherein m and j are natural numbers other than 0, and n > m and m > j, including:

[0013] The controller obtains the m-th layer image and divides the m-th layer image into a plurality of sub-images, and the plurality of sub-images form the m-th layer image.

[0014] If the controller determines that at least one side of a sub-image is not adjacent to another sub-image, the sub-image is determined to be a target sub-image.

[0015] The controller obtains all target sub-images, takes the target sub-images as nodes, and takes the connection between two adjacent target sub-images as a path, and obtains a plurality of first Euler paths formed by the target sub-images.

[0016] The controller determines a plurality of target regions based on the positional relationship between the plurality of first Euler paths, and the target region is a region separated by the first Euler path as a boundary.

[0017] The controller determines a filling region in the plurality of target regions based on the m-th layer image, and determines the m-th layer trajectory according to the filling region.

[0018] In combination with the first aspect, in some embodiments, the system is configured to:

[0019] The controller determines a filling region in the plurality of target regions based on the m-th layer image, and determines the m-th layer trajectory according to the filling region, including:

[0020] The controller obtains a sub-image in each filling region, takes the sub-image as a node, and takes the connection between two adjacent sub-images as a path, and obtains a plurality of second Euler paths formed by the sub-images.

[0021] The controller generates the m-th layer trajectory based on the plurality of second Euler paths.

[0022] In combination with the first aspect, in some embodiments, the system is configured to:

[0023] The controller generates the m-th layer trajectory based on the plurality of second Euler paths, including:

[0024] Based on the shortest path principle, the controller connects multiple second Euler paths to form the trajectory of the mj layer.

[0025] In conjunction with the first aspect, in some implementations, the system is configured as follows:

[0026] The controller acquires a pre-trained neural network model and inputs the trajectory of the mj-th layer and the trajectory of the m+j-th layer into the pre-trained neural network model. The neural network model is used to output the predicted trajectory of the intermediate layer between the two input layer trajectories, including:

[0027] The controller acquires a training set, which includes k adjacent layer trajectories, where k is a natural number greater than or equal to 3;

[0028] The controller imports the training set into the neural network model and performs iterative calculations. When the iterative calculation process meets the preset conditions, the calculation stops and the trained neural network model is obtained.

[0029] In conjunction with the first aspect, in some implementations, the system is configured as follows:

[0030] The controller imports the training set into the neural network model and performs iterative calculations. When the iterative calculation process meets preset conditions, it stops calculating and obtains the trained neural network model, including:

[0031] The controller selects the (i-1)th layer trajectory, the ith layer trajectory, and the (i+1)th layer trajectory from the k-th layer trajectory, where i is a non-zero natural number, and i <k;

[0032] The controller uses the data of the (i-1)th layer trajectory and the data of the (i+1)th layer trajectory as two different dimension vectors to form the feature vector of the i-th layer trajectory data and performs iterative calculation.

[0033] The calculation stops when the iterative calculation process meets the preset conditions, and the controller obtains the trained neural network model.

[0034] In conjunction with the first aspect, in some implementations, the system is configured as follows:

[0035] The calculation stops when the iterative calculation process meets the preset conditions, and the trained neural network model is obtained, including:

[0036] The controller, based on the neural network model, obtains k-2 error values ​​according to the training process of the training set, and determines the average error value based on the k-2 error values;

[0037] The controller compares the average error value with the preset value. When the average error value is less than or equal to the preset value, it stops training and outputs the trained neural network model.

[0038] In conjunction with the first aspect, in some implementations, the system is configured as follows:

[0039] The controller, based on the neural network model, obtains k-2 error values ​​during the training process using the training set, and determines the average error value based on these k-2 error values, including:

[0040] The controller inputs the data of the (i-1)th layer trajectory and the data of the (i+1)th layer trajectory into the neural network training model, and obtains the predicted layer trajectory output by the neural network training model;

[0041] The controller determines an error value based on the data of the predicted layer trajectory output by the neural network training model and the data of the i-th layer trajectory.

[0042] In conjunction with the first aspect, in some implementations, the system is configured as follows:

[0043] The controller determines an error value based on the predicted layer trajectory data output by the neural network training model and the trajectory data of the i-th layer, satisfying the following:

[0044] If the data for the predicted layer trajectory is greater than the data for the i-th layer trajectory, then the error value is determined to be:

[0045] t = y(yx)

[0046] Where t is the error value, y is the data of the predicted layer trajectory, and x is the data of the i-th layer trajectory.

[0047] In conjunction with the first aspect, in some implementations, the system is configured as follows:

[0048] The controller determines an error value based on the predicted layer trajectory data output by the neural network training model and the trajectory data of the i-th layer, satisfying the following:

[0049] If the data for the predicted layer trajectory is less than the data for the i-th layer trajectory, then the error value is determined to be:

[0050] t = y(xy)

[0051] Where t is the error value, y is the data of the predicted layer trajectory, and x is the data of the i-th layer trajectory.

[0052] The second aspect of this invention proposes an AI-powered intelligent 3D spraying control method for surface treatment of urban art installations, applicable to an AI-powered intelligent 3D spraying control system for surface treatment of urban art installations. The system includes a controller and an execution terminal, and the method includes:

[0053] The controller acquires the spraying data, which is the image data to be sprayed. The spraying data includes n layers and n layer images corresponding to the n layers, namely the first layer image, the second layer image, ..., the nth layer image, where n is a natural number greater than or equal to 3.

[0054] The controller acquires the image of the mj-th layer and the image of the m+j-th layer, and generates the trajectory of the mj-th layer and the trajectory of the m+j-th layer based on the image of the mj-th layer and the image of the m+j-th layer, where m and j are both non-zero natural numbers, and n>m, m>j;

[0055] The controller acquires the pre-trained neural network model and inputs the trajectory of the mj-th layer and the trajectory of the m+j-th layer into the pre-trained neural network model. The neural network model is used to output the predicted trajectory of the intermediate layer between the two layers of the image based on the input trajectories of the two layers.

[0056] The controller obtains the trajectory of the m-th layer based on the output of the neural network training model, and controls the execution terminal to perform spraying on the m-th layer image based on the trajectory of the m-th layer.

[0057] In conjunction with the second aspect, in some implementations, the controller acquires the image of the mj-th layer and the image of the (m+j-th)-th layer, and generates the trajectory of the mj-th layer and the trajectory of the (m+j-th)-th layer based on the image of the mj-th layer and the image of the (m+j-th)-th layer, where m and j are both non-zero natural numbers, and n>m, m>j, including:

[0058] The controller acquires the image of the mj-th layer and divides the image of the mj-th layer into multiple sub-images, which together form the image of the mj-th layer.

[0059] If the controller determines that at least one side of a sub-image is not adjacent to another sub-image, then the sub-image is determined to be the target sub-image;

[0060] The controller acquires all target sub-images, and uses the target sub-images as nodes and the lines between two adjacent target sub-images as paths to obtain multiple first Euler paths formed by the target sub-images.

[0061] The controller determines multiple target regions based on the positional relationship between multiple first Euler paths. Each target region is a region divided by the first Euler paths.

[0062] The controller determines the filled region among multiple target regions based on the image of the mj layer, and determines the trajectory of the mj layer based on the filled region.

[0063] In conjunction with the second aspect, in some implementations, the controller determines a filled region among multiple target regions based on the image of the mj-th layer, and determines the trajectory of the mj-th layer based on the filled region, including:

[0064] The controller acquires the sub-image in each filled region, and uses the sub-image as a node and the line connecting two adjacent sub-images as a path to obtain multiple second Euler paths formed by the sub-images.

[0065] The controller generates the trajectory of the mj-th layer based on multiple second Euler paths.

[0066] In conjunction with the second aspect, in some implementations, the controller generates the trajectory of the mj-th layer based on multiple second Euler paths, including:

[0067] Based on the shortest path principle, the controller connects multiple second Euler paths to form the trajectory of the mj layer.

[0068] In conjunction with the second aspect, in some implementations, the controller acquires a pre-trained neural network model and inputs the trajectory of the mj-th layer and the trajectory of the m+j-th layer into the pre-trained neural network model. The neural network model is used to output the predicted trajectory of the intermediate layer between the two image layers based on the input trajectories of the two layers, including:

[0069] The controller acquires a training set, which includes k adjacent layer trajectories, where k is a natural number greater than or equal to 3;

[0070] The controller imports the training set into the neural network model and performs iterative calculations. When the iterative calculation process meets the preset conditions, the calculation stops and the trained neural network model is obtained.

[0071] In conjunction with the second aspect, in some implementations, the controller imports the training set into the neural network model and performs iterative calculations. When the iterative calculation process meets preset conditions, the calculation stops, and the trained neural network training model is obtained, including:

[0072] The controller selects the (i-1)th layer trajectory, the ith layer trajectory, and the (i+1)th layer trajectory from the k-th layer trajectory, where i is a non-zero natural number, and i <k;

[0073] The controller uses the data of the (i-1)th layer trajectory and the data of the (i+1)th layer trajectory as two different dimension vectors to form the feature vector of the i-th layer trajectory data and performs iterative calculation.

[0074] The calculation stops when the iterative calculation process meets the preset conditions, and the controller obtains the trained neural network model.

[0075] In conjunction with the second aspect, in some implementations, the calculation stops when the iterative calculation process meets a preset condition, and a trained neural network training model is obtained, including:

[0076] The controller, based on the neural network model, obtains k-2 error values ​​according to the training process of the training set, and determines the average error value based on the k-2 error values;

[0077] The controller compares the average error value with the preset value. When the average error value is less than or equal to the preset value, it stops training and outputs the trained neural network model.

[0078] In conjunction with the second aspect, in some implementations, the controller, based on the neural network model and the training process of the training set, obtains k-2 error values ​​and determines an average error value based on the k-2 error values, including:

[0079] The controller inputs the data of the (i-1)th layer trajectory and the data of the (i+1)th layer trajectory into the neural network training model, and obtains the predicted layer trajectory output by the neural network training model;

[0080] The controller determines an error value based on the data of the predicted layer trajectory output by the neural network training model and the data of the i-th layer trajectory.

[0081] In conjunction with the second aspect, in some implementations, the controller determines an error value based on the predicted layer trajectory data output by the neural network training model and the i-th layer trajectory data, satisfying:

[0082] If the data for the predicted layer trajectory is greater than the data for the i-th layer trajectory, then the error value is determined to be:

[0083] t = y(yx)

[0084] Where t is the error value, y is the data of the predicted layer trajectory, and x is the data of the i-th layer trajectory.

[0085] In conjunction with the second aspect, in some implementations, the controller determines an error value based on the predicted layer trajectory data output by the neural network training model and the i-th layer trajectory data, satisfying:

[0086] If the data for the predicted layer trajectory is less than the data for the i-th layer trajectory, then the error value is determined to be:

[0087] t = y(xy)

[0088] Where t is the error value, y is the data of the predicted layer trajectory, and x is the data of the i-th layer trajectory.

[0089] A third aspect of this invention provides an electronic device, which includes:

[0090] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method proposed in the first aspect of the present invention.

[0091] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method as described in the first aspect of the present invention.

[0092] In summary, the above method and apparatus have the following technical effects:

[0093] This application proposes an AI-powered intelligent 3D spraying control system for surface treatment of urban art installations. First, the controller acquires spraying data, which consists of image data to be sprayed. This data includes n layers and corresponding n layer images, namely, layer 1, layer 2, ..., layer n, where n is a natural number greater than or equal to 3. Then, the controller acquires the mj-th layer image and the (m+j-th)-th layer image, and generates the mj-th layer trajectory and the (m+j-th)-th layer trajectory based on these images, where m and j are non-zero natural numbers, and n > m and m > j. Next, the controller acquires a pre-trained neural network model and inputs the mj-th layer trajectory and the (m+j-th)-th layer trajectory into it. The neural network model outputs a predicted layer trajectory between the two input layer trajectories. Finally, the controller acquires the m-th layer trajectory based on the output of the neural network training model and controls the execution terminal to perform spraying on the m-th layer image based on this trajectory. This application proposes an AI-powered intelligent 3D spraying control system for surface treatment of urban art installations. Through a trained neural network model, the system can generate the trajectory of only some layers from images during multi-layer image spraying construction, while the trajectory of the remaining layers is generated using the neural network model. This reduces the computational effort required to directly generate trajectories from images, increases the total computation time, and further improves the efficiency of the spraying work. Attached Figure Description

[0094] Figure 1 This is a flowchart illustrating an AI-powered intelligent 3D spraying control method for surface treatment of urban art installations, as proposed in an embodiment of this application. Detailed Implementation

[0095] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0096] This application proposes an AI-powered intelligent 3D spraying control method for surface treatment of urban art installations. It is applicable to an AI-powered intelligent 3D spraying control system for surface treatment of urban art installations. The system includes a controller and an execution terminal. The execution terminal can be a printing device or a spraying device, which is not limited in this application.

[0097] Please see Figure 1 The method includes the following steps:

[0098] S101: The controller acquires the spraying data, which is the image data to be sprayed. The spraying data includes n layers and n layer images corresponding to the n layers, namely the first layer image, the second layer image, ..., the nth layer image, where n is a natural number greater than or equal to 3.

[0099] Understandably, before spraying, the controller needs to acquire spraying data and control the execution terminal to actually perform spraying or printing based on the data. The working principle of 3D spraying has been disclosed in relevant patents and will not be elaborated upon in this application. During the 3D spraying process, spraying is performed on each layer according to a predetermined route. In this application, the spraying data includes n layers and n layer images corresponding to the n layers, where n can be any natural number greater than or equal to 3.

[0100] S102: The controller acquires the image of the mj-th layer and the image of the m+j-th layer, and generates the trajectory of the mj-th layer and the trajectory of the m+j-th layer based on the image of the mj-th layer and the image of the m+j-th layer, where m and j are both non-zero natural numbers, and n>m, m>j.

[0101] Understandably, during the spraying process, the controller needs to design the actual spraying route based on the image of the layer. Directly designing the route based on the image requires a large amount of computing power. In this embodiment, the path derived from the neural network model is the trajectory of the m-th layer, that is, the trajectory of the m-th layer is between the trajectory of the mj-th layer image and the trajectory of the (m+j)-th layer image.

[0102] Specifically, in this embodiment, taking the image of the mj layer as an example, the process of obtaining the spray path from the image may include the following steps:

[0103] S1021: The controller acquires the image of the mj-th layer and divides the image of the mj-th layer into multiple sub-images, which together form the image of the mj-th layer.

[0104] It is understandable that in the process of segmenting an image, the smaller the sub-image, the less refined it is. In this embodiment, the sub-image can be the smallest area in the actual spraying process of the execution terminal. Of course, it can also be some other calculation methods, which are not limited in this embodiment.

[0105] S1022: If the controller determines that at least one side of a sub-image is not adjacent to another sub-image, then the sub-image is determined to be the target sub-image.

[0106] Understandably, after the controller divides the image into multiple sub-images, if at least one side of a sub-image is not adjacent to another sub-image, it can be proven that the sub-image is located at the edge of the entire image. In this way, one side of the sub-image will not have any sub-image adjacent to it, while the other three sides or two sides will be adjacent to other sub-images. This method can determine whether a sub-image is an edge sub-image. In this embodiment, for ease of explanation, the sub-image confirmed as an edge is defined as the target sub-image.

[0107] S1023: The controller acquires all target sub-images, and uses the target sub-images as nodes and the lines connecting two adjacent target sub-images as paths to acquire multiple first Euler paths formed by the target sub-images.

[0108] After the target sub-image is determined, it is understood that multiple consecutive sub-images are the boundaries of the parts of the entire image that need to be sprayed. Therefore, in this embodiment, the target sub-image is used as a node, and the line connecting two adjacent target sub-images is used as a path to obtain multiple first Euler paths formed by the target sub-images. In this way, the first Euler paths can be used as the boundaries of the spraying area, including, for example, the boundaries of the hollowed-out parts of the image.

[0109] S1024: The controller determines multiple target regions based on the positional relationship between multiple first Euler paths. The target region is a region divided by the first Euler paths.

[0110] Understandably, after determining multiple boundaries, the entire area is divided into multiple regions based on the positional relationship of these boundaries, such as whether they are nested. These regions may be areas that need to be painted or areas that need to be cut out.

[0111] S1025: The controller determines the filling region among multiple target regions based on the image of the mj layer, and determines the trajectory of the mj layer based on the filling region.

[0112] Understandably, after identifying multiple regions, we can combine the image of layer mj to determine whether the region is a filled region or a cutout region. After identifying multiple filled regions, we can determine the corresponding image trajectory of the layer based on the distribution of the multiple filled regions.

[0113] Understandably, in order to completely fill each filling area, in this embodiment, sub-images within each filling area can be obtained. Using the sub-images as nodes and the lines connecting adjacent sub-images as paths, multiple second Eulerian paths are generated from the sub-images. In this way, the second Eulerian paths traverse the area containing each sub-image. Connecting these multiple second Eulerian paths can also generate the trajectory of the mj-th layer. It should be noted that the generation process should be based on the shortest path principle, connecting multiple second Eulerian paths to form the trajectory of the mj-th layer.

[0114] S103: The controller acquires the trained neural network model and inputs the trajectory of the mj-th layer and the trajectory of the m+j-th layer into the trained neural network model. The neural network model is used to output the predicted trajectory of the intermediate layer of the two images based on the input trajectories of the two layers.

[0115] Understandably, in this embodiment, inputting the trajectory of layer mj and layer m+j into the pre-trained neural network model yields the trajectory of layer m. Thus, this method can be used to generate trajectories for all layers except for layer 1 and layer n. Therefore, considering practical applications, the generated layer trajectories can also be used as input to the neural network training model, reducing the number of layers the system needs to directly generate. Of course, the proportion of layers that can be directly generated from images and images that can be generated by the neural network model can be determined according to the user's needs; this embodiment does not impose any limitations.

[0116] In this embodiment, steps S1031-S1032 can be used to obtain a trained neural network model. Of course, in other embodiments, there may be other neural network models with different calculation or generation methods than those in this embodiment, which are not limited in this embodiment.

[0117] S1031: The controller acquires a training set, which includes k adjacent layer trajectories, where k is a natural number greater than or equal to 3.

[0118] For example, the training set consists of trajectories generated from images for each layer, with a total of k layers, and the trajectories in each layer are adjacent trajectories.

[0119] S1032: The controller imports the training set into the neural network model and performs iterative calculations. When the iterative calculation process meets the preset conditions, the calculation stops, and a trained neural network training model is obtained.

[0120] Specifically, the controller selects the (i - 1)-th layer track, the i-th layer track, and the (i + 1)-th layer track from the k layer tracks. Here, i is a natural number other than 0, and i < k. Then, the controller uses the data of the (i - 1)-th layer track and the data of the (i + 1)-th layer track as two different dimensional vectors to form the feature vector of the data of the i-th layer track and performs iterative calculations. When the iterative calculation process meets the preset conditions, the calculation stops, and the controller obtains a trained neural network training model.

[0121] As an implementation, for the determination of the error value, based on the process of training the neural network model with the training set, k - 2 error values can be obtained, and the average error value is determined based on the k - 2 error values.

[0122] The controller can also adopt the method of cross-validation. Specifically, the training set is divided into several subsets, and one of the subsets is used as the validation set in turn, and the rest are used as the training set for model training. In this way, the overfitting phenomenon can be effectively avoided, and the generalization ability of the model on different data sets can be ensured.

[0123] Then, the controller compares the average error value with the preset value. When the average error value is less than or equal to the preset value, the training stops and the trained neural network model is output.

[0124] It can be understood that when the average error value is less than or equal to the preset value, the training stops and the trained neural network model is output. This not only improves the accuracy of the spraying effect, but also greatly shortens the model training time and improves the overall work efficiency.

[0125] In summary, the AI intelligent 3D spraying control system provides an efficient, accurate and personalized solution for the surface treatment of urban art installations, and has broad application prospects and market potential.

[0126] As an implementation, if the data of the predicted layer track is greater than the data of the i-th layer track, the error value is determined as:

[0127] t = y(y - x)

[0128] If the data of the predicted layer track is less than the data of the i-th layer track, the error value is determined as:

[0129] t = y(x - y)

[0130] Where, t is the error value, y is the data of the predicted layer track, and x is the data of the i-th layer track.

[0131] S104: The controller obtains the trajectory of the m-th layer based on the output of the neural network training model, and controls the execution terminal to perform spraying construction on the m-th layer image based on the trajectory of the m-th layer.

[0132] Understandably, this method can be used to generate trajectories for all layers except for layer 1 and layer n. Therefore, considering practical use cases, the generated layer trajectories can also be used as input sources into the neural network training model, reducing the number of layers the system needs to directly generate.

[0133] This application proposes an AI-powered intelligent 3D spraying control method for surface treatment of urban art installations. First, the controller acquires spraying data, which consists of image data to be sprayed. This data includes n layers and corresponding n layer images, namely, layer 1, layer 2, ..., layer n, where n is a natural number greater than or equal to 3. Then, the controller acquires the mj-th layer image and the (m+j-th)-th layer image, and generates the mj-th layer trajectory and the (m+j-th)-th layer trajectory based on these images, where m and j are non-zero natural numbers, and n > m and m > j. Next, the controller acquires a pre-trained neural network model and inputs the mj-th layer trajectory and the (m+j-th)-th layer trajectory into it. The neural network model outputs a predicted layer trajectory between the two input layer trajectories. Finally, the controller acquires the m-th layer trajectory based on the output of the neural network training model and controls the execution terminal to perform spraying on the m-th layer image based on this trajectory. This application proposes an AI-powered intelligent 3D spraying control method for surface treatment of urban art installations. Through a trained neural network model, the method can generate the trajectory of only some layers from the image during the multi-layer image spraying process, while the trajectory of the remaining layers is generated by the neural network model. This reduces the computational effort required to directly generate the trajectory from the image, increases the total computation time, and further improves the efficiency of the spraying work.

[0134] Based on the same inventive concept, this application also proposes an AI intelligent 3D spraying control system for surface treatment of urban art installations. The system includes a controller and an execution terminal, and is configured as follows:

[0135] The controller acquires the spraying data, which is the image data to be sprayed. The spraying data includes n layers and n layer images corresponding to the n layers, namely the first layer image, the second layer image, ..., the nth layer image, where n is a natural number greater than or equal to 3.

[0136] The controller acquires the image of the mj-th layer and the image of the m+j-th layer, and generates the trajectory of the mj-th layer and the trajectory of the m+j-th layer based on the image of the mj-th layer and the image of the m+j-th layer, where m and j are both non-zero natural numbers, and n>m, m>j;

[0137] The controller acquires the pre-trained neural network model and inputs the trajectory of the mj-th layer and the trajectory of the m+j-th layer into the pre-trained neural network model. The neural network model is used to output the predicted trajectory of the intermediate layer between the two layers of the image based on the input trajectories of the two layers.

[0138] The controller obtains the trajectory of the m-th layer based on the output of the neural network training model, and controls the execution terminal to perform spraying on the m-th layer image based on the trajectory of the m-th layer.

[0139] In some implementations, the system is configured as follows:

[0140] The controller acquires the image of layer mj and layer m+j, and generates the trajectory of layer mj and layer m+j based on the image of layer mj and layer m+j, where m and j are non-zero natural numbers, and n>m, m>j, including:

[0141] The controller acquires the image of the mj-th layer and divides the image of the mj-th layer into multiple sub-images, which together form the image of the mj-th layer.

[0142] If the controller determines that at least one side of a sub-image is not adjacent to another sub-image, then the sub-image is determined to be the target sub-image;

[0143] The controller acquires all target sub-images, and uses the target sub-images as nodes and the lines between two adjacent target sub-images as paths to obtain multiple first Euler paths formed by the target sub-images.

[0144] The controller determines multiple target regions based on the positional relationship between multiple first Euler paths. Each target region is a region divided by the first Euler paths.

[0145] The controller determines the filled region among multiple target regions based on the image of the mj layer, and determines the trajectory of the mj layer based on the filled region.

[0146] In some implementations, the system is configured as follows:

[0147] The controller determines the filled regions among multiple target regions based on the image of layer mj, and determines the trajectory of layer mj based on the filled regions, including:

[0148] The controller acquires the sub-image in each filled region, and uses the sub-image as a node and the line connecting two adjacent sub-images as a path to obtain multiple second Euler paths formed by the sub-images.

[0149] The controller generates the trajectory of the mj-th layer based on multiple second Euler paths.

[0150] In some implementations, the system is configured as follows:

[0151] The controller generates the trajectory of the mj-th layer based on multiple second Euler paths, including:

[0152] Based on the shortest path principle, the controller connects multiple second Euler paths to form the trajectory of the mj layer.

[0153] In some implementations, the system is configured as follows:

[0154] The controller acquires a pre-trained neural network model and inputs the trajectory of the mj-th layer and the trajectory of the m+j-th layer into the pre-trained neural network model. The neural network model is used to output the predicted trajectory of the intermediate layer between the two input layer trajectories, including:

[0155] The controller acquires a training set, which includes k adjacent layer trajectories, where k is a natural number greater than or equal to 3;

[0156] The controller imports the training set into the neural network model and performs iterative calculations. When the iterative calculation process meets the preset conditions, the calculation stops and the trained neural network model is obtained.

[0157] In some implementations, the system is configured as follows:

[0158] The controller imports the training set into the neural network model and performs iterative calculations. When the iterative calculation process meets preset conditions, it stops calculating and obtains the trained neural network model, including:

[0159] The controller selects the (i-1)th layer trajectory, the ith layer trajectory, and the (i+1)th layer trajectory from the k-th layer trajectory, where i is a non-zero natural number, and i <k;

[0160] The controller uses the data of the (i-1)th layer trajectory and the data of the (i+1)th layer trajectory as two different dimension vectors to form the feature vector of the i-th layer trajectory data and performs iterative calculation.

[0161] The calculation stops when the iterative calculation process meets the preset conditions, and the controller obtains the trained neural network model.

[0162] In some implementations, the system is configured as follows:

[0163] The calculation stops when the iterative calculation process meets the preset conditions, and the trained neural network model is obtained, including:

[0164] The controller, based on the neural network model, obtains k-2 error values ​​according to the training process of the training set, and determines the average error value based on the k-2 error values;

[0165] The controller compares the average error value with the preset value. When the average error value is less than or equal to the preset value, it stops training and outputs the trained neural network model.

[0166] In some implementations, the system is configured as follows:

[0167] The controller, based on the neural network model, obtains k-2 error values ​​during the training process using the training set, and determines the average error value based on these k-2 error values, including:

[0168] The controller inputs the data of the (i-1)th layer trajectory and the data of the (i+1)th layer trajectory into the neural network training model, and obtains the predicted layer trajectory output by the neural network training model;

[0169] The controller determines an error value based on the data of the predicted layer trajectory output by the neural network training model and the data of the i-th layer trajectory.

[0170] In some implementations, the system is configured as follows:

[0171] The controller determines an error value based on the predicted layer trajectory data output by the neural network training model and the trajectory data of the i-th layer, satisfying the following:

[0172] If the data for the predicted layer trajectory is greater than the data for the i-th layer trajectory, then the error value is determined to be:

[0173] t = y(yx)

[0174] Where t is the error value, y is the data of the predicted layer trajectory, and x is the data of the i-th layer trajectory.

[0175] In some implementations, the system is configured as follows:

[0176] The controller determines an error value based on the predicted layer trajectory data output by the neural network training model and the trajectory data of the i-th layer, satisfying the following:

[0177] If the data for the predicted layer trajectory is less than the data for the i-th layer trajectory, then the error value is determined to be:

[0178] t = y(xy)

[0179] Where t is the error value, y is the data of the predicted layer trajectory, and x is the data of the i-th layer trajectory.

[0180] This application proposes an AI-powered intelligent 3D spraying control system for surface treatment of urban art installations. First, the controller acquires spraying data, which consists of image data to be sprayed. This data includes n layers and corresponding n layer images, namely, layer 1, layer 2, ..., layer n, where n is a natural number greater than or equal to 3. Then, the controller acquires the mj-th layer image and the (m+j-th)-th layer image, and generates the mj-th layer trajectory and the (m+j-th)-th layer trajectory based on these images, where m and j are non-zero natural numbers, and n > m and m > j. Next, the controller acquires a pre-trained neural network model and inputs the mj-th layer trajectory and the (m+j-th)-th layer trajectory into it. The neural network model outputs a predicted layer trajectory between the two input layer trajectories. Finally, the controller acquires the m-th layer trajectory based on the output of the neural network training model and controls the execution terminal to perform spraying on the m-th layer image based on this trajectory. This application proposes an AI-powered intelligent 3D spraying control system for surface treatment of urban art installations. Through a trained neural network model, the system can generate the trajectory of only some layers from images during multi-layer image spraying construction, while the trajectory of the remaining layers is generated using the neural network model. This reduces the computational effort required to directly generate trajectories from images, increases the total computation time, and further improves the efficiency of the spraying work.

[0181] Based on the same inventive concept, embodiments of this application also propose an electronic device, which includes:

[0182] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the AI ​​intelligent 3D spraying control method for surface treatment of urban art installations according to embodiments of this application.

[0183] Furthermore, to achieve the above objectives, embodiments of this application also propose a computer-readable storage medium storing a computer program that, when executed by a processor, implements the AI ​​intelligent 3D spraying control method for surface treatment of urban art installations according to embodiments of this application.

[0184] The following is a detailed introduction to the various components of the electronic device:

[0185] In this context, the processor is the control center of the electronic device. It can be a single processor or a collective term for multiple processing elements. For example, a processor can be one or more central processing units (CPUs), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention, such as one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs).

[0186] Alternatively, the processor can perform various functions of the electronic device by running or executing software programs stored in memory, and by calling data stored in memory.

[0187] The memory is used to store the software program that executes the solution of the present invention, and the execution is controlled by the processor. The specific implementation method can be referred to the above method embodiment, and will not be repeated here.

[0188] Optionally, the memory can be read-only memory (ROM) or other types of static storage devices capable of storing static information and instructions, random access memory (RAM) or other types of dynamic storage devices capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory can be integrated with the processor or exist independently and coupled to the processor through an interface circuit of an electronic device; the embodiments of the present invention do not specifically limit this.

[0189] A transceiver is used to communicate with network devices or with terminal devices.

[0190] Optionally, the transceiver may include a receiver and a transmitter. The receiver is used to implement the receiving function, and the transmitter is used to implement the sending function.

[0191] Optionally, the transceiver can be integrated with the processor or exist independently and coupled to the processor through the router's interface circuit. This embodiment of the invention does not specifically limit this.

[0192] Furthermore, the technical effects of the electronic device can be referred to the technical effects of the data transmission method in the above method embodiments, and will not be repeated here.

[0193] It should be understood that the processor in the embodiments of the present invention can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0194] It should also be understood that the memory in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0195] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the flow or function according to the embodiments of the present invention is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. A computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0196] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0197] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0198] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0199] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

Claims

1. An AI intelligent 3D spraying control system for urban art device surface treatment, characterized in that, The system comprises a controller and an execution terminal, and is configured to: The controller acquires spraying data, which is image data that needs to be sprayed, and the spraying data comprises n layers and n layer images corresponding to the n layers, which are respectively a first layer image, a second layer image,..., and an n-th layer image, wherein n is a natural number greater than or equal to 3; The controller acquires an m-j-th layer image and an m+j-th layer image, and generates an m-j-th layer track and an m+j-th layer track according to the m-j-th layer image and the m+j-th layer image, wherein m and j are both natural numbers other than 0, and n > m and m > j; The controller acquires a neural network model that has been trained, and inputs the m-j-th layer track and the m+j-th layer track into the neural network model that has been trained, and the neural network model is used to output a predicted layer track of an intermediate layer of two layers according to the input of the two layer tracks; The controller acquires an m-th layer track according to an output result of the neural network model, and controls the execution terminal to spray and construct an m-th layer image according to the m-th layer track; The system is configured to: The controller acquires an m-j-th layer image and an m+j-th layer image, and generates an m-j-th layer track and an m+j-th layer track according to the m-j-th layer image and the m+j-th layer image, wherein m and j are both natural numbers other than 0, and n > m and m > j, including: The controller acquires an m-j-th layer image, and divides the m-j-th layer image into a plurality of sub-images, and the plurality of sub-images form the m-j-th layer image; If the controller determines that at least one side of one of the sub-images is not adjacent to another of the sub-images, the controller determines that the one of the sub-images is a target sub-image; The controller acquires all of the target sub-images, takes the target sub-images as nodes, takes a connection between two adjacent target sub-images as a path, and acquires a plurality of first Euler paths formed by the target sub-images; The controller determines a plurality of target regions based on a positional relationship between the plurality of first Euler paths, and the target regions are regions formed by being separated by the first Euler paths as boundaries; The controller determines a filling region in the plurality of target regions based on the m-j-th layer image, and determines an m-j-th layer track according to the filling region; The system is configured to: The controller determines a filling region in the plurality of target regions based on the m-j-th layer image, and determines an m-j-th layer track according to the filling region, including: The controller acquires the sub-images in each of the filling regions, takes the sub-images as nodes, takes a connection between two adjacent sub-images as a path, and acquires a plurality of second Euler paths formed by the sub-images; The controller generates the m-j-th layer track based on the plurality of second Euler paths; The system is configured to: The controller generates the m-j-th layer track based on the plurality of second Euler paths, including: The controller connects multiple second Euler paths to form the m-jth layer trajectory based on a shortest path principle.

2. The AI intelligent 3D spraying control system for urban art device surface treatment according to claim 1, characterized in that, The system is configured to: The controller obtains a trained neural network model, and inputs the m-jth layer trajectory and the m+jth layer trajectory into the trained neural network model, and the neural network model is used to output a predicted layer trajectory of an intermediate layer of two layers of images according to the input two-layer layer trajectory, including: The controller obtains a training set, wherein the training set includes k adjacent layer trajectories, and k is a natural number greater than or equal to 3; The controller imports the training set into the neural network model and performs iterative calculation, stops calculation when the process of iterative calculation meets a preset condition, and obtains the trained neural network model.

3. The AI intelligent 3D spraying control system for urban art device surface treatment according to claim 2, characterized in that, The system is configured to: The controller imports the training set into the neural network model and performs iterative calculation, stops calculation when the process of iterative calculation meets a preset condition, and obtains the trained neural network model, including: The controller selects an i-1th layer trajectory, an ith layer trajectory and an i+1th layer trajectory from the k layer trajectories, wherein i is a natural number not equal to 0, and i < k; The controller takes the data of the i-1th layer trajectory and the data of the i+1th layer trajectory as two different dimension vectors, constructs a feature vector of the data of the ith layer trajectory, and performs iterative calculation; When the process of iterative calculation meets a preset condition, the controller obtains the trained neural network model.

4. The AI intelligent 3D spraying control system for urban art device surface treatment according to claim 3, characterized in that, The system is configured to: When the process of iterative calculation meets a preset condition, the controller stops calculation and obtains the trained neural network model, including: The controller obtains k-2 error values based on the process of training the neural network model according to the training set, and determines an average error value based on the k-2 error values; The controller compares the average error value with a preset value, and when the average error value is less than or equal to the preset value, the training is stopped and the trained neural network model is output.

5. The AI intelligent 3D spraying control system for urban art device surface treatment according to claim 3, characterized in that, The system is configured to: The controller obtains k-2 error values based on the process of training the neural network model according to the training set, and determines an average error value based on the k-2 error values, including: The controller inputs the data of the i-1th layer trajectory and the data of the i+1th layer trajectory into the neural network model, and obtains the predicted layer trajectory output by the neural network model; The controller determines an error value according to the data of the predicted layer trajectory output by the neural network model and the data of the ith layer trajectory.

6. The AI intelligent 3D spraying control system for urban art device surface treatment according to claim 5, characterized in that, The system is configured to: The controller determines an error value according to the data of the predicted layer trajectory output by the neural network model and the data of the ith layer trajectory, which satisfies: If the data of the predicted layer trajectory is greater than the data of the ith layer trajectory, the error value is determined as: t=y(y-x) Wherein t is the error value, y is the data of the predicted layer trajectory, and x is the data of the ith layer trajectory.

7. The AI intelligent 3D spraying control system for urban art device surface treatment according to claim 5, characterized in that, The system is configured to: The controller determines one error value according to the data of the predicted layer trajectory output by the neural network model and the data of the ith layer trajectory, which satisfies: If the data of the predicted layer trajectory is less than the data of the ith layer trajectory, the error value is determined as: t = y(x - y) Wherein, t is the error value, y is the data of the predicted layer trajectory, and x is the data of the ith layer trajectory.

Images