Aluminum plate carving transfer control system and method for interior decoration

Abstract

The present application relates to the field of intelligent control, and specifically discloses a control system and method for transferring aluminum plate carvings for interior decoration, which mines the correlation feature information between the surface state characteristics of the transferred substrate and the implicit features of the pattern to be transferred by adopting a neural network model based on deep learning, and adaptively adjusts the baking transfer temperature value based on the surface state of the transfer substrate and the transfer pattern, thereby ensuring the transfer quality of the aluminum plate carvings and the decorative effect of the aluminum plate carvings.

Description

Technical Field

[0001] The present application relates to the field of intelligent control, and more specifically, to a control system and method for transferring carved patterns on aluminum plates for interior decoration. Background Art

[0002] As people's living standards improve, the requirements for interior decoration are getting higher and higher. More and more people are beginning to pursue solid wood and natural stone decorative materials. With the development of the decoration industry, some natural precious wood and stone are becoming less and less. In order to prevent people from overusing them, the government has begun to vigorously protect these wood and stone. In order to make the surface of decorative materials have the effect of real wood grain and stone grain, people have invented a thermal transfer technology that can transfer various required patterns on the surface of wood.

[0003] Traditional thermal transfer technology usually transfers on the surface of wood. Since wood is not fireproof, easy to deform and not moisture-proof, this kind of wood decorative material has gradually failed to meet people's needs. People began to look for a decorative material that is both beautiful and fireproof, moisture-proof and deformation-proof. Aluminum plates have the characteristics of fireproof, moisture-proof and deformation-proof.

[0004] Currently, there are very few patterns that can be transferred on the surface of aluminum plates. Even after successful transfer on the aluminum plate, some even have color differences, incomplete patterns, and different shades of color, which seriously affect the decorative effect.

[0005] Therefore, an optimized control solution for the transfer of carved patterns on aluminum plates for interior decoration is desired. Summary of the invention

[0006] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a control system and method for transferring aluminum plate carvings for interior decoration, which mines the correlation feature information between the surface state characteristics of the transferred substrate and the implicit features of the pattern to be transferred by using a neural network model based on deep learning, and adaptively adjusts the baking transfer temperature value based on the surface state of the transfer substrate and the transfer pattern, thereby ensuring the transfer quality of the aluminum plate carvings and the decorative effect of the aluminum plate carvings.

[0007] According to one aspect of the present application, a control system for engraving and transferring aluminum plates for interior decoration is provided, which comprises:

[0008] A camera module, used to obtain a front detection image of the aluminum material and a front detection image of the thermal transfer paper;

[0009] An aluminum material surface feature extraction module, used for passing the front detection image of the aluminum material through a first convolutional neural network model as a feature extractor to obtain an aluminum material surface state feature matrix;

[0010] A first transfer pattern feature extraction module, used for obtaining a first-scale pattern feature matrix by using a second convolutional neural network model with a two-dimensional convolution kernel of a first scale on the front detection image of the thermal transfer paper;

[0011] A second transfer pattern feature extraction module, used for obtaining a second-scale pattern feature matrix by using a third convolutional neural network model with a two-dimensional convolution kernel of a second scale on the front detection image of the thermal transfer paper;

[0012] A multi-scale fusion module, used for fusing the first-scale pattern feature matrix and the second-scale pattern feature matrix to obtain a pattern feature matrix;

[0013] A decoding feature generation module, used to calculate a transfer matrix of the pattern feature matrix relative to the aluminum material surface state feature matrix as a decoding feature matrix;

[0014] A decoding feature distribution optimization module, used for optimizing the feature convergence of the decoding feature matrix based on the pattern feature matrix and the aluminum material surface state feature matrix to obtain an optimized decoding feature matrix; and

[0015] The recommended control result generating module is used to pass the optimized decoding feature matrix through a decoder to obtain a decoding value, and the decoding value is used to represent the recommended baking temperature value.

[0016] In the above-mentioned aluminum plate carving transfer control system for interior decoration, the aluminum surface feature extraction module is further used to: use the layers of the first convolutional neural network model as a feature extractor to perform the following on the input data in the forward transfer of the layer: convolution processing on the input data to obtain a convolution feature map; pooling the convolution feature map along the channel dimension to obtain a pooled feature map; and, nonlinear activation on the pooled feature map to obtain an activation feature map; wherein the output of the last layer of the first convolutional neural network as a feature extractor is the aluminum surface state feature matrix, and the input of the first layer of the first convolutional neural network as a feature extractor is the front detection image of the aluminum.

[0017] In the above-mentioned aluminum plate carved transfer control system for interior decoration, the first transfer pattern feature extraction module is further used to: use the layers of the second convolutional neural network model using the two-dimensional convolution kernel with the first scale to respectively perform the following on the input data in the forward transfer of the layer: convolution processing on the input data to obtain a convolution feature map; pooling the convolution feature map along the channel dimension to obtain a pooled feature map; and, nonlinear activation on the pooled feature map to obtain an activation feature map; wherein the output of the last layer of the second convolutional neural network using the two-dimensional convolution kernel with the first scale is the first-scale pattern feature matrix, and the input of the first layer of the second convolutional neural network using the two-dimensional convolution kernel with the first scale is the front detection image of the thermal transfer paper.

[0018] In the above-mentioned aluminum plate carved transfer control system for interior decoration, the second transfer pattern feature extraction module is further used to: use the layers of the third convolutional neural network model using the two-dimensional convolution kernel with the second scale to perform the following on the input data in the forward transfer of the layer: convolution processing on the input data to obtain a convolution feature map; pooling the convolution feature map along the channel dimension to obtain a pooled feature map; and, nonlinear activation on the pooled feature map to obtain an activation feature map; wherein the output of the last layer of the third convolutional neural network using the two-dimensional convolution kernel with the second scale is the second-scale pattern feature matrix, and the input of the first layer of the third convolutional neural network using the two-dimensional convolution kernel with the second scale is the front detection image of the thermal transfer paper.

[0019] In the above-mentioned aluminum plate carving transfer control system for interior decoration, the multi-scale fusion module is further used to: fuse the first-scale pattern feature matrix and the second-scale pattern feature matrix according to the following formula to obtain a pattern feature matrix; wherein the formula is:

[0020]

[0021] Wherein M1 represents the first scale pattern feature matrix, M2 represents the second scale pattern feature matrix, M represents the pattern feature matrix, a1 and a2 represent weighting parameters of the first scale pattern feature matrix and the second scale pattern feature matrix respectively, Indicates sum by position.

[0022] In the above-mentioned aluminum plate carving transfer control system for interior decoration, the decoding feature generation module is further used to: calculate the transfer matrix of the pattern feature matrix relative to the aluminum surface state feature matrix as the decoding feature matrix using the following formula; wherein the formula is:

[0023]

[0024] Where M represents the pattern feature matrix, M c represents the aluminum surface state characteristic matrix, M n represents the decoding feature matrix.

[0025] In the above-mentioned aluminum plate carving transfer control system for interior decoration, the decoding feature distribution optimization module includes: a feature matrix expansion unit, which is used to expand the pattern feature matrix and the aluminum surface state feature matrix into a pattern feature vector and an aluminum surface state feature vector; a Hilbert space constraint unit, which is used to perform vector modulus Hilbert space constraints on the pattern feature vector and the aluminum surface state feature vector to obtain a weight feature vector; a vector reconstruction unit, which is used to reconstruct the weight feature vector into a weight feature matrix; and an action unit, which is used to calculate the position point multiplication between the weight feature matrix and the decoding feature matrix to obtain the optimized decoding feature matrix.

[0026] In the above-mentioned aluminum plate carving transfer control system for interior decoration, the Hilbert space constraint unit is further used to: perform vector modulus Hilbert space constraints on the pattern feature vector and the aluminum surface state feature vector to obtain the weight feature vector according to the following formula; wherein the formula is:

[0027]

[0028] Wherein V1 and V2 represent the pattern feature vector and the aluminum surface state feature vector respectively, ||·||2 represents the second norm of the vector, represents the positional addition of the vector, α and β are weighted hyperparameters, Cov 1D Represents the convolution operator (||V1||2,||V2||2,V1V2 T ) for vector Perform one-dimensional convolution, V w represents the weight feature vector.

[0029] In the above-mentioned aluminum plate carving transfer control system for interior decoration, the recommended control result generation module includes: using the decoder to perform decoding regression on the optimized decoding feature matrix according to the following formula to obtain a decoding value for representing the recommended baking temperature value; wherein the formula is: Where X represents the optimized decoding feature matrix, Y is the decoding value, and W is the weight matrix. Represents matrix multiplication.

[0030] According to another aspect of the present application, a method for controlling the transfer of carved patterns on an aluminum plate for interior decoration is provided, which comprises:

[0031] Acquire a front detection image of the aluminum material and a front detection image of the thermal transfer paper;

[0032] Passing the front detection image of the aluminum material through a first convolutional neural network model as a feature extractor to obtain a surface state feature matrix of the aluminum material;

[0033] The front detection image of the thermal transfer paper is subjected to a second convolutional neural network model having a two-dimensional convolution kernel of a first scale to obtain a first-scale pattern feature matrix;

[0034] The front detection image of the thermal transfer paper is subjected to a third convolutional neural network model having a two-dimensional convolution kernel of a second scale to obtain a second-scale pattern feature matrix;

[0035] fusing the first-scale pattern feature matrix and the second-scale pattern feature matrix to obtain a pattern feature matrix;

[0036] Calculating a transfer matrix of the pattern feature matrix relative to the aluminum material surface state feature matrix as a decoding feature matrix;

[0037] Based on the pattern feature matrix and the aluminum material surface state feature matrix, optimizing the feature convergence of the decoding feature matrix to obtain an optimized decoding feature matrix; and

[0038] The optimized decoding feature matrix is ​​passed through a decoder to obtain a decoding value, and the decoding value is used to represent a baking recommended temperature value.

[0039] According to another aspect of the present application, an electronic device is provided, comprising: a processor; and a memory, wherein computer program instructions are stored in the memory, and when the computer program instructions are executed by the processor, the processor executes the above-mentioned method for controlling the transfer of carved patterns on aluminum plates for interior decoration.

[0040] According to another aspect of the present application, a computer-readable medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes the above-mentioned method for controlling the transfer of carved patterns on aluminum plates for interior decoration.

[0041] Compared with the prior art, the present application provides a control system and method for transferring aluminum plate carvings for interior decoration, which uses a neural network model based on deep learning to mine the correlation feature information between the surface state characteristics of the transferred substrate and the implicit features of the pattern to be transferred, and adaptively adjusts the baking transfer temperature value based on the surface state of the transfer substrate and the transfer pattern, thereby ensuring the transfer quality of the aluminum plate carvings and the decorative effect of the aluminum plate carvings. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0043] Figure 1 This is an application scenario diagram of an aluminum plate carving transfer control system for interior decoration according to an embodiment of the present application;

[0044] Figure 2 It is a block diagram of a control system for carving and transferring of aluminum plates for interior decoration according to an embodiment of the present application;

[0045] Figure 3 This is a system architecture diagram of an interior decoration aluminum plate carving transfer control system according to an embodiment of the present application;

[0046] Figure 4 A flowchart of the first convolutional neural network encoding in the control system for carving and transfer printing of aluminum plates for interior decoration according to an embodiment of the present application;

[0047] Figure 5 It is a block diagram of a decoding feature distribution optimization module in a control system for interior decoration aluminum plate carving transfer according to an embodiment of the present application;

[0048] Figure 6 It is a flow chart of a method for controlling the transfer of carved patterns on an aluminum plate for interior decoration according to an embodiment of the present application;

[0049] Figure 7 is a block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0050] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0051] Application Overview

[0052] As mentioned in the previous background technology, aluminum plates have the characteristics of fire resistance, moisture resistance and deformation resistance. At present, there are very few patterns that can be transferred on the surface of aluminum plates. Even after successful transfer on the aluminum plate, some even have color difference, incomplete patterns, different color depths, etc., which seriously affect the decorative effect. Therefore, an optimized control solution for the transfer of carved patterns on aluminum plates for interior decoration is expected.

[0053] Accordingly, considering that after the actual successful transfer on the aluminum plate, there may be color difference, incomplete pattern, different color depth and other phenomena, which seriously affect the decorative effect. This is because the existing scheme is mostly based on a fixed oven temperature to carry out the baking process when the wrapped aluminum material is sent to the oven for baking. However, due to the different states of the surfaces of each aluminum material and the different patterns to be transferred, this will lead to the use of a fixed baking temperature for baking, which will reduce the quality of the transfer and thus reduce the decorative effect. Based on this, in the technical solution of the present application, in the actual transfer process of the aluminum plate surface, after the wrapped aluminum material is sent to the oven for baking, the adjustment of the high and low changes in the baking temperature should be based on the surface state characteristics of the transferred substrate and the implicit features such as the texture of the pattern to be transferred to perform adaptive control. In this process, the difficulty lies in how to dig out the correlation feature information between the surface state characteristics of the transferred substrate and the implicit features of the pattern to be transferred, so as to perform adaptive control of the baking transfer temperature, thereby ensuring the transfer quality of the aluminum plate carving and the decorative effect of the aluminum plate carving.

[0054] In recent years, deep learning and neural networks have been widely used in computer vision, natural language processing, text signal processing and other fields. In addition, deep learning and neural networks have also demonstrated a level close to or even beyond that of humans in image classification, object detection, semantic segmentation, text translation and other fields.

[0055] The development of deep learning and neural networks provides new solutions and schemes for mining the correlation feature information between the surface state characteristics of the transferred substrate and the implicit features of the pattern to be transferred. Those skilled in the art should know that the deep neural network model based on deep learning can adjust the parameters of the deep neural network model through appropriate training strategies, such as the back propagation algorithm of gradient descent, so that it can simulate the complex nonlinear associations between things, which is obviously suitable for simulating and mining the correlation feature information between the surface state characteristics of the transferred substrate and the implicit features of the pattern to be transferred.

[0056] Specifically, in the technical solution of the present application, first, a front detection image of the aluminum material and a front detection image of the thermal transfer paper are obtained. It is worth mentioning that here, the aluminum material is the substrate to be transferred, and the aluminum material is intact and without deformation. Then, a first convolutional neural network model as a feature extractor with excellent performance in extracting local implicit correlation features of images is used to perform feature mining of the front detection image of the aluminum material, so as to extract high-dimensional implicit feature information about the surface state of the aluminum material in the front detection image of the aluminum material, thereby obtaining a surface state feature matrix of the aluminum material.

[0057] Then, considering that when extracting the implicit features of the transfer pattern, it is found that there are feature distribution information of different scales in the front detection image of the thermal transfer paper, therefore, if the baking temperature value is to be accurately adaptively controlled, it is necessary to fully and accurately mine the implicit features in the front detection image of the thermal transfer paper. It should be understood that, considering that in the deep learning network, the texture features of the transfer pattern extracted by feature encoders of different depths carry detail information and semantic information of different proportions. For example, the feature resolution extracted by the shallow encoder is high, and the details in the front detection image of the thermal transfer paper are more retained and the detail information is rich; while the feature resolution of the deep encoder is low, the features are more abstract, but contain more semantic information.

[0058] Based on this, in the technical solution of the present application, the front detection image of the thermal transfer paper is encoded through a convolutional neural network model with two-dimensional convolution kernels of different scales to extract multi-scale fusion features of semantics and detail information of the transfer pattern of the thermal transfer paper in the front detection image of the thermal transfer paper, thereby obtaining a pattern feature matrix. Specifically, first, the front detection image of the thermal transfer paper is processed by a second convolutional neural network model with a two-dimensional convolution kernel of a first scale to extract the first scale implicit feature distribution information of the transfer pattern in the front detection image of the thermal transfer paper, thereby obtaining a first scale pattern feature matrix. Then, the front detection image of the thermal transfer paper is processed by a third convolutional neural network model with a two-dimensional convolution kernel of a second scale to extract the second scale implicit feature distribution information of the transfer pattern in the front detection image of the thermal transfer paper, thereby obtaining a second scale pattern feature matrix. Then, the first scale pattern feature matrix and the second scale pattern feature matrix are fused to obtain a pattern feature matrix with multi-scale fusion features of the transfer pattern.

[0059] Furthermore, the transfer matrix of the pattern feature matrix relative to the aluminum surface state feature matrix is ​​calculated to represent the high-dimensional implicit features of the aluminum surface state and the correlation feature distribution information of the multi-scale correlation features of the transfer pattern, and this is used as a decoding feature matrix for decoding processing in a decoder to obtain a decoding value for representing the recommended baking temperature value. In this way, the baking transfer temperature value can be adaptively adjusted based on the surface state of the transfer substrate and the transfer pattern to ensure the transfer quality of the aluminum plate carving and the decorative effect of the aluminum plate carving.

[0060] In particular, in the technical solution of the present application, considering that the aluminum surface state feature matrix is ​​obtained by directly extracting features from the aluminum image through the first convolutional neural network model as a feature extractor, and the pattern feature matrix is ​​obtained by fusing the features of thermal transfer paper images of different scales position by position based on the feature matrix, its feature distribution may not be consistent in the convergence direction corresponding to the spatial distribution of image semantics, which leads to poor convergence of the overall feature distribution of the transfer matrix of the pattern feature matrix relative to the aluminum surface state feature matrix, and poor fitting effect of decoding it as the decoding feature matrix through the decoder. On the other hand, if a threshold is directly set for the pattern feature matrix and the aluminum surface state feature matrix to fit their convergence direction, it may lead to a high correlation between the eigenvalues ​​of the decoding feature matrix as the transfer matrix, thereby reducing the decoding accuracy of the decoding feature matrix.

[0061] Therefore, the pattern feature matrix and the aluminum surface state feature matrix are first expanded into a pattern feature vector, for example, denoted as V1 and an aluminum surface state feature vector, for example, denoted as V2, and then the Hilbert space constraint of the vector modulus basis is performed to obtain a weight feature vector, for example, denoted as V w , expressed as:

[0062]

[0063] Cov 1D Represents a one-dimensional convolution operation, that is, the convolution operator (||V1||2,||V2||2,V1V2 T ) for vector Perform a one-dimensional convolution.

[0064] Here, the weighted feature vector V is convolved by a convolution operator in a Hilbert space that defines the inner product of the vector sum and the vector modulus w To constrain, the weight feature vector V w The characteristic distribution of is limited to a finite closed domain in the Hilbert space based on the modulus of the vector, and the weighted characteristic vector V is improved. w The orthogonality between the basis dimensions of the high-dimensional manifold of the feature distribution is achieved, thereby achieving sparse correlation between eigenvalues ​​while maintaining the overall convergence of the feature distribution. Then, the weighted feature vector V w By restoring it to a feature matrix form and performing dot multiplication with the decoding feature matrix, the fitting effect of the optimized decoding feature matrix through the decoder and the accuracy of the decoding result can be improved. In this way, the baking transfer temperature value can be accurately and adaptively adjusted based on the surface state of the transfer substrate and the transfer pattern to ensure the transfer quality and decorative effect of the aluminum plate carving.

[0065] Based on this, the present application proposes a control system for carved transfer of aluminum plates for interior decoration, which includes: a camera module, used to obtain a front detection image of an aluminum material and a front detection image of a thermal transfer paper; an aluminum surface feature extraction module, used to pass the front detection image of the aluminum material through a first convolutional neural network model as a feature extractor to obtain an aluminum surface state feature matrix; a first transfer pattern feature extraction module, used to pass the front detection image of the thermal transfer paper through a second convolutional neural network model with a first-scale two-dimensional convolution kernel to obtain a first-scale pattern feature matrix; a second transfer pattern feature extraction module, used to pass the front detection image of the thermal transfer paper through a third convolutional neural network model with a second-scale two-dimensional convolution kernel A multi-scale neural network model is used to obtain a second-scale pattern feature matrix; a multi-scale fusion module is used to fuse the first-scale pattern feature matrix and the second-scale pattern feature matrix to obtain a pattern feature matrix; a decoding feature generation module is used to calculate the transfer matrix of the pattern feature matrix relative to the aluminum surface state feature matrix as a decoding feature matrix; a decoding feature distribution optimization module is used to optimize the feature convergence of the decoding feature matrix based on the pattern feature matrix and the aluminum surface state feature matrix to obtain an optimized decoding feature matrix; and a recommended control result generation module is used to pass the optimized decoding feature matrix through a decoder to obtain a decoding value, and the decoding value is used to represent a recommended baking temperature value.

[0066] Figure 1 This is an application scenario diagram of the indoor decoration aluminum plate carving transfer control system according to the embodiment of the present application. Figure 1 As shown, in this application scenario, through a camera (for example, Figure 1 C) as shown in the figure, obtains the front detection image of the aluminum material (for example, Figure 1 F1 shown in FIG. 1 ) and a front detection image of the thermal transfer paper (e.g., Figure 1 Then, the above image is input to a server (for example, Figure 1 In S), the server is capable of processing the input image using the interior decoration aluminum plate carving transfer control algorithm to generate a decoded value, which is used to represent the recommended baking temperature value.

[0067] After introducing the basic principles of the present application, various non-limiting embodiments of the present application will be described in detail with reference to the accompanying drawings.

[0068] Exemplary Systems

[0069] Figure 2 FIG. 1 is a block diagram of a control system for interior decoration aluminum plate carving transfer according to an embodiment of the present application. Figure 2As shown, according to the embodiment of the present application, the control system 300 for the carved transfer of aluminum plates for interior decoration includes: a camera module 310; an aluminum surface feature extraction module 320; a first transfer pattern feature extraction module 330; a second transfer pattern feature extraction module 340; a multi-scale fusion module 350; a decoding feature generation module 360; a decoding feature distribution optimization module 370; and a recommended control result generation module 380.

[0070] Among them, the camera module 310 is used to obtain the front detection image of the aluminum material and the front detection image of the thermal transfer paper; the aluminum surface feature extraction module 320 is used to obtain the aluminum surface state feature matrix by passing the front detection image of the aluminum material through the first convolutional neural network model as a feature extractor; the first transfer pattern feature extraction module 330 is used to obtain the first scale pattern feature matrix by passing the front detection image of the thermal transfer paper through the second convolutional neural network model with a first scale two-dimensional convolution kernel; the second transfer pattern feature extraction module 340 is used to obtain the second scale pattern feature matrix by passing the front detection image of the thermal transfer paper through the third convolutional neural network model with a second scale two-dimensional convolution kernel. The multi-scale fusion module 350 is used to fuse the first-scale pattern feature matrix and the second-scale pattern feature matrix to obtain a pattern feature matrix; the decoding feature generation module 360 ​​is used to calculate the transfer matrix of the pattern feature matrix relative to the aluminum surface state feature matrix as a decoding feature matrix; the decoding feature distribution optimization module 370 is used to optimize the feature convergence of the decoding feature matrix based on the pattern feature matrix and the aluminum surface state feature matrix to obtain an optimized decoding feature matrix; and the recommended control result generation module 380 is used to pass the optimized decoding feature matrix through a decoder to obtain a decoding value, and the decoding value is used to represent the recommended baking temperature value.

[0071] Figure 3 1 is a system architecture diagram of a control system for interior decoration aluminum plate carving transfer according to an embodiment of the present application. Figure 3As shown, in the network architecture, the front detection image of the aluminum material and the front detection image of the thermal transfer paper are first acquired by the camera module 310; then, the aluminum surface feature extraction module 320 passes the front detection image of the aluminum material acquired by the camera module 310 through the first convolutional neural network model as a feature extractor to obtain the aluminum surface state feature matrix; the first transfer pattern feature extraction module 330 passes the front detection image of the thermal transfer paper acquired by the camera module 310 through the second convolutional neural network model with a first-scale two-dimensional convolution kernel to obtain a first-scale pattern feature matrix; the second transfer pattern feature extraction module 340 passes the front detection image of the thermal transfer paper acquired by the camera module 310 through the third convolutional neural network model with a second-scale two-dimensional convolution kernel to obtain a second-scale pattern feature matrix; then, the multi- The scale fusion module 350 fuses the first scale pattern feature matrix obtained by the first transfer pattern feature extraction module 330 and the second scale pattern feature matrix obtained by the second transfer pattern feature extraction module 340 to obtain a pattern feature matrix; the decoding feature generation module 360 ​​calculates the transfer matrix of the pattern feature matrix fused by the multi-scale fusion module 350 relative to the aluminum surface state feature matrix obtained by the aluminum surface feature extraction module 320 as a decoding feature matrix; the decoding feature distribution optimization module 370 optimizes the feature convergence of the decoding feature matrix based on the pattern feature matrix and the aluminum surface state feature matrix to obtain an optimized decoding feature matrix; further, the recommended control result generation module 380 passes the optimized decoding feature matrix through a decoder to obtain a decoding value, and the decoding value is used to represent the recommended baking temperature value.

[0072] Specifically, during the operation of the control system 300 for transferring carved aluminum plates for interior decoration, the camera module 310 is used to obtain the front detection image of the aluminum material and the front detection image of the thermal transfer paper. In the technical solution of the present application, during the actual transfer process on the surface of the aluminum plate, after the wrapped aluminum material is sent to the oven for baking, the adjustment of the high and low changes in the baking temperature should be adaptively controlled based on the surface state characteristics of the transferred substrate and the implicit characteristics such as the texture of the pattern to be transferred. Therefore, first of all, the front detection image of the aluminum material and the front detection image of the thermal transfer paper can be obtained by a camera. It is worth mentioning that here, the aluminum material is the substrate to be transferred, and the aluminum material is intact and without deformation.

[0073] Specifically, during the operation of the control system 300 for interior decoration aluminum plate carving transfer, the aluminum surface feature extraction module 320 is used to pass the front detection image of the aluminum through the first convolutional neural network model as a feature extractor to obtain the aluminum surface state feature matrix. That is, the first convolutional neural network model as a feature extractor with excellent performance in extracting local implicit correlation features of the image is used to perform feature mining of the front detection image of the aluminum, so as to extract high-dimensional implicit feature information about the surface state of the aluminum in the front detection image of the aluminum, thereby obtaining the aluminum surface state feature matrix. In a specific example, the first convolutional neural network includes a plurality of neural network layers cascaded to each other, wherein each neural network layer includes a convolution layer, a pooling layer, and an activation layer. In the encoding process of the first convolutional neural network, each layer of the first convolutional neural network uses the convolution layer to perform convolution processing based on the convolution kernel on the input data during the forward transmission process of the layer, uses the pooling layer to perform pooling processing on the convolution feature map output by the convolution layer, and uses the activation layer to perform activation processing on the pooling feature map output by the pooling layer, wherein the input data of the first layer of the first convolutional neural network as a feature extractor is the front detection image of the aluminum material.

[0074] Figure 4 The flowchart of the first convolutional neural network encoding in the control system for the transfer of carved patterns on aluminum plates for interior decoration according to the embodiment of the present application is shown in FIG. Figure 4 As shown, in the encoding process of the first convolutional neural network, it includes: using each layer of the first convolutional neural network model as a feature extractor to perform the following on the input data in the forward transfer of the layer: S210, convolution processing on the input data to obtain a convolution feature map; S220, pooling the convolution feature map along the channel dimension to obtain a pooled feature map; and, S230, nonlinear activation on the pooled feature map to obtain an activation feature map; wherein the output of the last layer of the first convolutional neural network as a feature extractor is the surface state feature matrix of the aluminum material, and the input of the first layer of the first convolutional neural network as a feature extractor is the front detection image of the aluminum material.

[0075] Specifically, during the operation of the interior decoration aluminum plate carved transfer control system 300, the first transfer pattern feature extraction module 330 and the second transfer pattern feature extraction module 340 are used to obtain a first scale pattern feature matrix by using a second convolutional neural network model with a two-dimensional convolution kernel of a first scale for the front detection image of the thermal transfer paper; and, the front detection image of the thermal transfer paper is obtained by using a third convolutional neural network model with a two-dimensional convolution kernel of a second scale to obtain a second scale pattern feature matrix. Considering that when the implicit features of the transfer pattern are extracted, it is found that there are feature distribution information of different scales in the front detection image of the thermal transfer paper. Therefore, if the baking temperature value is to be accurately adaptively controlled, it is necessary to fully and accurately mine the implicit features in the front detection image of the thermal transfer paper. It should be understood that considering that in the deep learning network, the texture features of the transfer pattern extracted by the feature encoders of different depths carry different proportions of detail information and semantic information. For example, the feature resolution extracted by the shallow encoder is high, and the details in the front detection image of the thermal transfer paper are more retained and the detail information is rich; while the feature resolution of the deep encoder is low, the features are more abstract, but contain more semantic information. Based on this, in the technical solution of the present application, the front detection image of the thermal transfer paper is encoded in a convolutional neural network model with two-dimensional convolution kernels of different scales to extract multi-scale fusion features of the transfer pattern of the thermal transfer paper in the front detection image of the thermal transfer paper, which have both semantics and detail information, thereby obtaining a pattern feature matrix. Specifically, first, the front detection image of the thermal transfer paper is processed in a second convolutional neural network model using a two-dimensional convolution kernel with a first scale to extract the first-scale implicit feature distribution information about the transfer pattern in the front detection image of the thermal transfer paper, thereby obtaining a first-scale pattern feature matrix. More specifically, the front detection image of the thermal transfer paper is passed through a second convolutional neural network model using a two-dimensional convolution kernel with a first scale to obtain a first-scale pattern feature matrix, including: using each layer of the second convolutional neural network model using the two-dimensional convolution kernel with the first scale to respectively perform the following on the input data in the forward pass of the layer: convolution processing on the input data to obtain a convolution feature map; pooling the convolution feature map along the channel dimension to obtain a pooled feature map; and nonlinear activation on the pooled feature map to obtain an activation feature map; wherein the output of the last layer of the second convolutional neural network using the two-dimensional convolution kernel with the first scale is the first-scale pattern feature matrix, and the input of the first layer of the second convolutional neural network using the two-dimensional convolution kernel with the first scale is the front detection image of the thermal transfer paper.Next, the front detection image of the thermal transfer paper is subjected to a third convolutional neural network model with a second-scale two-dimensional convolution kernel to extract the second-scale implicit feature distribution information about the transfer pattern in the front detection image of the thermal transfer paper, thereby obtaining a second-scale pattern feature matrix. Then, the first-scale pattern feature matrix and the second-scale pattern feature matrix are fused to obtain a pattern feature matrix having multi-scale fusion features of the transfer pattern. More specifically, the method of obtaining a second-scale pattern feature matrix by using a third convolutional neural network model with a two-dimensional convolution kernel of a second scale for the front detection image of the thermal transfer paper includes: using each layer of the third convolutional neural network model with a two-dimensional convolution kernel of a second scale to respectively perform the following on the input data in the forward pass of the layer: convolution processing on the input data to obtain a convolution feature map; pooling the convolution feature map along the channel dimension to obtain a pooled feature map; and nonlinearly activating the pooled feature map to obtain an activation feature map; wherein the output of the last layer of the third convolutional neural network using the two-dimensional convolution kernel of the second scale is the second-scale pattern feature matrix, and the input of the first layer of the third convolutional neural network using the two-dimensional convolution kernel of the second scale is the front detection image of the thermal transfer paper.

[0076] Specifically, during the operation of the interior decoration aluminum plate carving transfer control system 300, the multi-scale fusion module 350 is used to fuse the first-scale pattern feature matrix and the second-scale pattern feature matrix to obtain a pattern feature matrix. In the technical solution of the present application, the first-scale pattern feature matrix and the second-scale pattern feature matrix can be fused by weighted sum according to position. More specifically, the first-scale pattern feature matrix and the second-scale pattern feature matrix are fused by the following formula to obtain a pattern feature matrix; wherein the formula is:

[0077]

[0078] Wherein M1 represents the first scale pattern feature matrix, M2 represents the second scale pattern feature matrix, M represents the pattern feature matrix, a1 and a2 represent weighting parameters of the first scale pattern feature matrix and the second scale pattern feature matrix respectively, Indicates sum by position.

[0079] Specifically, during the operation of the control system 300 for transfer of carved aluminum plates for interior decoration, the decoding feature generation module 360 ​​is used to calculate the transfer matrix of the pattern feature matrix relative to the aluminum surface state feature matrix as a decoding feature matrix. In the technical solution of the present application, the transfer matrix of the pattern feature matrix relative to the aluminum surface state feature matrix is ​​calculated to represent the high-dimensional implicit features of the aluminum surface state and the correlation feature distribution information of the multi-scale correlation features of the transfer pattern, and this is used as a decoding feature matrix for decoding processing in the decoder to obtain a decoding value for representing the recommended baking temperature value. In this way, the baking transfer temperature value can be adaptively regulated based on the surface state of the transfer substrate and the transfer pattern to ensure the transfer quality of the aluminum plate carving and the decorative effect of the aluminum plate carving. Specifically, the calculation of the transfer matrix of the pattern feature matrix relative to the aluminum surface state feature matrix as a decoding feature matrix includes: calculating the transfer matrix of the pattern feature matrix relative to the aluminum surface state feature matrix as a decoding feature matrix using the following formula; wherein the formula is: Where M represents the pattern feature matrix, M c represents the aluminum surface state characteristic matrix, M n represents the decoding feature matrix.

[0080] Specifically, during the operation of the control system 300 for interior decoration aluminum plate carving transfer, the decoding feature distribution optimization module 370 is used to optimize the feature convergence of the decoding feature matrix based on the pattern feature matrix and the aluminum surface state feature matrix to obtain an optimized decoding feature matrix. In particular, in the technical solution of the present application, considering that the aluminum surface state feature matrix is ​​obtained by directly extracting features from the aluminum image through the first convolutional neural network model as a feature extractor, and the pattern feature matrix is ​​obtained by fusing the thermal transfer paper image features of different scales position by position based on the feature matrix, its feature distribution may not be consistent in the convergence direction corresponding to the spatial distribution of image semantics, which leads to the poor convergence of the overall feature distribution of the transfer matrix of the pattern feature matrix relative to the aluminum surface state feature matrix, and the poor fitting effect of using it as the decoding feature matrix for decoding by the decoder. On the other hand, if the threshold is directly set for the pattern feature matrix and the aluminum surface state feature matrix to fit their convergence direction, it may also lead to a high correlation between the eigenvalues ​​of the decoding feature matrix as the transfer matrix, thereby reducing the decoding accuracy of the decoding feature matrix. Therefore, the pattern feature matrix and the aluminum surface state feature matrix are first expanded into a pattern feature vector, for example, denoted as V1 and an aluminum surface state feature vector, for example, denoted as V2, and then the Hilbert space constraint of the vector modulus basis is performed to obtain a weight feature vector, for example, denoted as Vw , expressed as:

[0081]

[0082] Wherein V1 and V2 represent the pattern feature vector and the aluminum surface state feature vector respectively, ||·||2 represents the second norm of the vector, represents the positional addition of the vector, α and β are weighted hyperparameters, Cov 1D Represents the convolution operator (||V1||2,||V2||2,V1V2 T ) for vector Perform one-dimensional convolution, V w Here, the weight feature vector V is convolved by a convolution operator in a Hilbert space that defines the inner product of the vector and the modulus with the vector. w To constrain, the weight feature vector V w The characteristic distribution of is limited to a finite closed domain in the Hilbert space based on the modulus of the vector, and the weighted characteristic vector V is improved. w The orthogonality between the basis dimensions of the high-dimensional manifold of the feature distribution is achieved, thereby achieving sparse correlation between eigenvalues ​​while maintaining the overall convergence of the feature distribution. Then, the weighted feature vector V w By restoring it to a feature matrix form and performing dot multiplication with the decoding feature matrix, the fitting effect of the optimized decoding feature matrix through the decoder and the accuracy of the decoding result can be improved. In this way, the baking transfer temperature value can be accurately and adaptively adjusted based on the surface state of the transfer substrate and the transfer pattern to ensure the transfer quality and decorative effect of the aluminum plate carving.

[0083] Figure 5 FIG. 1 is a block diagram of a decoding feature distribution optimization module in a control system for interior decoration aluminum plate carving transfer printing according to an embodiment of the present application. Figure 5 As shown, the decoding feature distribution optimization module 370 includes: a feature matrix expansion unit 371, used to expand the pattern feature matrix and the aluminum surface state feature matrix into a pattern feature vector and an aluminum surface state feature vector; a Hilbert space constraint unit 372, used to perform vector modulus Hilbert space constraints on the pattern feature vector and the aluminum surface state feature vector to obtain a weight feature vector; a vector reconstruction unit 373, used to reconstruct the weight feature vector into a weight feature matrix; an action unit 374, used to calculate the position point multiplication between the weight feature matrix and the decoding feature matrix to obtain the optimized decoding feature matrix.

[0084] Specifically, during the operation of the control system 300 for transferring the carved aluminum plate for interior decoration, the recommended control result generation module 380 is used to pass the optimized decoding feature matrix through a decoder to obtain a decoding value, and the decoding value is used to represent the recommended baking temperature value. In this way, the baking transfer temperature value can be accurately and adaptively regulated based on the surface state of the transfer substrate and the transfer pattern to ensure the transfer quality of the aluminum plate carving and the decorative effect of the aluminum plate carving. In a specific example of the present application, the decoder is used to decode and regress the optimized decoding feature matrix using the following formula to obtain a decoding value used to represent the recommended baking temperature value; wherein the formula is: Where X represents the optimized decoding feature matrix, Y is the decoding value, and W is the weight matrix. Represents matrix multiplication.

[0085] In summary, according to the embodiment of the present application, the control system 300 for interior decoration of aluminum plate carving is explained, which mines the correlation feature information between the surface state characteristics of the transferred substrate and the implicit features of the pattern to be transferred by adopting a neural network model based on deep learning, and adaptively adjusts the baking transfer temperature value based on the surface state of the transfer substrate and the transfer pattern, thereby ensuring the transfer quality of the aluminum plate carving and the decorative effect of the aluminum plate carving.

[0086] As described above, the interior decoration aluminum plate carving transfer control system according to the embodiment of the present application can be implemented in various terminal devices. In one example, the interior decoration aluminum plate carving transfer control system 300 according to the embodiment of the present application can be integrated into the terminal device as a software module and / or a hardware module. For example, the interior decoration aluminum plate carving transfer control system 300 can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the interior decoration aluminum plate carving transfer control system 300 can also be one of the many hardware modules of the terminal device.

[0087] Alternatively, in another example, the interior decoration aluminum plate carved transfer control system 300 and the terminal device may also be separate devices, and the interior decoration aluminum plate carved transfer control system 300 may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0088] Exemplary Methods

[0089] Figure 6 Flow chart of the control method of the interior decoration aluminum plate carving transfer according to the embodiment of the present application. Figure 6As shown, according to the embodiment of the present application, the control method for carving transfer of aluminum plate for interior decoration includes the following steps: S110, obtaining a front detection image of an aluminum material and a front detection image of a thermal transfer paper; S120, passing the front detection image of the aluminum material through a first convolutional neural network model as a feature extractor to obtain a surface state feature matrix of the aluminum material; S130, passing the front detection image of the thermal transfer paper through a second convolutional neural network model using a two-dimensional convolution kernel of a first scale to obtain a first-scale pattern feature matrix; S140, passing the front detection image of the thermal transfer paper through a third convolutional neural network model using a two-dimensional convolution kernel of a second scale A neural network model is used to obtain a second-scale pattern feature matrix; S150, the first-scale pattern feature matrix and the second-scale pattern feature matrix are integrated to obtain a pattern feature matrix; S160, a transfer matrix of the pattern feature matrix relative to the aluminum surface state feature matrix is ​​calculated as a decoding feature matrix; S170, based on the pattern feature matrix and the aluminum surface state feature matrix, the decoding feature matrix is ​​optimized for feature convergence to obtain an optimized decoding feature matrix; and, S180, the optimized decoding feature matrix is ​​passed through a decoder to obtain a decoding value, and the decoding value is used to represent a recommended baking temperature value.

[0090] In one example, in the above-mentioned control method for carving and transfer of aluminum plates for interior decoration, the step S120 includes: using each layer of the first convolutional neural network model as a feature extractor to perform the following on the input data in the forward pass of the layer: convolution processing on the input data to obtain a convolution feature map; pooling the convolution feature map along the channel dimension to obtain a pooled feature map; and, nonlinear activation on the pooled feature map to obtain an activation feature map; wherein the output of the last layer of the first convolutional neural network as a feature extractor is the surface state feature matrix of the aluminum material, and the input of the first layer of the first convolutional neural network as a feature extractor is the front detection image of the aluminum material.

[0091] In one example, in the above-mentioned control method for the transfer of carved patterns on aluminum plates for interior decoration, the step S130 includes: using each layer of the second convolutional neural network model using a two-dimensional convolutional kernel with a first scale to perform the following on the input data in the forward pass of the layer: convolution processing on the input data to obtain a convolution feature map; pooling the convolution feature map along the channel dimension to obtain a pooled feature map; and, nonlinear activation on the pooled feature map to obtain an activation feature map; wherein the output of the last layer of the second convolutional neural network using a two-dimensional convolutional kernel with a first scale is the first-scale pattern feature matrix, and the input of the first layer of the second convolutional neural network using a two-dimensional convolutional kernel with a first scale is the front detection image of the thermal transfer paper.

[0092] In one example, in the above-mentioned control method for carved transfer of aluminum plates for interior decoration, the step S140 includes: using each layer of the third convolutional neural network model using the two-dimensional convolution kernel with the second scale to perform the following on the input data in the forward pass of the layer: convolution processing on the input data to obtain a convolution feature map; pooling the convolution feature map along the channel dimension to obtain a pooled feature map; and, nonlinear activation on the pooled feature map to obtain an activation feature map; wherein the output of the last layer of the third convolutional neural network using the two-dimensional convolution kernel with the second scale is the second-scale pattern feature matrix, and the input of the first layer of the third convolutional neural network using the two-dimensional convolution kernel with the second scale is the front detection image of the thermal transfer paper.

[0093] In one example, in the above-mentioned control method for the transfer of carved patterns on aluminum plates for interior decoration, the step S150 includes: fusing the first scale pattern feature matrix and the second scale pattern feature matrix using the following formula to obtain a pattern feature matrix; wherein the formula is:

[0094]

[0095] Wherein M1 represents the first scale pattern feature matrix, M2 represents the second scale pattern feature matrix, M represents the pattern feature matrix, a1 and a2 represent weighting parameters of the first scale pattern feature matrix and the second scale pattern feature matrix respectively, Indicates sum by position.

[0096] In one example, in the above-mentioned control method for the transfer of carved patterns on aluminum plates for interior decoration, the step S160 includes: calculating the transfer matrix of the pattern feature matrix relative to the aluminum surface state feature matrix as a decoding feature matrix using the following formula; wherein the formula is:

[0097]

[0098] Where M represents the pattern feature matrix, M c represents the aluminum surface state characteristic matrix, M n represents the decoding feature matrix.

[0099] In one example, in the above-mentioned control method for the transfer of carved patterns on aluminum plates for interior decoration, the step S170 includes: expanding the pattern feature matrix and the aluminum surface state feature matrix into a pattern feature vector and an aluminum surface state feature vector; performing Hilbert space constraints on the pattern feature vector and the aluminum surface state feature vector to obtain a weight feature vector; reconstructing the weight feature vector into a weight feature matrix; calculating the position point multiplication between the weight feature matrix and the decoding feature matrix to obtain the optimized decoding feature matrix. Wherein, performing Hilbert space constraints on the pattern feature vector and the aluminum surface state feature vector to obtain the weight feature vector includes: performing Hilbert space constraints on the pattern feature vector and the aluminum surface state feature vector to obtain the weight feature vector according to the following formula; wherein, the formula is:

[0100]

[0101] Wherein V1 and V2 represent the pattern feature vector and the aluminum surface state feature vector respectively, ||·||2 represents the second norm of the vector, represents the positional addition of the vector, α and β are weighted hyperparameters, Cov 1D Represents the convolution operator (||V1||2,||V2||2,V1V2 T ) for vector Perform one-dimensional convolution, V w represents the weight feature vector.

[0102] In one example, in the above-mentioned control method for engraving transfer of aluminum plate for interior decoration, the step S180 includes: using the decoder to perform decoding regression on the optimized decoding feature matrix using the following formula to obtain a decoding value for representing the recommended baking temperature value; wherein the formula is: Where X represents the optimized decoding feature matrix, Y is the decoding value, and W is the weight matrix. Represents matrix multiplication.

[0103] In summary, according to the embodiment of the present application, the control method for transfer of aluminum plate carvings for interior decoration is explained, which mines the correlation feature information between the surface state characteristics of the transferred substrate and the implicit features of the pattern to be transferred by adopting a neural network model based on deep learning, and adaptively adjusts the baking transfer temperature value based on the surface state of the transfer substrate and the transfer pattern, thereby ensuring the transfer quality of the aluminum plate carvings and the decorative effect of the aluminum plate carvings.

[0104] Exemplary Electronic Devices

[0105] Below, reference Figure 7To describe an electronic device according to an embodiment of the present application.

[0106] Figure 7 A block diagram of an electronic device according to an embodiment of the present application is illustrated.

[0107] like Figure 7 As shown, the electronic device 10 includes one or more processors 11 and a memory 12 .

[0108] The processor 11 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.

[0109] The memory 12 may include one or more computer program products, and the computer program product may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 11 may run the program instructions to implement the functions of the interior decoration aluminum plate carved transfer control system of each embodiment of the present application described above and / or other desired functions. Various contents such as optimized decoding feature matrix may also be stored in the computer-readable storage medium.

[0110] In one example, the electronic device 10 may further include: an input device 13 and an output device 14, and these components are interconnected via a bus system and / or other forms of connection mechanisms (not shown).

[0111] The input device 13 may include, for example, a keyboard, a mouse, etc.

[0112] The output device 14 can output various information to the outside, including decoded values, etc. The output device 14 can include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto, and the like.

[0113] Of course, to simplify, Figure 7 Only some of the components related to the present application in the electronic device 10 are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, according to specific application situations, the electronic device 10 may also include any other appropriate components.

[0114] Exemplary computer program products and computer-readable storage media

[0115] In addition to the above-mentioned methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps in the functions of the method for controlling the transfer of carved patterns on aluminum plates for interior decoration according to various embodiments of the present application described in the above-mentioned "Exemplary System" section of this specification.

[0116] The computer program product may be written in any combination of one or more programming languages ​​to write program codes for performing the operations of the embodiments of the present application, including object-oriented programming languages, such as Java, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as an independent software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0117] In addition, an embodiment of the present application may also be a computer-readable storage medium on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes the steps in the functions of the method for controlling the transfer of carved patterns on aluminum plates for interior decoration according to various embodiments of the present application described in the above "Exemplary System" section of this specification.

[0118] The computer readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0119] The basic principles of the present application are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present application are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present application. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, not for limitation, and the above details do not limit the present application to being implemented by adopting the above specific details.

[0120] The block diagrams of the devices, apparatuses, equipment, and systems involved in this application are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The words "such as" used here refer to the phrase "such as but not limited to", and can be used interchangeably with them.

[0121] It should also be noted that in the apparatus, device and method of the present application, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0122] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

[0123] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. A control system for aluminum plate carving transfer for interior decoration, characterized in that: include: A camera module, used to obtain a front detection image of the aluminum material and a front detection image of the thermal transfer paper; An aluminum surface feature extraction module, used for passing the front detection image of the aluminum through a first convolutional neural network model as a feature extractor to obtain an aluminum surface state feature matrix; A first transfer pattern feature extraction module, used for obtaining a first-scale pattern feature matrix by using a second convolutional neural network model with a two-dimensional convolution kernel of a first scale on the front detection image of the thermal transfer paper; A second transfer pattern feature extraction module, used for obtaining a second-scale pattern feature matrix by using a third convolutional neural network model with a two-dimensional convolution kernel of a second scale on the front detection image of the thermal transfer paper; A multi-scale fusion module, used for fusing the first-scale pattern feature matrix and the second-scale pattern feature matrix to obtain a pattern feature matrix; A decoding feature generation module, used for calculating a transfer matrix of the pattern feature matrix relative to the aluminum material surface state feature matrix as a decoding feature matrix; A decoding feature distribution optimization module, used for optimizing the feature convergence of the decoding feature matrix based on the pattern feature matrix and the aluminum material surface state feature matrix to obtain an optimized decoding feature matrix; as well as The recommended control result generating module is used to pass the optimized decoding feature matrix through a decoder to obtain a decoding value, and the decoding value is used to represent the recommended baking temperature value.

2. The control system for interior decoration aluminum plate carving transfer according to claim 1 is characterized in that: The aluminum surface feature extraction module is further used to: use each layer of the first convolutional neural network model as a feature extractor to perform the following operations on the input data in the forward transfer of the layer: Perform convolution on the input data to obtain a convolution feature map; Pooling the convolutional feature map along the channel dimension to obtain a pooled feature map; as well as Performing nonlinear activation on the pooled feature map to obtain an activated feature map; Among them, the output of the last layer of the first convolutional neural network serving as a feature extractor is the surface state feature matrix of the aluminum material, and the input of the first layer of the first convolutional neural network serving as a feature extractor is the front detection image of the aluminum material.

3. The control system for interior decoration aluminum plate carving transfer according to claim 2 is characterized in that: The first transfer pattern feature extraction module is further used to: use each layer of the second convolutional neural network model using the two-dimensional convolution kernel with the first scale to respectively perform the following on the input data in the forward transfer of the layer: Perform convolution on the input data to obtain a convolution feature map; Pooling the convolutional feature map along the channel dimension to obtain a pooled feature map; as well as Performing nonlinear activation on the pooled feature map to obtain an activated feature map; Among them, the output of the last layer of the second convolutional neural network using the two-dimensional convolutional kernel with the first scale is the first-scale pattern feature matrix, and the input of the first layer of the second convolutional neural network using the two-dimensional convolutional kernel with the first scale is the front detection image of the thermal transfer paper.

4. The control system for interior decoration aluminum plate carving transfer according to claim 3 is characterized in that: The second transfer pattern feature extraction module is further used to: use each layer of the third convolutional neural network model using the two-dimensional convolution kernel with the second scale to respectively perform the following on the input data in the forward transfer of the layer: Perform convolution on the input data to obtain a convolution feature map; Pooling the convolutional feature map along the channel dimension to obtain a pooled feature map; as well as Performing nonlinear activation on the pooled feature map to obtain an activated feature map; Among them, the output of the last layer of the third convolutional neural network using the two-dimensional convolution kernel with the second scale is the second-scale pattern feature matrix, and the input of the first layer of the third convolutional neural network using the two-dimensional convolution kernel with the second scale is the front detection image of the thermal transfer paper.

5. The control system for interior decoration aluminum plate carving transfer according to claim 4 is characterized in that: The multi-scale fusion module is further used to: fuse the first-scale pattern feature matrix and the second-scale pattern feature matrix according to the following formula to obtain a pattern feature matrix; Wherein, the formula is: Wherein M1 represents the first scale pattern feature matrix, M2 represents the second scale pattern feature matrix, M represents the pattern feature matrix, a1 and a2 represent weighting parameters of the first scale pattern feature matrix and the second scale pattern feature matrix respectively, Indicates sum by position.

6. The control system for interior decoration aluminum plate carving transfer according to claim 5, characterized in that: The decoding feature generation module is further used to: calculate the transfer matrix of the pattern feature matrix relative to the aluminum material surface state feature matrix as a decoding feature matrix using the following formula; Wherein, the formula is: Where M represents the pattern feature matrix, M c represents the aluminum surface state characteristic matrix, M n represents the decoding feature matrix.

7. The control system for interior decoration aluminum plate carving transfer according to claim 6, characterized in that: The decoding feature distribution optimization module comprises: A feature matrix expansion unit, used for expanding the pattern feature matrix and the aluminum material surface state feature matrix into a pattern feature vector and an aluminum material surface state feature vector; A Hilbert space constraint unit, used for performing vector modulus basis Hilbert space constraints on the pattern feature vector and the aluminum material surface state feature vector to obtain a weight feature vector; A vector reconstruction unit, used for reconstructing the weight feature vector into a weight feature matrix; An action unit is used to calculate the position point multiplication between the weight feature matrix and the decoding feature matrix to obtain the optimized decoding feature matrix.

8. The control system for interior decoration aluminum plate carving transfer according to claim 7, characterized in that: The Hilbert space constraint unit is further used to: perform vector modulus basis Hilbert space constraint on the pattern feature vector and the aluminum material surface state feature vector according to the following formula to obtain the weight feature vector; Wherein, the formula is: Wherein V1 and V2 represent the pattern feature vector and the aluminum surface state feature vector respectively, ||·||2 represents the second norm of the vector, represents the positional addition of the vector, α and β are weighted hyperparameters, Cov 1D Represents the convolution operator (||V1||2,||V2||2,V1V2 T ) for vector Perform one-dimensional convolution, V w represents the weight feature vector.

9. The control system for interior decoration aluminum plate carving transfer according to claim 8, characterized in that: The recommended control result generating module includes: using the decoder to perform decoding regression on the optimized decoding feature matrix according to the following formula to obtain a decoding value for representing a baking recommended temperature value; Wherein, the formula is: Where X represents the optimized decoding feature matrix, Y is the decoding value, and W is the weight matrix. Represents matrix multiplication.

10. A method for controlling the transfer of carved patterns on aluminum plates for interior decoration, characterized in that: include: Acquire a front detection image of the aluminum material and a front detection image of the thermal transfer paper; Passing the front detection image of the aluminum material through a first convolutional neural network model as a feature extractor to obtain a surface state feature matrix of the aluminum material; The front detection image of the thermal transfer paper is subjected to a second convolutional neural network model having a two-dimensional convolution kernel of a first scale to obtain a first-scale pattern feature matrix; The front detection image of the thermal transfer paper is subjected to a third convolutional neural network model having a two-dimensional convolution kernel of a second scale to obtain a second-scale pattern feature matrix; fusing the first-scale pattern feature matrix and the second-scale pattern feature matrix to obtain a pattern feature matrix; Calculating a transfer matrix of the pattern feature matrix relative to the aluminum material surface state feature matrix as a decoding feature matrix; Based on the pattern feature matrix and the aluminum material surface state feature matrix, optimizing the feature convergence of the decoding feature matrix to obtain an optimized decoding feature matrix; as well as The optimized decoding feature matrix is ​​passed through a decoder to obtain a decoding value, and the decoding value is used to represent a baking recommended temperature value.

Images