Intelligent rate parameter control method based on information entropy of Softmax-like function

Abstract

The present application relates to intelligent sealing operation parameter control in the field of smart manufacturing, and specifically discloses an intelligent rate parameter control method based on information entropy of a Softmax-like function, which uses a convolutional neural network to extract the spatial features of the gluing position in an image, obtains the correlation between the local spatial features of the gluing position and the distribution characteristics of the overall spatial features of the windshield in a high-dimensional space through the information entropy expression of the Softmax-like function, and fuses the information entropy matrix of the Softmax-like function with the local spatial high-dimensional features of the gluing position to construct a feature matrix, so as to obtain the desired parameters, namely, the gluing rate and the movement rate, from the feature matrix through an encoder.

Description

Technical Field

[0001] The present invention relates to intelligent sealing operation parameter control in the field of smart manufacturing, and more specifically, to an intelligent rate parameter control method based on Softmax-like function information entropy, an intelligent rate parameter control system based on Softmax-like function information entropy, and an electronic device. Background Art

[0002] When installing or replacing a car's windshield, a glue sealing operation is required to effectively bond the windshield to the frame. During the current glue sealing operation, an automatic glue dispensing device has been used to replace manual glue extrusion, greatly improving the glue dispensing operation of the windshield.

[0003] Current automatic glue dispensing devices dispense glue at a constant rate at each application location and move at a constant speed along the windshield frame, ensuring even application of glue at each application location. However, because the windshield is not flat in the frame during the sealing operation, and the glue itself may move due to its fluid properties, this method of glue application can sometimes result in uneven application.

[0004] Therefore, it is expected to provide a solution for intelligently controlling the glue dispensing rate and movement rate of an automatic glue dispensing device.

[0005] Currently, deep learning and neural networks are widely used in fields such as computer vision, natural language processing, and text signal processing. Furthermore, deep learning and neural networks have demonstrated capabilities approaching or even surpassing those of humans in areas such as image classification, object detection, semantic segmentation, and text translation.

[0006] The development of deep learning and neural networks provides new solutions and options for intelligently controlling the glue dispensing rate and movement rate of automatic glue dispensing devices. Summary of the Invention

[0007] In order to solve the above technical problems, the present application is proposed. The embodiments of the present application provide an intelligent rate parameter control method based on information entropy of a Softmax function, an intelligent rate parameter control system and an electronic device based on information entropy of a Softmax function, which use a convolutional neural network to extract the spatial features of the gluing position in the image, obtain the correlation between the local spatial features of the gluing position and the distribution characteristics of the overall spatial features of the windshield in high-dimensional space through the information entropy expression of the Softmax function, and fuse the information entropy matrix of the Softmax function with the local spatial high-dimensional features of the gluing position to construct a feature matrix, so as to obtain the desired parameters, namely the gluing rate and the moving rate, from the feature matrix through an encoder.

[0008] According to one aspect of the present application, a method for intelligent rate parameter control based on information entropy of a Softmax-like function is provided, which includes:

[0009] Acquire a first image of the entire windshield on which the gluing operation is being performed and a second image of the gluing position on which the gluing operation is being performed;

[0010] Passing the first image and the second image through a deep convolutional neural network, respectively, to obtain a first feature map corresponding to the first image and a second feature map corresponding to the second image, wherein the last layer of the deep convolutional neural network is activated with a sigmoid activation function so that the feature value of each position in the first feature map and the second feature map is in the range of 0 to 1;

[0011] For the eigenvalue at each position in the second feature map, the information entropy of the eigenvalue at each position relative to the Softmax function of the first feature map is calculated using the following formula to obtain the information entropy matrix of the Softmax function of the second feature map relative to the first feature map, where the formula is:

[0012]

[0013] Wherein, P represents the information entropy of the Softmax-like function, yi represents the eigenvalue of each position in the second feature map, and xj represents the eigenvalue of each position in the first feature map;

[0014] Calculating a position-weighted sum of the information entropy matrix of the Softmax-like function and the second feature map to obtain a fused feature map; and

[0015] The fused feature map is passed through an encoder, the output bit number of the last layer of the encoder is two, to obtain two encoding values, and the two encoding values ​​are used to respectively represent the glue dispensing rate and the moving rate of the automatic glue dispensing device.

[0016] In the above-mentioned intelligent rate parameter control method based on information entropy of a Softmax-like function, a first image of the entire windshield undergoing the gluing operation and a second image of the gluing position where the gluing is being performed are obtained, including: obtaining an original second image of the gluing position where the gluing is being performed; and magnifying the original second image at a preset magnification to obtain the second image.

[0017] In the above-mentioned intelligent rate parameter control method based on information entropy of a Softmax-like function, the resolution of the first image is smaller than the resolution of the original second image.

[0018] In the above-mentioned intelligent rate parameter control method based on information entropy of a Softmax-like function, the encoder is a deep fully connected network.

[0019] In the above-mentioned intelligent rate parameter control method based on information entropy of a Softmax-like function, the deep fully connected network includes a fully connected layer with a preset depth, and the preset depth is greater than or equal to ten.

[0020] In the above-mentioned intelligent rate parameter control method based on information entropy of a Softmax-like function, the deep convolutional neural network is a deep residual network.

[0021] According to another aspect of the present application, an intelligent rate parameter control system based on information entropy of a Softmax-like function includes:

[0022] an image acquisition unit, configured to acquire a first image of the entire windshield on which the gluing operation is being performed and a second image of the gluing position on which the gluing operation is being performed;

[0023] a feature map generating unit, configured to pass the first image and the second image obtained by the image acquisition unit through a deep convolutional neural network, respectively, to obtain a first feature map corresponding to the first image and a second feature map corresponding to the second image, wherein the last layer of the deep convolutional neural network is activated with a Sigmoid activation function so that the feature value of each position in the first feature map and the second feature map is within the range of 0 to 1;

[0024] An information entropy matrix generating unit is configured to calculate the information entropy of the Softmax-like function of the eigenvalue at each position in the second feature map obtained by the feature map generating unit using the following formula to obtain an information entropy matrix of the Softmax-like function of the second feature map relative to the first feature map obtained by the feature map generating unit, wherein the formula is:

[0025]

[0026] Wherein, P represents the information entropy of the Softmax-like function, yi represents the eigenvalue of each position in the second feature map, and xj represents the eigenvalue of each position in the first feature map;

[0027] a fused feature map generating unit, configured to calculate a position-weighted sum of the information entropy matrix of the Softmax-like function obtained by the information entropy matrix generating unit and the second feature map obtained by the feature map generating unit to obtain a fused feature map; and

[0028] The encoding value generating unit is used to pass the fused feature map obtained by the fused feature map generating unit through an encoder, where the output bit number of the last layer of the encoder is two, to obtain two encoding values, and the two encoding values ​​are used to respectively represent the glue dispensing rate and the moving rate of the automatic glue dispensing device.

[0029] In the above-mentioned intelligent rate parameter control system based on information entropy of a Softmax-like function, the image acquisition unit includes: an original second image acquisition subunit, used to acquire the original second image of the gluing position where gluing is being performed; and a second image generation subunit, used to magnify the original second image obtained by the original second image acquisition subunit at a preset magnification to obtain the second image.

[0030] In the above-mentioned intelligent rate parameter control system based on information entropy of a Softmax-like function, the resolution of the first image is smaller than the resolution of the original second image.

[0031] In the above-mentioned intelligent rate parameter control system based on Softmax-like function information entropy, the encoder is a deep fully connected network.

[0032] In the above-mentioned intelligent rate parameter control system based on information entropy of a Softmax-like function, the deep fully connected network includes a fully connected layer with a preset depth, and the preset depth is greater than or equal to ten.

[0033] In the above-mentioned intelligent rate parameter control system based on information entropy of a Softmax-like function, the deep convolutional neural network is a deep residual network.

[0034] According to another aspect of the present application, an electronic device is provided, comprising: a processor; and a memory, in which computer program instructions are stored, and when the computer program instructions are executed by the processor, the processor executes the intelligent rate parameter control method based on Softmax-like function information entropy as described above.

[0035] According to yet 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 intelligent rate parameter control method based on Softmax-like function information entropy as described above.

[0036] Compared with the prior art, the intelligent rate parameter control method based on information entropy of a Softmax function, the intelligent rate parameter control system and the electronic device based on information entropy of a Softmax function provided by the present application use a convolutional neural network to extract the spatial features of the gluing position in the image, obtain the correlation between the local spatial features of the gluing position and the distribution characteristics of the overall spatial features of the windshield in high-dimensional space through the information entropy expression of a Softmax function, and fuse the information entropy matrix of the Softmax function with the local spatial high-dimensional features of the gluing position to construct a feature matrix, so as to obtain the desired parameters, namely the gluing rate and the moving rate, from the feature matrix through an encoder. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The above and other purposes, features, and advantages of the present application will become more apparent through a more detailed description of the embodiments of the present application in conjunction with the accompanying drawings. The accompanying drawings are intended 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 drawings, the same reference numerals generally represent the same components or steps.

[0038] Figure 1 The figure illustrates an application scenario diagram of the intelligent rate parameter control method based on Softmax-like function information entropy according to an embodiment of the present application;

[0039] Figure 2 The figure illustrates a flow chart of an intelligent rate parameter control method based on Softmax-like function information entropy according to an embodiment of the present application;

[0040] Figure 3 FIG2 illustrates a schematic diagram of a system architecture of an intelligent rate parameter control method based on Softmax-like function information entropy according to an embodiment of the present application;

[0041] Figure 4 FIG2 illustrates a flow chart of obtaining a first image of the entire windshield undergoing gluing and a second image of the gluing position undergoing gluing in an intelligent rate parameter control method based on information entropy of a Softmax-like function according to an embodiment of the present application;

[0042] Figure 5 The figure illustrates a block diagram of an intelligent rate parameter control system based on Softmax-like function information entropy according to an embodiment of the present application.

[0043] Figure 6 The figure illustrates a block diagram of an image acquisition unit in an intelligent rate parameter control system based on Softmax-like function information entropy according to an embodiment of the present application.

[0044] Figure 7The figure shows a block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0045] 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 herein.

[0046] Scenario Overview

[0047] As mentioned above, the current automatic glue dispensing device dispenses glue at a constant rate at each gluing position and moves at a uniform speed along the windshield frame. However, since the windshield is not placed flat in the frame during the windshield sealing operation, and the glue itself may move due to its own fluid properties, this gluing method sometimes results in uneven glue coating.

[0048] Based on this, the inventors of the present application hope to be able to control the glue dispensing rate and movement rate of the automatic glue dispensing device so as to adjust the current glue dispensing rate and movement rate according to the specific glue coating position, thereby improving the uniformity of glue coating.

[0049] The inventors of this application took into account that the gluing position reflects the overall spatial characteristics of the position itself relative to the windshield to be installed, and therefore it is appropriate to use a convolutional neural network to extract the spatial features of a predetermined area in the image, and construct a feature vector through the correlation between it and the spatial features of the entire image, so as to obtain the desired parameters, namely the gluing rate and the moving rate, from the feature vector through an encoder.

[0050] Furthermore, considering the high-dimensional nature of the spatial features extracted by convolutional neural networks, the solution of this application uses a Softmax-like information entropy expression to obtain the correlation between the distribution characteristics of local spatial features and global spatial features in high-dimensional space. In addition, since the scales of local spatial features and global spatial features in this application differ significantly, when sampling the source image space, the source image of the local spatial features is obtained at a higher resolution to compensate for the large scale difference.

[0051] Based on this, the present application proposes an intelligent rate parameter control method based on Softmax function-like information entropy, which includes: obtaining a first image of the entire windshield undergoing a gluing operation and a second image of the gluing position undergoing gluing; passing the first image and the second image through a deep convolutional neural network respectively to obtain a first feature map corresponding to the first image and a second feature map corresponding to the second image, wherein the last layer of the deep convolutional neural network is activated with a Sigmoid activation function so that the eigenvalue of each position in the first feature map and the second feature map is in the range of 0 to 1; for the eigenvalue of each position in the second feature map, the information entropy of the Softmax function of the eigenvalue of each position relative to the first feature map is calculated using the following formula to obtain the information entropy matrix of the Softmax function of the second feature map relative to the first feature map, wherein the formula is:

[0052]

[0053] Wherein, P represents the information entropy of the Softmax-like function, yi represents the eigenvalue of each position in the second feature map, and xj represents the eigenvalue of each position in the first feature map; the information entropy matrix of the Softmax-like function and the position-weighted sum of the second feature map are calculated to obtain a fused feature map; and the fused feature map is passed through an encoder whose last layer has an output bit number of two to obtain two encoding values, and the two encoding values ​​are used to respectively represent the glue dispensing rate and the moving rate of the automatic glue dispensing device.

[0054] Figure 1 The figure illustrates an application scenario diagram of the intelligent rate parameter control method based on Softmax-like function information entropy according to an embodiment of the present application.

[0055] like Figure 1 As shown, in this application scenario, through a camera (for example, Figure 1 C) obtaining a first image of the entire windshield undergoing the gluing operation and a second image of the gluing position undergoing the gluing operation; then, the first image and the second image are input to a server that is deployed with an intelligent rate parameter control algorithm based on information entropy of a Softmax function (for example, Figure 1 S) as shown in , wherein the server is capable of processing the first image and the second image based on an intelligent rate parameter control algorithm of information entropy of a Softmax function class to generate a coded value representing the glue dispensing rate and the moving rate of the automatic glue dispensing device.

[0056] 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.

[0057] Exemplary Methods

[0058] Figure 2 The figure shows a flow chart of the intelligent rate parameter control method based on the information entropy of the Softmax function. Figure 2 As shown, according to an embodiment of the present application, the intelligent rate parameter control method based on the information entropy of the Softmax function includes: S110, obtaining a first image of the entire windshield undergoing a gluing operation and a second image of the gluing position undergoing gluing; S120, passing the first image and the second image through a deep convolutional neural network respectively to obtain a first feature map corresponding to the first image and a second feature map corresponding to the second image, wherein the last layer of the deep convolutional neural network is activated with a Sigmoid activation function so that the eigenvalue of each position in the first feature map and the second feature map is in the range of 0 to 1; S130, for the eigenvalue of each position in the second feature map, calculating the information entropy of the Softmax function of the eigenvalue of each position relative to the first feature map using the following formula to obtain an information entropy matrix of the Softmax function of the second feature map relative to the first feature map, wherein the formula is:

[0059]

[0060] Wherein, P represents the information entropy of the Softmax-like function, yi represents the eigenvalue of each position in the second feature map, and xj represents the eigenvalue of each position in the first feature map; S140, calculates the position-weighted sum of the information entropy matrix of the Softmax-like function and the second feature map to obtain a fused feature map; and, S150, passes the fused feature map through an encoder, the output bit number of the last layer of the encoder is two, to obtain two encoding values, and the two encoding values ​​are used to respectively represent the glue dispensing rate and the moving rate of the automatic glue dispensing device.

[0061] Figure 3 The figure shows the architecture diagram of the intelligent rate parameter control method based on the information entropy of the Softmax function according to the embodiment of the present application. Figure 3 As shown, in the network architecture of the intelligent rate parameter control method based on information entropy of a Softmax-like function, first, the first image of the entire windshield undergoing the glue coating operation (for example, Figure 3 IN1) and a second image of the gluing location where gluing is being performed (e.g., as shown in FIG. Figure 3 IN2) are respectively trained by deep convolutional neural networks (e.g., Figure 3) to obtain a first feature map corresponding to the first image (e.g., as shown in Figure 3 F1 as shown in FIG) and a second feature map corresponding to the second image (e.g., as Figure 3 Then, for the eigenvalues ​​at each position in the second feature map, the information entropy of the eigenvalue at each position relative to the Softmax function of the first feature map is calculated to obtain the information entropy matrix of the Softmax function of the second feature map relative to the first feature map (for example, as Figure 3 Then, the information entropy matrix of the Softmax-like function and the position-weighted sum of the second feature map are calculated to obtain a fused feature map (for example, as Figure 3 Then, the fused feature map is passed through an encoder (e.g., Figure 3 The encoder shown in FIG), the number of output bits of the last layer of the encoder is two, so as to obtain two code values, and the two code values ​​are used to respectively represent the glue dispensing rate of the automatic glue dispensing device (for example, Figure 3 K1 as shown in the figure) and the movement speed (e.g. Figure 3 K2 shown in the figure).

[0062] In step S110, a first image of the entire windshield undergoing gluing and a second image of the gluing location are acquired. It should be understood that the gluing location reflects the spatial characteristics of the location relative to the entire windshield to be installed. Therefore, a convolutional neural network is suitable for extracting spatial features of a predetermined region in the image and processing them based on their correlation with the spatial features of the entire image to obtain the desired parameters.

[0063] Specifically, in an embodiment of the present application, the process of acquiring a first image of the entire windshield undergoing gluing and a second image of the gluing location undergoing gluing includes: first, acquiring an original second image of the gluing location undergoing gluing, wherein the resolution of the first image is lower than the resolution of the original second image; that is, the original second image of the gluing location undergoing gluing is acquired at a higher resolution. It should be understood that, because the scale of the local spatial features of the gluing location differs significantly from the scale of the overall spatial features of the windshield in this application, the large scale difference is compensated for by acquiring a source image of the local spatial features at a higher resolution when sampling the source image space. Then, the original second image is magnified by a preset magnification to obtain the second image. It should be understood that the size of the original second image is smaller than the size of the first image of the entire windshield. To facilitate subsequent calculations, the original second image is magnified by a preset magnification to ensure that the size of the second image is consistent with that of the first image.

[0064] Figure 4 The figure shows a flow chart of obtaining a first image of the entire windshield undergoing gluing and a second image of the gluing position undergoing gluing in an intelligent rate parameter control method based on information entropy of a Softmax-like function according to an embodiment of the present application. Figure 4 As shown, obtaining a first image of the entire windshield undergoing a gluing operation and a second image of the gluing position where gluing is being performed includes: S210, obtaining an original second image of the gluing position where gluing is being performed; and, S220, magnifying the original second image at a preset magnification to obtain the second image.

[0065] In step S120, the first image and the second image are respectively passed through a deep convolutional neural network to obtain a first feature map corresponding to the first image and a second feature map corresponding to the second image, wherein the last layer of the deep convolutional neural network is activated with a Sigmoid activation function so that the eigenvalue of each position in the first feature map and the second feature map is in the range of 0 to 1. That is, the deep convolutional neural network is used to extract each high-dimensional feature in the first image and the second image, and then the Sigmoid activation function is used to map each high-dimensional feature in the first image and the second image to a high-dimensional nonlinear interval for interpretation, thereby solving problems that cannot be solved by a linear model. Through activation with the Sigmoid activation function, the eigenvalue of each position in the first feature map and the second feature map is mapped to the range of 0 to 1.

[0066] In particular, in an embodiment of the present application, the deep convolutional neural network may adopt a deep residual neural network, for example, ResNet 50. Those skilled in the art should know that, compared with the traditional convolutional neural network, the deep residual network is an optimized network structure proposed on the basis of the traditional convolutional neural network, which mainly solves the problem of gradient disappearance during training. The deep residual network introduces a residual network structure, through which the network layer can be made deeper, and the problem of gradient disappearance does not occur. The residual network draws on the cross-layer connection idea of ​​the high-speed network, which breaks the convention that the input layer of the traditional neural network from the N-1 layer can only give the N layer as input, so that the output of a certain layer can directly cross several layers as the input of a certain layer behind it. Its significance lies in providing a new direction for the difficult problem of stacking multiple layers of networks and causing the error rate of the entire learning model to increase instead of decrease.

[0067] In step S130, for the eigenvalues ​​at each position in the second feature map, the information entropy of the eigenvalue at each position relative to the Softmax function of the first feature map is calculated using the following formula to obtain an information entropy matrix of the Softmax function of the second feature map relative to the first feature map, wherein the formula is:

[0068]

[0069] Where P represents the information entropy of the Softmax-like function, yi represents the eigenvalue at each position in the second feature map, and xj represents the eigenvalue at each position in the first feature map. That is, in the solution of this application, the correlation between the distribution characteristics of local spatial features and global spatial features in high-dimensional space is obtained based on the information entropy expression of the Softmax-like function.

[0070] In step S140, a weighted sum of the information entropy matrix of the Softmax-like function and the position of the second feature map is calculated to obtain a fused feature map. That is, a weighted sum of the information entropy matrix of the Softmax-like function and the feature matrix of the second feature map is calculated. The information entropy matrix of the Softmax-like function is equivalent to a mask that represents the weights that different regions in the second feature map should be paid attention to.

[0071] In step S150, the fused feature map is passed through an encoder, the output bit number of the last layer of the encoder is two, to obtain two encoding values, and the two encoding values ​​are used to respectively represent the glue dispensing rate and the moving rate of the automatic glue dispensing device.

[0072] Specifically, in the embodiment of the present application, the encoder is a deep fully connected network, which includes fully connected layers with a preset depth, where the preset depth is greater than or equal to ten. It should be understood that the deeper the neural network, the more abstract the features it extracts and the more attention it pays to details. Here, the deeper the fully connected layer, the more fully connected it is, the more effective the encoding can be.

[0073] In summary, the intelligent rate parameter control method based on the information entropy of the Softmax function-like embodiment of the present application is explained, which uses a convolutional neural network to extract the spatial features of the gluing position in the image, and obtains the correlation between the local spatial features of the gluing position and the distribution characteristics of the overall spatial features of the windshield in the high-dimensional space through the information entropy expression of the Softmax function-like function, and fuses the information entropy matrix of the Softmax function-like function with the local spatial high-dimensional features of the gluing position to construct a feature matrix, so as to obtain the desired parameters, namely the gluing rate and the moving rate, from the feature matrix through the encoder.

[0074] Exemplary Systems

[0075] Figure 5 The figure illustrates a block diagram of an intelligent rate parameter control system based on Softmax-like function information entropy according to an embodiment of the present application.

[0076] like Figure 5 As shown, according to an embodiment of the present application, an intelligent rate parameter control system 500 based on Softmax-like function information entropy includes: an image acquisition unit 510, used to acquire a first image of the entire windshield undergoing a gluing operation and a second image of the gluing position undergoing gluing; a feature map generation unit 520, used to pass the first image and the second image obtained by the image acquisition unit 510 through a deep convolutional neural network respectively to obtain a first feature map corresponding to the first image and a second feature map corresponding to the second image, wherein the last layer of the deep convolutional neural network is activated with a Sigmoid activation function so that the eigenvalue of each position in the first feature map and the second feature map is in the range of 0 to 1; an information entropy matrix generation unit 530, used to calculate the Softmax-like function information entropy of the eigenvalue of each position in the second feature map obtained by the feature map generation unit 520 relative to the first feature map obtained by the feature map generation unit using the following formula to obtain an information entropy matrix of the Softmax-like function of the second feature map relative to the first feature map, wherein the formula is:

[0077]

[0078] Wherein, P represents the information entropy of the Softmax-like function, yi represents the eigenvalue of each position in the second feature map, and xj represents the eigenvalue of each position in the first feature map; the fused feature map generation unit 540 is used to calculate the position-weighted sum of the information entropy matrix of the Softmax-like function obtained by the information entropy matrix generation unit 530 and the second feature map obtained by the feature map generation unit 520 to obtain a fused feature map; and the parameter generation unit 550 is used to pass the fused feature map obtained by the fused feature map generation unit 540 through an encoder, the output bit number of the last layer of the encoder is two, to obtain two encoding values, and the two encoding values ​​are used to respectively represent the glue dispensing rate and the moving rate of the automatic glue dispensing device.

[0079] In one example, in the above-mentioned intelligent rate parameter control system 500, as Figure 6As shown, the image acquisition unit 510 includes: an original second image acquisition subunit 511, used to acquire an original second image of the gluing position where gluing is being performed; and a second image generation subunit 512, used to magnify the original second image obtained by the original second image acquisition subunit 511 at a preset magnification to obtain the second image.

[0080] In one example, in the intelligent rate parameter control system 500 , the resolution of the first image is smaller than the resolution of the original second image.

[0081] In one example, in the intelligent rate parameter control system 500 , the encoder is a deep fully connected network.

[0082] In one example, in the intelligent rate parameter control system 500 , the deep fully connected network includes a fully connected layer with a preset depth, and the preset depth is greater than or equal to ten.

[0083] In one example, in the intelligent rate parameter control system 500 , the deep convolutional neural network is a deep residual network.

[0084] Here, those skilled in the art will appreciate that the specific functions and operations of the various units and modules in the intelligent rate parameter control system 500 have been described in detail above. Figures 1 to 4 The description of the intelligent rate parameter control method based on the information entropy of the Softmax function has been introduced in detail, and therefore, its repeated description will be omitted.

[0085] As described above, the intelligent rate parameter control system 500 according to the embodiment of the present application can be implemented in various terminal devices, such as a server for controlling the glue dispensing rate and movement rate of an automatic glue dispensing device. In one example, the intelligent rate parameter control system 500 according to the embodiment of the present application can be integrated into the terminal device as a software module and / or hardware module. For example, the intelligent rate parameter control system 500 can be a software module in the operating system of the terminal device, or it can be an application developed for the terminal device; of course, the intelligent rate parameter control system 500 can also be one of the many hardware modules of the terminal device.

[0086] Alternatively, in another example, the intelligent rate parameter control system 500 and the terminal device may be separate devices, and the intelligent rate parameter control system 500 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.

[0087] Exemplary electronic devices

[0088] Below, reference Figure 7 To describe the electronic device according to the embodiment of the present application.

[0089] Figure 7 The figure shows a block diagram of an electronic device according to an embodiment of the present application.

[0090] like Figure 7 As shown, the electronic device 10 includes one or more processors 11 and a memory 12 .

[0091] 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.

[0092] The memory 12 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may, for example, include random access memory (RAM) and / or cache memory (cache), etc. The non-volatile memory may, for example, include read-only memory (ROM), a hard disk, a flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 11 may execute the program instructions to implement the functions of the intelligent rate parameter control method based on the Softmax-like function information entropy of the various embodiments of the present application described above and / or other desired functions. Various contents such as the information entropy matrix of the Softmax-like function, the fusion feature map, etc. may also be stored in the computer-readable storage medium.

[0093] In one example, the electronic device 10 may further include an input system 13 and an output system 14 , and these components are interconnected via a bus system and / or other forms of connection mechanisms (not shown).

[0094] The input system 13 may include, for example, a keyboard, a mouse, and the like.

[0095] The output system 14 can output various information to the outside, including code values, etc. The output system 14 can include, for example, a display, a speaker, a printer, a communication network and remote output devices connected thereto, and the like.

[0096] 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, the electronic device 10 may further include any other appropriate components according to specific application scenarios.

[0097] Exemplary computer program products and computer-readable storage media

[0098] 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 perform the steps in the functions of the intelligent rate parameter control method based on Softmax-like function information entropy according to various embodiments of the present application described in the above "Exemplary Method" section of this specification.

[0099] The computer program product may be written in any combination of one or more programming languages ​​to implement the program code for performing the operations of the embodiments of the present application, including object-oriented programming languages ​​such as Java, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0100] In addition, an embodiment of the present application may also be a computer-readable storage medium having computer program instructions stored thereon, which, when executed by a processor, enable the processor to execute the steps of the intelligent rate parameter control method based on Softmax-like function information entropy described in the above-mentioned "Exemplary Method" section of this specification.

[0101] 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, for example, include but is not limited to a system, system or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination thereof. 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 thereof.

[0102] The basic principles of the present application have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, and effects mentioned in this application are merely illustrative and not restrictive, and it should not be assumed that these advantages, strengths, and effects are required of each embodiment of this application. In addition, the specific details disclosed above are merely illustrative and facilitating understanding, and are not restrictive. The above details do not limit this application to necessarily being implemented using the above specific details.

[0103] The block diagrams of the devices, devices, equipment, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As will be appreciated by those skilled in the art, these devices, devices, equipment, and systems can be connected, arranged, or configured in any manner. Words such as "include," "comprise," "have," and the like are open-ended words, meaning "including but not limited to," and can be used interchangeably therewith. The words "or" and "and" used herein refer to the words "and / or" and can be used interchangeably therewith, unless the context clearly indicates otherwise. The word "such as" used herein refers to the phrase "such as but not limited to," and can be used interchangeably therewith.

[0104] 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, and such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0105] 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 be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0106] The above description has been provided for the purpose of illustration and description. Furthermore, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although a number of example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. An intelligent rate parameter control method based on information entropy of a Softmax-like function, characterized in that: include: Acquire a first image of the entire windshield on which the gluing operation is being performed and a second image of the gluing position on which the gluing operation is being performed; Passing the first image and the second image through a deep convolutional neural network, respectively, to obtain a first feature map corresponding to the first image and a second feature map corresponding to the second image, wherein the last layer of the deep convolutional neural network is activated with a sigmoid activation function so that the feature value of each position in the first feature map and the second feature map is in the range of 0 to 1; Calculating the information entropy of the Softmax function of the eigenvalue at each position in the second feature map relative to the first feature map to obtain an information entropy matrix of the Softmax function of the second feature map relative to the first feature map; Calculating a position-weighted sum of the information entropy matrix of the Softmax-like function and the second feature map to obtain a fused feature map; and Passing the fused feature map through an encoder, wherein the number of output bits of the last layer of the encoder is two, to obtain two encoding values, wherein the two encoding values ​​are used to represent the glue dispensing rate and the moving rate of the automatic glue dispensing device respectively; The method of acquiring a first image of the entire windshield on which the gluing operation is being performed and a second image of the gluing position on which the gluing operation is being performed includes: Acquiring an original second image of the gluing location where gluing is being performed; and The original second image is magnified at a preset magnification to obtain the second image.

2. The intelligent rate parameter control method based on information entropy of a Softmax-like function according to claim 1, wherein: The resolution of the first image is smaller than the resolution of the original second image.

3. The intelligent rate parameter control method based on Softmax-like function information entropy according to claim 1, wherein: The encoder is a deep fully connected network.

4. The intelligent rate parameter control method based on Softmax-like function information entropy according to claim 3, wherein: The deep fully connected network includes a fully connected layer with a preset depth, and the preset depth is greater than or equal to ten.

5. The intelligent rate parameter control method based on Softmax-like function information entropy according to claim 1, wherein: The deep convolutional neural network is a deep residual network.

6. An intelligent rate parameter control system based on information entropy of a Softmax-like function, characterized in that: include: an image acquisition unit, configured to acquire a first image of the entire windshield on which the gluing operation is being performed and a second image of the gluing position on which the gluing operation is being performed; a feature map generating unit, configured to pass the first image and the second image obtained by the image acquisition unit through a deep convolutional neural network, respectively, to obtain a first feature map corresponding to the first image and a second feature map corresponding to the second image, wherein the last layer of the deep convolutional neural network is activated with a Sigmoid activation function so that the feature value of each position in the first feature map and the second feature map is within the range of 0 to 1; an information entropy matrix generating unit, configured to calculate the information entropy of the Softmax-like function of the eigenvalue at each position in the second feature map relative to the first feature map obtained by the feature map generating unit, so as to obtain an information entropy matrix of the Softmax-like function of the second feature map relative to the first feature map; a fused feature map generating unit, configured to calculate a position-weighted sum of the information entropy matrix of the Softmax-like function obtained by the information entropy matrix generating unit and the second feature map obtained by the feature map generating unit to obtain a fused feature map; and a coding value generating unit, configured to pass the fused feature map obtained by the fused feature map generating unit through an encoder, wherein the number of output bits of the last layer of the encoder is two, to obtain two coding values, wherein the two coding values ​​are used to respectively represent a glue dispensing rate and a moving rate of the automatic glue dispensing device; Wherein, the image acquisition unit includes: an original second image acquisition subunit, configured to acquire an original second image of the gluing position where gluing is being performed; and The second image generating subunit is configured to amplify the original second image obtained by the original second image acquiring subunit at a preset magnification to obtain the second image.

7. The intelligent rate parameter control system based on Softmax-like information entropy according to claim 6, wherein: The resolution of the first image is smaller than the resolution of the original second image.

8. An electronic device comprising: processor; as well as 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 intelligent rate parameter control method based on Softmax-like function information entropy according to any one of claims 1 to 5.

Images