A face micro-expression recognition method based on a double attention mechanism

Abstract

The application discloses a face micro-expression recognition method based on a double attention mechanism and belongs to the technical field of image processing. The method solves the problem of large splicing error of a series feature splicing mode of an existing method, thereby causing inaccurate recognition. The method comprises the following steps: acquiring a face micro-expression image; processing the face micro-expression image based on a pre-trained double attention recognition model and converting the processed face micro-expression image into image element information of different dimensions; and constructing the converted image element information into a three-dimensional micro-expression model, which is displayed through stereoscopic data information output. The method processes the face micro-expression image through the pre-trained double attention recognition model, converts the processed face micro-expression image into image element information of different dimensions, and then accurately recognizes the three-dimensional micro-expression model, thereby improving the recognition efficiency.

Description

Technical Field

[0001] This invention belongs to the field of image processing technology, specifically relating to a method for facial micro-expression recognition based on a dual attention mechanism. Background Technology

[0002] Microexpressions are a psychological term. People express their inner feelings to others by making certain facial expressions. Between different expressions or within a single expression, the face can "reveal" other information. Microexpressions are short-lived; a subconscious expression may only last for a moment, but it can easily expose emotions.

[0003] To improve the efficiency and applicability of micro-expression recognition methods, machine recognition methods are currently commonly used to identify human micro-expressions. This requires acquiring video data of the object to be identified and then using a deep learning-based recognition model to perform micro-expression recognition on the video data. Chinese patent CN113963423A discloses a micro-expression recognition method, system, device, and storage medium based on neural networks. The method includes: preprocessing micro-expression images in the video to be identified to obtain preprocessed micro-expression images; extracting the spatiotemporal features of micro-expressions from the preprocessed micro-expression images; and inputting the spatiotemporal features of micro-expressions from the preprocessed micro-expression images into a micro-expression recognition model to identify the micro-expressions in the video to be identified. However, existing methods using concatenated feature splicing have the problem of large splicing errors, resulting in inaccurate recognition. Based on this, this invention proposes a facial micro-expression recognition method based on a dual attention mechanism. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by providing a facial micro-expression recognition method based on a dual attention mechanism. This method solves the problem of large splicing errors in existing methods that use serial feature splicing, which leads to inaccurate recognition.

[0005] This invention is implemented as follows: a facial micro-expression recognition method based on a dual attention mechanism, comprising:

[0006] Acquire facial micro-expression images; wherein, the facial micro-expression images are captured through omnidirectional image information capture to obtain facial micro-expression information from different angles;

[0007] The pre-trained dual attention recognition model is used to process facial micro-expression images and convert the processed facial micro-expression images into image element information of different dimensions, which meets the requirements of AR space three-dimensional modeling.

[0008] The converted image element information is used to construct a three-dimensional micro-expression model, which is then displayed through stereoscopic data information output.

[0009] Preferably, acquiring facial micro-expression images specifically includes:

[0010] Data information is extracted from the working process of the facial micro-expression acquisition device. The monitored data information is statistically analyzed from the data information of the captured images. Then, the clearest captured image data is found based on the following function formula.

[0011] F(x) = min(min(T) σ (y)))→0

[0012] Where F(x) is the set of image data acquired by the device; T σ (y) represents the information variable of the acquiring device; y represents the high-definition image set of the acquired image data information; T represents the recognition time of the decomposed image; σ represents the fuzzing and denoising constant of the decomposed monitoring image data information.

[0013] Preferably, the acquisition of facial micro-expression images further includes:

[0014] The clearest captured image data is obtained, and the blurry area data information of all captured images is filtered and removed based on the dark channel method to obtain the cleaned image dataset.

[0015] Preferably, the training method for the dual-attention recognition model specifically includes:

[0016] Obtain sample data for training the dual attention recognition model, and divide the sample data into training set, test set and validation set in a 3:1:1 ratio;

[0017] The pre-established dual attention recognition model is trained based on the training set and the preset dual attention mechanism CNN-LSTM model algorithm to obtain the trained dual attention recognition model.

[0018] Extract the test set and prune the trained dual attention recognition model to obtain multiple pruned dual attention recognition models.

[0019] The optimal dual attention recognition model is selected from all the pruned dual attention recognition models based on cross-validation.

[0020] Preferably, the training method for the dual-attention recognition model further includes:

[0021] A validation set is extracted, and a channel attention mechanism is used to assign weights to the max pooling features and the average pooling features in the dual attention recognition model to output image spatial features.

[0022] Preferably, the step of extracting the test set to prune the trained dual-attention recognition model to obtain multiple pruned dual-attention recognition models specifically includes:

[0023] All data in the test dataset are input into the initial dual-attention recognition model. After all the data enters the initial decision tree model, the loss of each node in the initial dual-attention recognition model is calculated.

[0024] Determine whether the loss of each node in the initial dual-attention recognition model is less than a preset threshold, where the preset threshold for each node is obtained based on a preset threshold set; if the loss of the current node is less than the preset threshold, stop building the tree for the current node. If the loss of the current node is greater than the preset threshold, adjust the node parameters of the initial decision tree model and prune it again based on the dual-attention mechanism.

[0025] Preferably, the pre-trained dual-attention recognition model processes facial micro-expression images, specifically including:

[0026] Obtain the image dataset after clearing;

[0027] The image is decomposed using the decomposition function of the dual attention recognition model, and the recognition coordinates are marked in the decomposed monitoring image data.

[0028] The image is decomposed into data, and tolerance calculations are performed on invalid regions in the decomposed data to obtain image element information of different dimensions after processing.

[0029] Preferably, the decomposition function is:

[0030]

[0031] in, The result is the image decomposition, where n is the total number of images in the image set, i is the index of an image in the image set, and K is the index of the image. σ σ represents the influence factor of external environmental data information when acquiring images; σ represents the fuzzy denoising constant for blurry areas generated during the information acquisition process.

[0032] Preferably, the converted image element information is used to construct a three-dimensional micro-expression model, specifically including:

[0033] At least one interactive display area is formed in the 3D micro-expression display interface;

[0034] Obtain image element information from different dimensions;

[0035] The display of virtual image element information is achieved using the MFC algorithm model.

[0036] Preferably, the display of virtual image element information using the MFC algorithm model specifically includes:

[0037] Image element information of different dimensions is defined as dimensional features based on a dual attention recognition model;

[0038] The dimensional features of image data information modules of different dimensions are defined by a dual attention recognition model. The dimensional feature definition is implemented by a convolution algorithm model.

[0039] Based on a virtual 3D modeling environment, the dynamic demonstration of 3D micro-expression models is achieved through a monitor;

[0040] The dual-attention recognition model identifies facial micro-expression images based on the content displayed in the 3D micro-expression model, and obtains the recognition results. The results are presented as a dynamic 3D model demonstration and a demonstration of local feature information, depending on the display of the 3D micro-expression model.

[0041] Compared with the prior art, the embodiments of this application have the following main advantages:

[0042] This application processes facial micro-expression images using a pre-trained dual-attention recognition model, converts the processed images into image element information of different dimensions, and then accurately recognizes the three-dimensional micro-expression model. This solves the problem of large splicing errors in existing methods that use serial feature splicing, which leads to inaccurate recognition and improves recognition efficiency. Attached Figure Description

[0043] Figure 1 This is a schematic diagram illustrating the implementation process of the facial micro-expression recognition method based on the dual attention mechanism provided by the present invention.

[0044] Figure 2 This is a schematic diagram illustrating the process of acquiring facial micro-expression images provided by the present invention.

[0045] Figure 3 This is a schematic diagram illustrating the implementation process of the training method for the dual attention recognition model provided by this invention.

[0046] Figure 4 This is a schematic diagram illustrating the implementation process of extracting a test set to prune the trained dual-attention recognition model, thereby obtaining multiple pruned dual-attention recognition models.

[0047] Figure 5 This is a schematic diagram illustrating the implementation process of processing facial micro-expression images using a pre-trained dual attention recognition model provided by the present invention.

[0048] Figure 6This is a schematic diagram illustrating the implementation process of facial micro-expression recognition based on a dual attention mechanism provided by the present invention.

[0049] Figure 7 This is a schematic diagram illustrating the implementation process of displaying virtual image element information using the MFC algorithm model provided by the present invention.

[0050] Figure 8 This is a schematic diagram of the facial micro-expression recognition system based on the dual attention mechanism provided by the present invention. Detailed Implementation

[0051] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0052] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs; the terminology used herein in the specification of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "comprising" and "having," and any variations thereof, in the specification, claims, and foregoing drawings of this application are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification, claims, or foregoing drawings of this application are used to distinguish different objects, not to describe a particular order.

[0053] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0054] Existing methods using serial feature stitching suffer from large stitching errors, leading to inaccurate recognition. To address this, this invention proposes a facial micro-expression recognition method based on a dual-attention mechanism. The method includes: acquiring facial micro-expression images; processing the facial micro-expression images using a pre-trained dual-attention recognition model; constructing a three-dimensional micro-expression model from the converted image element information; and displaying the three-dimensional micro-expression model through stereoscopic data output. This invention processes facial micro-expression images using a pre-trained dual-attention recognition model, converts the processed images into image element information of different dimensions, and then accurately recognizes the three-dimensional micro-expression model. This solves the problem of large stitching errors in existing methods using serial feature stitching, thus improving recognition efficiency.

[0055] This invention provides a method for facial micro-expression recognition based on a dual attention mechanism, such as... Figure 1 The diagram illustrates the implementation flow of a facial micro-expression recognition method based on a dual attention mechanism, specifically including:

[0056] Step S10: Acquire facial micro-expression images; wherein, the facial micro-expression images are captured by omnidirectional image information to obtain facial micro-expression information from different angles;

[0057] Step S20: The facial micro-expression image is processed based on the pre-trained dual attention recognition model, and the processed facial micro-expression image is converted into image element information of different dimensions, wherein the image element information meets the requirements of AR space three-dimensional modeling.

[0058] Step S30: The converted image element information is used to construct a three-dimensional micro-expression model, which is then displayed through stereoscopic data information output.

[0059] In this embodiment, the present invention processes facial micro-expression images using a pre-trained dual-attention recognition model, converts the processed facial micro-expression images into image element information of different dimensions, and then accurately recognizes the three-dimensional micro-expression model. This solves the problem of large splicing errors in existing methods that use serial feature splicing, which leads to inaccurate recognition and improves recognition efficiency.

[0060] The method for obtaining facial micro-expression images in step S10 above is as follows: Figure 2 The diagram illustrates the process of acquiring facial micro-expression images, specifically including:

[0061] Step S101: Extract data information from the working process of the facial micro-expression acquisition device, and statistically analyze the monitored data information from the data information of the captured images, such as the amount of image data contained in the captured images, the network protocol (Internet Protocol, IP) address of the monitoring device, the SNMP version number, and the SNMP community number, and then find the clearest captured image data based on the following function formula;

[0062] F(x) = min(min(T) σ (y))) →0 (1)

[0063] Where F(x) is the set of image data acquired by the device, and T σ(y) is the information variable of the acquisition device (i.e. the information variable of the image acquired by the monitoring device), where y represents the high-definition image set of the acquired image data information, T represents the recognition time of the decomposed image, and σ represents the fuzzing and denoising constant of the decomposed monitoring image data information; the clearest captured image data can be output through formula (1).

[0064] Step S102: Obtain the clearest captured image data, filter and remove the blurry area data information of all captured images based on the dark channel method, and obtain the cleared image dataset.

[0065] In this embodiment, the facial micro-expression acquisition device can be a 360-degree panoramic camera or a night vision surveillance camera. The facial micro-expression acquisition device uses 360's cloud recording function, which can automatically store the video recording of the scene changes in the cloud. The facial micro-expression acquisition device is a blind-spot-free monitoring device, and the monitoring coverage area of ​​the facial micro-expression acquisition device is 30 square meters. In a specific embodiment, a fisheye panoramic imaging optical system may be used.

[0066] The training method for the dual-attention recognition model in step S20 above is as follows: Figure 3 The diagram illustrates the implementation flow of the training method for the dual-attention recognition model. The training method specifically includes:

[0067] Step S201: Obtain sample data for training the dual attention recognition model, and divide the sample data into training set, test set and validation set in a ratio of 3:1:1;

[0068] Step S202: Based on the training set and the preset dual attention mechanism CNN-LSTM model algorithm, train the pre-established dual attention recognition model to obtain the trained dual attention recognition model.

[0069] Step S203: Extract the test set and prune the trained dual attention recognition model to obtain multiple pruned dual attention recognition models.

[0070] Step S204: Select the optimal dual attention recognition model from all the pruned dual attention recognition models based on cross-validation;

[0071] Step S205: Extract the validation set, and use the channel attention mechanism to assign weights to the max pooling feature and the average pooling feature in the dual attention recognition model to output image spatial features.

[0072] In this embodiment, the CNN-LSTM model is a neural network model, specifically a neural network model in machine learning. It is a complex neural network system formed by extensive interconnection of a large number of simple processing units (called neurons). It reflects many fundamental characteristics of human brain function and is a highly complex nonlinear dynamic learning system. Simply put, it is a mathematical model. In this embodiment, sample data is input into the dual-attention recognition model for continuous training, making the matching degree of the dual-attention recognition model more optimized and accurate.

[0073] This invention provides a method for extracting a test set and pruning the trained dual-attention recognition model to obtain multiple pruned dual-attention recognition models, such as... Figure 4 The diagram illustrates the implementation process of pruning the trained dual-attention recognition model using a test set to obtain multiple pruned dual-attention recognition models; specifically, it includes:

[0074] Step S2031: Input all data in the test dataset into the initial dual attention recognition model;

[0075] Step S2032: After all the data enters the initial decision tree model, calculate the loss of each node in the initial dual attention recognition model.

[0076] Step S2033: Determine whether the loss of each node in the initial dual attention recognition model is less than a preset threshold, wherein the preset threshold of each node is obtained based on a preset threshold set.

[0077] Step S2034: If the loss of the current node is less than a preset threshold, stop building the tree for the current node; if the loss of the current node is greater than the preset threshold, adjust the node parameters of the initial decision tree model and prune it again based on the dual attention mechanism.

[0078] This invention provides a process for processing facial micro-expression images based on a pre-trained dual-attention recognition model, such as... Figure 5 The diagram illustrates the implementation process of processing facial micro-expression images based on a pre-trained dual-attention recognition model, specifically including:

[0079] Step S301: Obtain the image dataset after cleaning;

[0080] Step S302: Decompose the image using the decomposition function of the dual attention recognition model, and mark the recognition coordinates in the decomposed monitoring image data;

[0081] Step S303: Decompose the image data, perform tolerance calculation on invalid regions in the decomposed data, and obtain image element information of different dimensions after processing.

[0082] In this application, the decomposition function is:

[0083]

[0084] in, The result is the image decomposition, where n is the total number of images in the image set, i is the index of an image in the image set, and K is the index of the image. σ This indicates the influence factor of external environmental data information when acquiring images, and σ represents the fuzzy denoising constant for blurry areas generated during the information acquisition process.

[0085] This invention provides a method for facial micro-expression recognition based on a dual attention mechanism, such as... Figure 6 The diagram illustrates the implementation process of the facial micro-expression recognition based on the dual attention mechanism, specifically including:

[0086] Step S401: Form at least one interactive display area in the 3D micro-expression display interface;

[0087] Step S402: Obtain image element information in different dimensions;

[0088] Step S403: Display virtual image element information using the MFC algorithm model.

[0089] This invention provides a method for displaying virtual image element information using an MFC algorithm model, such as... Figure 7 The diagram illustrates the implementation process of displaying virtual image element information using the MFC algorithm model, specifically including:

[0090] Step S4031: Define the dimensional features of image element information of different dimensions according to the dual attention recognition model;

[0091] Step S4032: Define the dimensional features of image data information modules of different dimensions through a dual attention recognition model. The dimensional feature definition adopts the convolution algorithm model to define the dimensional features of the data.

[0092] Step S4033 is based on a virtualized 3D modeling environment, and the 3D micro-expression model is dynamically demonstrated through a monitor;

[0093] In step S4034, the dual attention recognition model recognizes the facial micro-expression image based on the display content of the three-dimensional micro-expression model to obtain the recognition result. The display of the three-dimensional micro-expression model is a dynamic demonstration of the 3D model and a demonstration of local feature information.

[0094] Based on the same inventive concept, embodiments of the present invention also provide a facial micro-expression recognition system based on a dual attention mechanism, such as... Figure 8The diagram shows the structure of a facial micro-expression recognition system based on a dual attention mechanism, specifically including:

[0095] Image acquisition module 100, which is used to acquire facial micro-expression images; wherein, the facial micro-expression images are captured by omnidirectional image information to acquire facial micro-expression information from different angles;

[0096] The model processing module 200 processes facial micro-expression images based on a pre-trained dual attention recognition model and converts the processed facial micro-expression images into image element information of different dimensions, wherein the image element information meets the requirements of AR space three-dimensional modeling.

[0097] The three-dimensional model construction module 300 constructs a three-dimensional micro-expression model from the converted image element information, and the three-dimensional micro-expression model is displayed through stereoscopic data information output.

[0098] In this embodiment, the present invention processes facial micro-expression images using a pre-trained dual-attention recognition model, converts the processed facial micro-expression images into image element information of different dimensions, and then accurately recognizes the three-dimensional micro-expression model. This solves the problem of large splicing errors in existing methods that use serial feature splicing, which leads to inaccurate recognition and improves recognition efficiency.

[0099] Based on the same inventive concept, this embodiment of the invention also provides a schematic diagram of a computer device, which includes a display screen, a memory, a processor, and a computer program. The memory stores the computer program, and when the computer program is executed by the processor, the processor performs the steps of the facial micro-expression recognition method based on the dual attention mechanism.

[0100] It is understood that, in the preferred embodiments provided by the present invention, the computer device may also be a laptop computer, a personal digital assistant (PDA), a mobile phone, or other device capable of communication.

[0101] Based on the same inventive concept, the present invention also provides a readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the facial micro-expression recognition method based on a dual attention mechanism.

[0102] For example, a computer program can be divided into one or more modules, one or more of which are stored in memory and executed by a processor to perform the present invention. One or more modules can be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in a terminal device. For example, the aforementioned computer program can be divided into units or modules of the facial micro-expression recognition system based on a dual-attention mechanism provided in the various system embodiments described above.

[0103] The processor referred to can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. This processor is the control center of the terminal device, connecting various parts of the user terminal via various interfaces and lines.

[0104] The aforementioned memory can be used to store computer programs and / or modules. The aforementioned processor implements various functions of the aforementioned terminal device by running or executing the computer programs and / or modules stored in the memory, and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area; the data storage area may store data created based on the use of a facial micro-expression system based on a dual-attention mechanism. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, RAM, plug-in hard disk, SmartMedia Card (SMC), Secure Digital (SD) card, Flash Card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0105] Finally, it should be noted that the computer-readable storage medium (e.g., memory) described herein can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. By way of example, and not limitation, non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM), which can act as external cache memory. By way of example, and not limitation, RAM can be obtained in various forms, such as synchronous RAM (DRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct Rambus RAM (DRRAM). The storage devices disclosed herein are intended to include, but are not limited to, these and other suitable types of memory.

[0106] The various exemplary logic blocks, modules, and circuits described herein can be implemented or performed using the following components designed to perform the functions herein: general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic, discrete hardware components, or any combination of these components. A general-purpose processor may be a microprocessor, but alternatively, the processor may be any conventional processor, controller, microcontroller, or state machine. The processor may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors combined with a DSP, and / or any other such configuration.

[0107] In summary, this invention provides a facial micro-expression recognition method based on a dual attention mechanism. It processes facial micro-expression images using a pre-trained dual attention recognition model, converts the processed images into image element information of different dimensions, and then accurately recognizes the three-dimensional micro-expression model. This solves the problem of large splicing errors in existing methods that use serial feature splicing, which leads to inaccurate recognition and improves recognition efficiency.

[0108] It should be noted that, for the sake of simplicity, the foregoing embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to the present invention. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0109] It should be understood that the disclosed apparatus can be implemented in other ways, given the several embodiments provided in this application. For example, the apparatus embodiments described above are merely illustrative; the division of units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or communication connections shown or discussed may be through some interfaces; the indirect coupling or communication connections between devices or units may be telecommunications or other forms.

[0110] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0111] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit the scope of protection of the invention.

[0112] Obviously, the described embodiments are merely some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on these embodiments without inventive effort are within the scope of protection of the present invention. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art can still combine, add to, delete from, or otherwise adjust the features of the various embodiments of the present invention as appropriate without conflict or inventive effort, thereby obtaining different technical solutions that do not fundamentally depart from the concept of the present invention. These technical solutions are also within the scope of protection of the present invention.

Claims

1. A facial micro-expression recognition method based on a dual attention mechanism, characterized in that, The facial micro-expression recognition method based on dual attention mechanism includes: Acquire facial micro-expression images; wherein, the facial micro-expression images are captured through omnidirectional image information capture to obtain facial micro-expression information from different angles; The pre-trained dual attention recognition model is used to process facial micro-expression images and convert the processed facial micro-expression images into image element information of different dimensions, which meets the requirements of AR space three-dimensional modeling. The converted image element information is used to construct a three-dimensional micro-expression model, which is then displayed by outputting stereoscopic data information. Specifically, acquiring facial micro-expression images includes: Data information is extracted from the working process of the facial micro-expression acquisition device. The monitored data information is statistically analyzed from the data information of the captured images. Then, the clearest captured image data is found based on the following function formula. ; Where F(x) is the set of image data acquired by the device; T σ (y) represents the information variable of the acquiring device; y represents the high-definition image set of the acquired image data information; T represents the recognition time of the decomposed image; σ represents the fuzzing and denoising constant of the decomposed monitoring image data information; The converted image element information is used to construct a three-dimensional micro-expression model, specifically including: 1) Form at least one interactive display area in the 3D micro-expression display interface; 2) Obtain image element information from different dimensions; 3) Displaying virtual image element information using the MFC algorithm model; specifically including: Image element information of different dimensions is defined as dimensional features based on a dual attention recognition model; The dimensional features of image data information modules of different dimensions are defined by a dual attention recognition model. The dimensional feature definition is implemented by a convolution algorithm model. Based on a virtual 3D modeling environment, the dynamic demonstration of 3D micro-expression models is achieved through a monitor; The dual-attention recognition model identifies facial micro-expression images based on the content displayed in the 3D micro-expression model, and obtains the recognition results. The results are presented as a dynamic 3D model demonstration and a demonstration of local feature information, depending on the display of the 3D micro-expression model.

2. The facial micro-expression recognition method based on dual attention mechanism as described in claim 1, characterized in that: The acquisition of facial micro-expression images specifically includes: The clearest captured image data is obtained, and the blurry area data information of all captured images is filtered and removed based on the dark channel method to obtain the cleaned image dataset.

3. The facial micro-expression recognition method based on dual attention mechanism as described in any one of claims 1-2, characterized in that: The training method for the dual-attention recognition model specifically includes: Obtain sample data for training the dual attention recognition model, and divide the sample data into training set, test set and validation set in a 3:1:1 ratio; The pre-established dual attention recognition model is trained based on the training set and the preset dual attention mechanism CNN-LSTM model algorithm to obtain the trained dual attention recognition model. Extract the test set and prune the trained dual attention recognition model to obtain multiple pruned dual attention recognition models. The optimal dual attention recognition model is selected from all the pruned dual attention recognition models based on cross-validation.

4. The facial micro-expression recognition method based on dual attention mechanism as described in claim 3, characterized in that: The training method for the dual attention recognition model further includes: A validation set is extracted, and a channel attention mechanism is used to assign weights to the max pooling features and average pooling features in the dual attention recognition model to output image spatial features.

5. The facial micro-expression recognition method based on dual attention mechanism as described in claim 4, characterized in that: The process of extracting a test set to prune the trained dual-attention recognition model yields multiple pruned dual-attention recognition models, specifically including: All data in the test set are input into the initial dual-attention recognition model. After all the data enters the initial decision tree model, the loss of each node in the initial dual-attention recognition model is calculated. Determine whether the loss of each node in the initial dual-attention recognition model is less than a preset threshold, wherein the preset threshold for each node is obtained based on a preset threshold set; if the loss of the current node is less than the preset threshold, stop building the tree for the current node; if the loss of the current node is greater than the preset threshold, adjust the node parameters of the initial decision tree model and prune it again based on the dual-attention mechanism.

6. The facial micro-expression recognition method based on dual attention mechanism as described in any one of claims 1-2, characterized in that: The pre-trained dual-attention recognition model processes facial micro-expression images, specifically including: Obtain the image dataset after clearing; The image is decomposed using the decomposition function of the dual attention recognition model, and the recognition coordinates are marked in the decomposed monitoring image data. The image is decomposed into data, and tolerance calculations are performed on invalid regions in the decomposed data to obtain image element information of different dimensions after processing.

7. The facial micro-expression recognition method based on dual attention mechanism as described in claim 6, characterized in that: The decomposition function is: ; in, The result is the image decomposition, where n is the total number of images in the image set, and i is the index of an image in the image set. σ represents the influence factor of external environmental data information when acquiring images; σ represents the fuzzy denoising constant for blurry areas generated during the information acquisition process.

Images