An AR glasses and a safety assistance method based on AR glasses

Abstract

This invention discloses AR glasses and a safety assistance method based on AR glasses, including: supplementing the environment with supplemental lighting through a supplemental lighting device to facilitate video acquisition by a video acquisition device; processing and identifying the video information acquired by the video acquisition device using an auxiliary recognition module; and stopping projection by a projection module when the host computer or the auxiliary recognition module detects a stop mark on the gloves to avoid obstructing the operator's view. This invention uses both the auxiliary recognition module and the host computer in the backend system to perform dual recognition of video information, effectively reducing interference from AR glasses on users when working in an environment without power interruption, thus improving user safety. Furthermore, the cooperation between the auxiliary recognition module and the host computer enables real-time monitoring and data processing of the work site, further enhancing the system's reliability and practicality.

Description

Technical Field

[0001] This invention belongs to the field of AR glasses technology, and relates to AR glasses and a safety assistance method based on AR glasses. Background Technology

[0002] AR glasses, as a cutting-edge augmented reality device, integrate the real environment with digital information, enabling real-time feedback of images and videos to users through the lenses, providing an immersive interactive experience. With continuous technological advancements, AR glasses are increasingly gaining popularity and demonstrating broad application potential in various fields such as industrial manufacturing and healthcare. The quality of their materials directly affects user comfort and efficiency; therefore, durability, comfort, and practicality must be considered in the design to ensure reliable performance even in harsh industrial environments.

[0003] While AR glasses technology has made some progress, it still faces some challenges in practical applications: current AR glasses on the market are generally quite heavy, and prolonged wear may cause discomfort; the functions of existing AR glasses are relatively limited, making it difficult to meet the needs of complex working environments, such as safety assistance functions in special scenarios like uninterrupted power supply operations; and in some specific working environments, existing AR glasses cannot provide users with sufficient safety guarantees and cannot alert users to potential safety hazards in the external environment. Summary of the Invention

[0004] The purpose of this invention is to solve the problem that existing AR glasses cannot provide users with sufficient safety protection and cannot alert users to potential safety hazards in the external environment, and to provide an AR glasses and a safety assistance method based on AR glasses.

[0005] To achieve the above objectives, the present invention employs the following technical solution:

[0006] An AR glasses system includes: a frame, a lens, a video capture device, an auxiliary recognition module, an optical display element, a fill light device, a microphone, and a nose bridge;

[0007] The frame is connected to the lens; video capture devices are located on both sides of the lens frame, and an auxiliary recognition module is located below the video capture device; the auxiliary recognition module is connected to the video capture device, and the auxiliary recognition module is connected to the projection module, which is located on the lens frame and connected to a host computer; the projection module projects the information sent by the host computer onto the optical display element; the optical display element is located on the lens frame, and a supplementary lighting device is located on one side of the video capture device to provide supplementary lighting for the environment, facilitating video capture; the bridge of the nose is located between the optical display elements; a microphone is located in the bridge of the nose and is used for communication between the background monitoring personnel and the wearer of the glasses.

[0008] A further improvement of the present invention is that:

[0009] Furthermore, the auxiliary recognition module is connected to the video acquisition device via a connecting circuit; the connecting circuit is externally connected to a control circuit; the control circuit controls the connecting circuit to send the video acquired by the video acquisition device to the auxiliary recognition module.

[0010] Furthermore, the video acquisition device and the auxiliary recognition module are simultaneously electrically connected to the host computer. The host computer stores the video information acquired by the video acquisition device. The auxiliary recognition module sends the real-time recognition results to the host computer for storage. The host computer is connected to an alarm device and a microphone. Based on the stored video information, the host computer identifies the stop marker on the glove in the video. If the stop marker is identified, the host computer sends a command to the projection module to stop the projection. The alarm device is used to sound an alarm when the recognition result is a stop marker, reminding the monitoring personnel to forcibly power off the AR glasses.

[0011] Furthermore, the auxiliary recognition module includes a video segmentation module, an image preprocessing module, an image recognition module, and a judgment module. The video segmentation module is used to segment the video into images of different frames. The image preprocessing module performs image denoising, feature enhancement, edge detection, morphological operations, and image segmentation on the segmented images. The image recognition module is used to recognize the preprocessed images. The judgment module determines whether the recognized image is a stop marker on the glove. If so, the judgment module instructs the projection module to stop projecting onto the optical display element to prevent interference with user operations. If not, the recognition continues until a stop marker is obtained.

[0012] Furthermore, the video acquisition device connects to the host computer using WebRTC technology based on the UDP communication protocol, and the auxiliary recognition module uploads the recognition results to the host computer via WebRTC. The video acquisition device includes a first camera 3 and a second camera; the optical display element includes a first AR lens and a second AR lens; the first camera is located on one side of the first AR lens; the second camera is located on the other side of the second AR lens; the auxiliary recognition module includes a first auxiliary recognition module and a second auxiliary recognition module; the first auxiliary recognition module is connected to the first camera; the second auxiliary recognition module is connected to the second camera.

[0013] Furthermore, both the connection circuit and the control circuit are located on the frame; a front top cover is provided on the frame; the front top cover is fastened to the frame, sealing the connection circuit and the control circuit.

[0014] A safety assistance method based on AR glasses includes: when the environment is dim, turning on a supplementary lighting device to provide supplementary lighting; a video acquisition device transmitting the acquired video information to an auxiliary recognition module for auxiliary recognition; when the image recognized by the auxiliary recognition module is a stop mark, the projection module stops projecting onto the optical display element to prevent interference with the user's work; otherwise, recognition continues until a stop mark is obtained; the host computer, based on the stored video information, identifies the stop identifier on the glove in the video; if a stop mark is identified, the host computer sends a command to the projection module to stop projection, and the background alarm device sounds an alarm to remind the monitoring personnel to forcibly power off the AR glasses.

[0015] Furthermore, the video capture device transmits the captured video information to the auxiliary recognition module for auxiliary recognition, specifically as follows:

[0016] The video captured by the video acquisition device is divided into different frames of images;

[0017] Preprocess the segmented image;

[0018] The preprocessed image is input into a trained support vector machine learning model to obtain the recognized image;

[0019] The system determines whether the identified image is a stop sign. If it is, the projection module stops projecting onto the optical display element to prevent interference with user operations. If not, the system continues to identify the image until a stop sign is obtained.

[0020] Furthermore, the segmented image is preprocessed, specifically: image denoising, feature enhancement, edge detection, morphological operations, and image segmentation are performed on the segmented image; bilateral filtering is used for denoising, and histogram equalization algorithm is used to enhance image contrast, thereby extracting the most relevant information from the original image data; edge detection uses the Canny algorithm, morphological operations include dilation and erosion steps, and image segmentation is achieved through the watershed algorithm to eliminate irrelevant details and background noise.

[0021] Furthermore, the trained support vector machine machine learning model is specifically formed by: dividing the image dataset into a training set and a test set; using the training set to learn the model parameters and the test set to evaluate the model's generalization performance; inputting the training set into the support vector machine machine learning model for training, and evaluating the training results based on the test set, until the support vector machine machine learning model reaches the convergence condition, thus obtaining the trained support vector machine machine learning model; the image dataset is obtained by segmenting the video pre-captured by the video capture device.

[0022] The training representation of a support vector machine machine learning model is as follows:

[0023]

[0024] Where w is the weight vector, b is the bias term, and X... i Y is the i-th eigenvector. i These are the corresponding category tags;

[0025] During training, the hinge loss function is used as the optimization objective, and the loss function is minimized through an iterative optimization algorithm until the preset convergence condition is met, thus obtaining the trained support vector machine machine learning model.

[0026] Compared with the prior art, the present invention has the following beneficial effects:

[0027] This invention provides supplemental lighting to the environment through a supplemental lighting device, facilitating video acquisition by the video capture device. An auxiliary recognition module processes and identifies the video information captured by the video capture device. When the host computer or the auxiliary recognition module detects a stop marker on the glove, the projection module stops projecting to avoid obstructing the operator's view. This invention utilizes dual recognition of video information through the auxiliary recognition module and the host computer in the backend system, effectively reducing interference from AR glasses for users operating in uninterrupted power environments and improving user safety. Through the cooperation of the auxiliary recognition module and the host computer, this invention achieves real-time monitoring and data processing of the work site. Wireless network communication enables efficient and stable data transmission between the AR support device and the AR field device, further enhancing the system's reliability and practicality. Attached Figure Description

[0028] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 This is a schematic diagram of the structure of the AR glasses of the present invention;

[0030] Figure 2 This is a schematic diagram of the method for assisting recognition using the auxiliary recognition module of the present invention;

[0031] Figure 3 This is a schematic diagram of the safety assistance method based on AR glasses according to the present invention;

[0032] Figure 4 This is a stop sign diagram.

[0033] Among them, 1-frame; 2-frame; 3-first camera; 4-first auxiliary recognition module; 5-first AR lens; 6-microphone; 7-bridge of nose; 8-second AR lens; 9-second auxiliary recognition module; 10-second camera; 12-front top cover; 13-control circuit; 14-connection circuit. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0035] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0036] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0037] In the description of the embodiments of the present invention, it should be noted that if terms such as "upper," "lower," "horizontal," or "inner" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of the invention is in use, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, terms such as "first" and "second" are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0038] Furthermore, the use of the term "horizontal" does not imply that the component must be absolutely horizontal, but rather that it can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal than "vertical," and does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0039] In the description of the embodiments of the present invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in the present invention according to the specific circumstances.

[0040] The present invention will now be described in further detail with reference to the accompanying drawings:

[0041] See Figure 1 The present invention discloses an AR glasses, including: a frame 1, a frame 2, a video acquisition device, an auxiliary recognition module, an optical display element, a supplementary lighting device, a microphone 6, and a nose bridge 7;

[0042] The frame 2 is connected to the lens rim 1; video capture devices are located on both sides of the lens rim 1, and an auxiliary recognition module is located below the video capture device; the auxiliary recognition module is connected to the video capture device and a projection module, which is located on the lens rim 1 and connected to a host computer; the projection module projects the information sent by the host computer onto the optical display element; the optical display element is located on the lens rim 1, and a supplementary lighting device is located on one side of the video capture device to provide supplementary lighting for the environment, facilitating video capture; the nose bridge 7 is located between the optical display elements; a microphone 6 is located in the nose bridge 7 and is used for communication between the monitoring personnel and the wearer.

[0043] The auxiliary recognition module is connected to the video acquisition device via a connection circuit 14; the connection circuit 14 is externally connected to a control circuit 13; the control circuit 13 controls the connection circuit 14 to send the video acquired by the video acquisition device to the auxiliary recognition module.

[0044] The video acquisition device and the auxiliary recognition module are electrically connected to the host computer. The host computer stores the video information acquired by the video acquisition device. The auxiliary recognition module sends the real-time recognition results to the host computer for storage. The host computer is connected to an alarm device and a microphone. Based on the stored video information, the host computer identifies the stop marker on the glove in the video. If the stop marker is identified, the host computer sends a command to the projection module to stop the projection. The alarm device is used to sound an alarm when the recognition result is a stop marker, reminding the monitoring personnel to forcibly power off the AR glasses.

[0045] The auxiliary recognition module includes a video segmentation module, an image preprocessing module, an image recognition module, and a judgment module. The video segmentation module is used to segment the video into images of different frames. The image preprocessing module performs image denoising, feature enhancement, edge detection, morphological operations, and image segmentation on the segmented images. The image recognition module is used to recognize the preprocessed images. The judgment module determines whether the recognized image is a stop marker on the glove. If it is, the judgment module instructs the projection module to stop projecting onto the optical display element to prevent interference with user operations. If not, the recognition continues until a stop marker is obtained.

[0046] The video acquisition device connects to the host computer using WebRTC technology based on the UDP communication protocol. The auxiliary recognition module uploads the recognition results to the host computer via WebRTC. The video acquisition device includes a first camera 3 and a second camera 10. The optical display elements include a first AR lens 5 and a second AR lens 8. The first camera 3 is located on one side of the first AR lens 5. The second camera 10 is located on the other side of the second AR lens 8. The auxiliary recognition module includes a first auxiliary recognition module 4 and a second auxiliary recognition module 9. The first auxiliary recognition module 4 is connected to the first camera 3. The second auxiliary recognition module 9 is connected to the second camera 10.

[0047] Both the connecting circuit 14 and the control circuit 13 are mounted on the frame 1; a front top cover 12 is mounted on the frame 1; the front top cover 12 is fastened to the frame 1, sealing the connecting circuit 14 and the control circuit 13.

[0048] A safety assistance method based on AR glasses includes: when the environment is dim, turning on a supplementary lighting device to provide supplementary lighting; a video acquisition device transmitting the acquired video information to an auxiliary recognition module for auxiliary recognition; when the image recognized by the auxiliary recognition module is a stop mark, the projection module stops projecting onto the optical display element to prevent interference with the user's work; otherwise, recognition continues until a stop mark is obtained; the host computer, based on the stored video information, identifies the stop identifier on the glove in the video; if a stop mark is identified, the host computer sends a command to the projection module to stop projection, and the background alarm device sounds an alarm to remind the monitoring personnel to forcibly power off the AR glasses.

[0049] See Figure 2 This invention discloses a method for auxiliary recognition module to perform auxiliary recognition, specifically as follows:

[0050] S101, the video captured by the video acquisition device is divided into images of different frames;

[0051] S102, preprocess the segmented image;

[0052] The segmented image is preprocessed, specifically by performing image denoising, feature enhancement, edge detection, morphological operations, and image segmentation. Bilateral filtering is used for denoising, and histogram equalization is used to enhance image contrast, thereby extracting the most relevant information from the original image data. Edge detection uses the Canny algorithm, morphological operations include dilation and erosion steps, and image segmentation is achieved using the watershed algorithm to eliminate irrelevant details and background noise.

[0053] S103, The preprocessed image is input into the trained support vector machine machine learning model to obtain the recognized image;

[0054] The trained support vector machine learning model is specifically formed by: dividing the image dataset into a training set and a test set; using the training set to learn the model parameters and the test set to evaluate the model's generalization performance; inputting the training set into the support vector machine learning model for training, and evaluating the training results based on the test set, until the support vector machine learning model reaches the convergence condition, thus obtaining the trained support vector machine learning model; the image dataset is obtained by segmenting the video pre-captured by the video capture device.

[0055] The training representation of a support vector machine machine learning model is as follows:

[0056]

[0057] Where w is the weight vector, b is the bias term, and x i It is the i-th eigenvector, y i These are the corresponding category tags;

[0058] During training, the hinge loss function is used as the optimization objective, and the loss function is minimized through an iterative optimization algorithm until the preset convergence condition is met, thus obtaining the trained support vector machine machine learning model.

[0059] S104, determine whether the identified image is a stop mark. If it is, the projection module stops projecting onto the optical display element to prevent affecting the user's operation; if not, continue to identify until a stop mark is obtained.

[0060] Example:

[0061] This invention provides AR glasses, including a frame 1, a frame 2, a camera, a video acquisition device, a supplementary lighting device, an auxiliary recognition module, an optical display element, a nose bridge, a front top cover 12, a control circuit 13, and a connection circuit 14. The auxiliary recognition module is connected to the video acquisition device via the connection circuit 14. The connection circuit 14 is externally connected to the control circuit 13. The control circuit 13 controls the connection circuit 14 to send the video acquired by the video acquisition device to the auxiliary recognition module. Both the connection circuit 14 and the control circuit 13 are mounted on the frame 1. The front top cover 12 is mounted on the frame 1, sealing the connection circuit 14 and the control circuit 13. The supplementary lighting device is located on one side of the video acquisition device and is used to supplement the ambient light, facilitating video acquisition by the video acquisition device.

[0062] The outer side of the frame 1 is made of plastic, while the inner side of the frame 2 is a metal support, wrapped in plastic to improve the user's wearing experience. The frame 2 has expansion interfaces on both sides for adding functionality and upgrading firmware. Sufficient space is provided on both sides of the frame 1 to house the first camera 3 and the second camera 10; the auxiliary recognition module and connecting circuit 14 are also included, preventing damage from collisions and minimizing overall weight and deformation from impacts, thus improving the user experience.

[0063] Furthermore, sufficient space is reserved on both sides of the frame 1 to place the first camera 3 and the second camera 10. Through relevant computer network communication protocols, WebRTC technology based on the UDP communication protocol can be used to remotely establish a connection between the wearer and the host computer, transmit video streaming media, transcribe the scene currently viewed by the wearer in real time, and send it to the supporting background system. The auxiliary recognition module receives the video information captured by the first camera and the second camera, and can end the transcription mode when it recognizes the relevant stop mark.

[0064] Furthermore, the first AR lens 5 and the second AR lens 8 can work with the first camera 3 and the second camera 10 to recognize the current scene. At the same time, they can receive video streaming media processed by the backend system through WebRTC technology to mark and select specific items, helping to remind the wearer of points to pay attention to during work and prevent misoperation or dangerous behavior.

[0065] Furthermore, the auxiliary recognition module embeds computer vision processing algorithms, utilizing the open-source OpenCV library to perform image processing tasks. These tasks include, but are not limited to, image denoising, feature enhancement, edge detection, morphological operations, and image segmentation. In the image preprocessing stage, bilateral filtering is used for denoising, while histogram equalization enhances image contrast, thereby extracting the most relevant information from the original image data. Edge detection employs the Canny algorithm, morphological operations include dilation and erosion steps, and image segmentation is achieved through the watershed algorithm to eliminate irrelevant details and background noise, improving the accuracy and speed of image recognition.

[0066] To achieve efficient image classification and recognition, this invention employs a Support Vector Machine (SVM) machine learning model, implemented through the LIBSVM library, which provides an easy-to-use interface for rapid integration into auxiliary recognition modules. During SVM model training, the image dataset is divided into training and test sets. The training set is used to learn the model parameters, while the test set is used to evaluate the model's generalization performance. The training process involves solving a quadratic programming problem: maximizing the margin between two classes of data points while minimizing the classification error. Specifically, SVM training involves solving the following quadratic programming problem:

[0067]

[0068] Where w is the weight vector, b is the bias term, and x i It is the i-th eigenvector, y i These are the corresponding category labels.

[0069] During training, the hinge loss function is used as the optimization objective, and the loss function is minimized through iterative optimization algorithms, such as Sequence Minimum Optimization (SMO), until the preset convergence conditions are met, such as reaching a preset number of iterations or the rate of decrease of the loss function is lower than a predetermined threshold.

[0070] LIBSVM's interface allows system designers to directly call pre-trained SVM models without complex programming, thus simplifying the development process. SVM models trained in the OpenCV environment can be exported as XML text files, enabling reuse on different platforms and devices, including resource-constrained embedded systems. In real-time applications, AR glasses can read XML parameter files through a programming interface and use these parameters for real-time image recognition and classification inference, ensuring rapid response even in dynamic environments.

[0071] Specifically, when the AR glasses detect a STOP sign in the user's field of vision based on the above algorithm, the auxiliary recognition module immediately triggers a series of safety measures. These measures may include issuing a warning sound to the user, displaying a stop signal on the AR glasses' screen, or activating an emergency stop command in the remote assistance system. This instant feedback mechanism effectively prevents the user from continuing potentially harmful operations in potentially dangerous situations. Furthermore, after recognizing the stop sign and activating the relevant auxiliary functions, the system can automatically turn off the AR glasses' display optics to prevent user distraction when full attention is required to focus on the external environment. This safety feature is particularly important in scenarios where users may need to respond immediately to external emergencies, such as avoiding obstacles.

[0072] Furthermore, a microphone 6 is installed in the middle of the frame 1, which allows the wearer to communicate with back-end staff in real time when the remote assisted recognition function is activated, preventing dangerous or unnecessary operations by the staff wearing the glasses.

[0073] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A safety assistance method based on AR glasses, characterized in that, include: When the environment is dim, turn on the supplementary lighting device to provide additional illumination. The video acquisition device transmits the acquired video information to the auxiliary recognition module for auxiliary recognition. When the image recognized by the auxiliary recognition module is a stop mark, the projection module stops projecting onto the optical display element to prevent interference with the user's work; otherwise, it continues to recognize until a stop mark is obtained. The host computer identifies the stop identifier on the glove in the video based on the stored video information. If a stop mark is recognized, the host computer sends a command to the projection module to stop projection, and the alarm device in the background sounds an alarm to remind the monitoring personnel to forcibly power off the AR glasses. Among them, the safety assistance method for AR glasses is applied to AR glasses, which include: a frame (1), a frame (2), a video acquisition device, an auxiliary recognition module, an optical display element, a supplementary light device, a microphone (6), and a nose bridge (7). The frame (2) is connected to the lens frame (1); the video acquisition device is set on both sides of the lens frame (1), and the auxiliary recognition module is set below the video acquisition device; the auxiliary recognition module is connected to the video acquisition device, the auxiliary recognition module is connected to the projection module, the projection module is set on the lens frame (1), and the projection module is connected to the host computer; the projection module projects the information sent by the host computer onto the optical display element; the optical display element is set on the lens frame (1), the supplementary lighting device is set on one side of the video acquisition device, and the supplementary lighting device is used to supplement the ambient light to facilitate video acquisition by the video acquisition device; the bridge of the nose (7) is set between the optical display elements; the microphone (6) is set in the bridge of the nose (7), and the microphone (6) is used for the background monitoring personnel to communicate with the wearer of the glasses; The auxiliary recognition module includes a video segmentation module, an image preprocessing module, an image recognition module, and a judgment module. The video segmentation module is used to segment the video into images of different frames. The image preprocessing module performs image denoising, feature enhancement, edge detection, morphological operations, and image segmentation on the segmented images. The image recognition module is used to recognize the preprocessed images. The judgment module determines whether the recognized image is a stop marker on the glove. If it is, the judgment module instructs the projection module to stop projecting onto the optical display element to prevent interference with the user's work. If not, the recognition continues until a stop marker is obtained. The video acquisition device is connected to the host computer based on WebRTC technology of UDP communication protocol. The auxiliary recognition module uploads the recognition result to the host computer through WebRTC. The video acquisition device includes a first camera (3) and a second camera (10). The optical display element includes a first AR lens (5) and a second AR lens (8). The first camera (3) is set on one side of the first AR lens (5). The second camera (10) is set on the other side of the second AR lens (8). The auxiliary recognition module includes a first auxiliary recognition module (4) and a second auxiliary recognition module (9). The first auxiliary recognition module (4) is connected to the first camera (3). The second auxiliary recognition module (9) is connected to the second camera (10).

2. The safety assistance method based on AR glasses according to claim 1, characterized in that, The auxiliary recognition module is connected to the video acquisition device via a connection circuit (14); the connection circuit (14) is connected to an external control circuit (13); the control circuit (13) controls the connection circuit (14) to send the video acquired by the video acquisition device to the auxiliary recognition module.

3. The safety assistance method based on AR glasses according to claim 2, characterized in that, The video acquisition device and the auxiliary recognition module are both electrically connected to the host computer, and the host computer stores the video information acquired by the video acquisition device. The auxiliary recognition module sends the real-time recognition results to the host computer for storage. The host computer is connected to the alarm device and the microphone. Based on the stored video information, the host computer identifies the stop marker on the glove in the video. If the stop marker is identified, the host computer sends a command to the projection module to stop the projection. The alarm device is used to sound an alarm when the recognition result is a stop marker, reminding the monitoring personnel to forcibly power off the AR glasses.

4. The safety assistance method based on AR glasses according to claim 3, characterized in that, The connection circuit (14) and the control circuit (13) are both set on the frame (1); a front top cover (12) is set on the frame (1); the front top cover (12) is fastened on the frame (1) to seal the connection circuit (14) and the control circuit (13).

5. The safety assistance method based on AR glasses according to claim 4, characterized in that, The video acquisition device transmits the acquired video information to the auxiliary recognition module for auxiliary recognition, specifically: The video captured by the video capture device is divided into different frames of images; Preprocess the segmented image; The preprocessed image is input into the trained support vector machine machine learning model to obtain the recognized image; The system determines whether the identified image is a stop sign. If it is, the projection module stops projecting onto the optical display element to prevent interference with user operations. If not, the system continues to identify the image until a stop sign is obtained.

6. The safety assistance method based on AR glasses according to claim 5, characterized in that, The preprocessing of the segmented image specifically includes: image denoising, feature enhancement, edge detection, morphological operations, and image segmentation; bilateral filtering is used for denoising, and histogram equalization is used to enhance image contrast, thereby extracting the most relevant information from the original image data; edge detection uses the Canny algorithm, morphological operations include dilation and erosion steps, and image segmentation is achieved through the watershed algorithm to eliminate irrelevant details and background noise.

7. The safety assistance method based on AR glasses according to claim 6, characterized in that, The trained support vector machine learning model is specifically defined as follows: the image dataset is divided into a training set and a test set; the training set is used to learn the model parameters, and the test set is used to evaluate the generalization performance of the model; the training set is input into the support vector machine learning model for training, and the training results are evaluated based on the test set, until the support vector machine learning model reaches the convergence condition, thus obtaining the trained support vector machine learning model; the image dataset is obtained by segmenting the video pre-captured by the video capture device. The training representation of a support vector machine machine learning model is as follows: Where w is the weight vector and b is the bias term. It is the i-th eigenvector. These are the corresponding category tags; During training, the hinge loss function is used as the optimization objective, and the loss function is minimized through an iterative optimization algorithm until the preset convergence condition is met, thus obtaining the trained support vector machine machine learning model.

Images