Induction control glasses for human body signal recognition and induction control method

Abstract

The invention discloses an inductive control method for human signal recognition, which is applied to inductive control glasses for human signal recognition. The inductive control glasses for human signal recognition include a glasses frame detachably connected by temples and frames, a control center and a signal capture module. The inductive control method includes: step S1, temples contact skin, generate electrical signals and wake up glasses; step S2, signal capture module captures human signals and transmits the signals to the control center; step S3, control center preprocesses human signals; step S4, status judgment is performed on the preprocessed signals, and glasses functions are automatically adjusted according to the current status of the user. The inductive control method cooperates with glasses to optimize user experience and greatly improve the function of human-computer interaction.

Description

Technical Field

[0001] The present invention relates to the technical field of glasses, and in particular to a pair of induction control glasses for human body signal recognition and an induction control method. Background Art

[0002] Glasses are simple optical devices made to correct vision or protect the eyes. They are used to improve vision, protect the eyes or for decorative purposes. Glasses can correct a variety of vision problems. Glasses can also be used as a tool for vision training. They conduct scientific physical exercises and training for the eyes and visual system, enhance the eye's ability to adjust, help prevent and control myopia, repair amblyopia, improve the health of naked eye vision, and reduce dependence on glasses, surgery and other vision correction methods. With the popularization of intelligence, smart glasses, as an important product, have become a hot spot for research and application.

[0003] Smart glasses currently rely mainly on display technologies such as optical waveguides, micro-projectors, and OLEDs, which can realize functions such as augmented reality (AR) experience, virtual reality (VR) support, voice control, image and environment recognition. They integrate a variety of sensors, such as eye tracking, accelerometers, gyroscopes, etc., which can realize head posture, gesture recognition and real-time environmental perception. At the same time, smart glasses usually have wireless connection capabilities, such as Bluetooth, Wi-Fi, 5G, etc., can exchange data with smartphones, computers and other devices, and support independent communication functions. However, the existing interaction methods still have limitations. Smart glasses and electronic accessories need to be connected through wired methods such as data cables, and manual operations are required to achieve the purpose of using smart glasses. The user experience in human-computer interaction is poor.

[0004] Therefore, it is necessary to propose a new sensor control glasses and a sensor control method for human signal recognition to solve the defects of poor human-computer interaction function in the prior art. Summary of the invention

[0005] The purpose of the present invention is to provide a human body signal recognition induction control glasses and an induction control method to solve the problem of poor human-computer interaction function in the prior art.

[0006] To achieve this object, the present invention adopts the following technical solutions: A human body signal recognition inductive control method is applied to a human body signal recognition inductive control glasses, the human body signal recognition inductive control glasses comprising a glasses frame detachably connected by temples and frames, a control center and a signal capture module, the inductive control method comprising: Step S1, the temples touch the skin, generating an electrical signal and waking up the glasses; Step S2: The signal capture module captures human body signals and transmits the signals to the control center; Step S3, the control center pre-processes the human body signal; Step S4: determine the status of the preprocessed signal and automatically adjust the glasses function according to the current status of the user.

[0007] Furthermore, before step S1, the following steps are also included: Step S0: pre-input one or more sets of user's sclera feature information as an unlocking key, and set corresponding permissions.

[0008] Furthermore, the signal capture module includes an acceleration sensor and an image sensor, and the step S2 further includes the following steps: Step S21, capturing a human face signal, a motion signal, and a blinking signal.

[0009] Furthermore, the step S2 further includes the following steps: Step S211: The image sensor captures the user's sclera signal and transmits it to the control center; Step S212: The control center extracts characteristic information from the sclera signal according to the image processing algorithm, and performs data processing on the characteristic information to obtain a first signal; Step S213, performing feature comparison on the first signal; Step S214: If the calculation result is lower than the set threshold, it is determined to be a match, and it is considered that this feature corresponds to an identity in the database, and the glasses are unlocked at the same time; Step S215: At predetermined time intervals, the image sensor and the acceleration sensor continuously capture human body signals and transmit the signals to the control center.

[0010] Furthermore, the step S3 includes the following steps: Step S31, filtering and invalid data processing; Step S32: linear interpolation; Step S33, filtering and denoising; Step S34: Standardization processing.

[0011] Furthermore, the linear interpolation formula used in step S32 is as follows: ; in, : The data point to be interpolated, the known data point on its left is , the known data points on the right are ; : The missing value after interpolation, the known data point on the left is , the known data points on the right are .

[0012] Furthermore, the formula used for filtering and denoising in step S33 is as follows: ; in: is the cut-off frequency, is the passband edge frequency.

[0013] Furthermore, the standardization in step S34 is mean-variance standardization, and its formula is as follows: ; in, is the mean of the input sample x, is the standard deviation of the input sample x.

[0014] A pair of induction-controlled glasses for human signal recognition, comprising a glasses frame detachably connected by temples and a frame, a control center, and a signal capture module, wherein the control center further comprises: A repository for pre-inputting scleral feature information of more than one group of users as unlocking keys and setting corresponding permissions; and for setting environmental information corresponding to instructions for mobilizing various structures of the glasses; A data processing center, used for processing the information from the signal capture module; A verification module is used to make a similarity match judgment based on the collected information and the information in the repository; The authorization module is used to open the corresponding permissions of the glasses and make corresponding adjustments based on the judgment results and the information recorded in the database.

[0015] A pair of induction control glasses for human body signal recognition includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, the steps of the above-mentioned induction control method are implemented.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The glasses can be easily awakened by touching the skin with the temples, and the sclera feature unlocking is highly secure and protects privacy. It can capture a variety of human body signals and intelligently adjust the functions of the glasses based on them to enhance the user experience. It performs multi-step pre-processing of human body signals to improve data quality and ensure accurate function adjustment. The control center modules are divided into two parts for easy maintenance and expansion. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0018] The structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with this technology. They are not used to limit the conditions under which the present invention can be implemented, and therefore have no substantive technical significance. Any structural modification, change in proportion or adjustment of size, without affecting the effects and purposes that can be achieved by the present invention, should still fall within the scope of the technical contents disclosed by the present invention.

[0019] Figure 1 A flow chart of an embodiment of a sensing control method for human body signal recognition of the present invention; Figure 2 A flowchart of human activity recognition of an embodiment of a human signal recognition induction control method of the present invention; Figure 3 A schematic diagram of sclera information collection of an embodiment of the induction control glasses for human signal recognition of the present invention; Figure 4 The figure is a schematic structural diagram of an embodiment of the inductive control glasses for human body signal recognition of the present invention.

[0020] Illustration description: 100, glasses frame; 200, signal capture module; 300, control center; 400, battery; 110, temple; 120, glasses frame. DETAILED DESCRIPTION

[0021] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] In the description of the present invention, it should be understood that the terms "upper", "lower", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present invention. It should be noted that when a component is considered to be "connected" to another component, it may be directly connected to the other component or there may be a centrally arranged component at the same time.

[0023] The technical solution of the present invention is further described below with reference to the accompanying drawings and through specific implementation methods.

[0024] Embodiment 1: The present invention provides an inductive control method for human body signal recognition to solve the problem of poor human-computer interaction. A signal capture module is set to capture human body signals, and then the glasses are adjusted according to preset rules to achieve coordination between multiple functions and maximize the user experience.

[0025] Reference Figure 1 , a flow chart of an embodiment of a human signal recognition inductive control method of the present invention is shown, which is applied to a human signal recognition inductive control glasses, the human signal recognition inductive control glasses comprising a glasses frame 100 detachably connected by temples 110 and frames 120, a control center 300 and a signal capture module 200, the inductive control method comprising: Step S1, the temple 110 contacts the skin, generates an electrical signal and wakes up the glasses; Step S2: the signal capture module 200 captures human body signals and transmits the signals to the control center 300; Step S3, the control center 300 pre-processes the human body signal; Step S4: determine the status of the preprocessed signal and automatically adjust the glasses function according to the current status of the user.

[0026] It should be noted that as a power-on function, the part that contacts the human skin is made of conductive material and connected to the entire glasses. The principle of this function is that when different materials contact the skin, due to the different binding abilities of the atoms on the material and the surface of the skin to electrons, at the moment of contact, the electrons will be transferred from the party with weaker binding ability to the party with stronger binding ability. Due to this feature, when the user wears the glasses to the ears and the glasses touch the skin, an electrical signal will be quickly generated. This signal will serve as a power-on command and wake up the glasses. There is no need to manually control the switch, further realizing human-computer interaction and greatly improving the user experience. It should be added that at this time it only serves as a switch, and further unlocking is required for use, similar to the lock screen password set on the mobile phone after powering on. You need to enter the pre-set password for further use.

[0027] Furthermore, the signal capture module 200 captures human body signals, such as human sclera signals, blink signals, and motion signals. Various types of sensors need to be set on the glasses to collect human body signals. As a more preferred embodiment, there is an acceleration sensor to sense the motion state of the human body, corresponding to the states of motion and rest; there is also an image sensor to capture human sclera signals, blink signals, and environmental signals; there is also a temperature sensor to sense human body temperature. The above signals generally correspond to the user's usage status.

[0028] Furthermore, the user's usage status includes, but is not limited to, states such as exercise, rest, and work (or concentration), which can be preset and input with the corresponding values ​​of each signal. Taking the acceleration sensor as an example, as a preferred embodiment, the acceleration sensor can be a gravity sensor. It should be added that the gravity sensor mainly utilizes the characteristic of crystal deformation caused by acceleration inside it. Since this deformation will generate voltage, as long as the relationship between the generated voltage and the applied acceleration is calculated, the acceleration can be converted into a voltage output. Of course, there are many other methods to make acceleration sensors, such as capacitance effect, thermal bubble effect, and optical effect, but the most basic principle is that the acceleration generates a deformation of a certain medium, and the deformation is measured and converted into a voltage output by a related circuit. Then, the corresponding relationship between voltage and acceleration is used to obtain the acceleration corresponding to the output voltage, thereby obtaining acceleration data.

[0029] When the acceleration signal output by the gravity sensor remains relatively stable for a period of time and its value is close to the local gravity acceleration (approximately ), it can be determined that the user is in a stationary state; when walking, the human body will produce regular up and down and forward and backward swings; the acceleration signal detected by the gravity sensor will show periodic changes, and there will be obvious fluctuations in the vertical and horizontal directions. By analyzing the frequency and amplitude range of the signal, the walking state can be identified. Generally speaking, the walking frequency is about 0.8-2Hz. Depending on the stride and speed, the acceleration amplitude will also be different; when going up and down stairs, in addition to the horizontal movement, the vertical acceleration changes more significantly. When going up the stairs, the vertical acceleration is positive at the beginning of the rising stage after taking a step, and the amplitude is When going down the stairs, the vertical acceleration is negative at the beginning of the descending stage after taking a step, and the amplitude is also large. Combined with the change pattern of the horizontal and vertical acceleration signals, the state of going up and down the stairs can be effectively identified; when riding in a car, train or other means of transportation, the gravity sensor will detect complex acceleration signals due to the movement of the vehicle. Unlike the movement of the human body, the acceleration changes when riding in a vehicle are usually smoother and may contain low-frequency vibration signals, such as the low-frequency vibration caused by bumps on the road when the car is driving. By analyzing the frequency components and change trends of the signals, it is possible to distinguish between riding in a vehicle and other motion states.

[0030] Furthermore, before step S1, the following steps are also included: Step S0: pre-input one or more sets of user's sclera feature information as an unlocking key, and set corresponding permissions.

[0031] It should be noted that this signal is used as an unlocking key and is compared with the user's signal. If they are consistent, the various functions of the glasses are unlocked. If they are inconsistent, the various functions of the glasses, such as playing music via Bluetooth, making phone calls, and keeping the temples 110 close to the skin during exercise, are unavailable. Face recognition builds a security line for access to data and functions of smart glasses. Only authorized users can unlock and use all functions to prevent data leakage and device abuse. Even if the glasses are accidentally lost, others cannot obtain the information therein.

[0032] Furthermore, the signal capture module 200 includes an acceleration sensor and an image sensor, and the step S2 further includes the following steps: Step S21, capturing a human face signal, a motion signal, and a blinking signal.

[0033] It should be noted that the acceleration sensor is used to capture the motion signals of the human body. The gravity sensor has been used as an example in the previous article, so it will not be repeated here. Before unlocking, the image sensor is used to capture the user's sclera signal and compare it with the pre-stored sclera signal. After unlocking, the user's blinking signal is captured at regular intervals.

[0034] Furthermore, the step S2 further includes the following steps: Step S211: The image sensor captures the user's sclera signal and transmits it to the control center 300; Step S212: the control center 300 extracts characteristic information from the sclera signal according to the image processing algorithm, and performs data processing on the characteristic information to obtain a first signal; Image features of the sclera typically include the following: Vascular distribution: The vascular texture of the eye is highly variable among individuals.

[0035] Texture pattern: The texture details of the scleral area, especially the vascular structure, local grayscale changes, etc., are important identity features.

[0036] Local contrast, texture roughness, etc.: Slight changes in lighting and texture roughness may also vary between individuals.

[0037] It is more preferred to use a deep convolutional neural network image processing model (CNN) as an image processing algorithm to extract the sclera and obtain a feature vector, which is a point in a high-dimensional space and contains unique information of the sclera area.

[0038] Step S213, performing feature comparison on the first signal; The Euclidean distance formula is preferably used as the comparison standard, which calculates the straight-line distance between two feature vectors. The formula is as follows: ; in: and are the first signal and the feature vector of the first signal (sclera feature) respectively.

[0039] and is the i-th element in the eigenvector (i.e., the specific eigenvalue of the sclera image).

[0040] The closer the values ​​of the eigenvectors are, the greater the Euclidean distance The smaller the distance, the higher their similarity. If the distance is lower than the set threshold (0.5), the two are considered to be the same identity. Otherwise, they are judged to be different identities.

[0041] Step S214: If the calculation result is lower than the set threshold, it is determined to be a match, and it is considered that this feature corresponds to an identity in the database, and the glasses are unlocked at the same time; Step S215 : The image sensor and the acceleration sensor continuously capture human body signals at predetermined time intervals, and transmit the signals to the control center 300 .

[0042] More preferably, the predetermined time interval means that the image sensor and the acceleration sensor capture the human body signal every one second.

[0043] Furthermore, the image sensor continuously monitors the user's blinking signals according to the user's line of sight. The blinking signal is used as one of the biofeedback signals to control different functions of the glasses. The user's usage status is judged based on the frequency of the user's blinking. For example, more frequent blinking may mean that the user is processing information or feeling tired. The system can make appropriate adjustments based on these signals (such as adjusting brightness, turning on blue light filtering, etc.). It should be added that mild fatigue is when the number of blinks reaches 20-25 times per minute, moderate fatigue is when the number of blinks reaches 25-30 times per minute, and severe fatigue is when the number of blinks exceeds 30 times per minute. As a more preferred embodiment, when mild fatigue is reached, for myopic users, on the basis of the original correction degree, the myopia degree is temporarily and slightly reduced (such as reducing it by 25-50 degrees) to put the eyes in a relatively relaxed adjustment state. The duration can be set to 15-20 minutes. The proportion of filtered blue light can also be increased to 40%-45% to reduce the stimulation of blue light to the eyes, and a gentle reminder is issued to suggest users to take a proper rest. When moderate fatigue is reached, the intensity of high-saturation colors is appropriately reduced to make bright colors such as red and yellow look less dazzling, while enhancing the perception of colors such as green that have a soothing effect on the eyes. To make the visual experience more comfortable, the polarization angle can also be fine-tuned to a direction that is more suitable for the current ambient light, thereby enhancing the filtering effect of the polarized lenses on horizontally reflected light. At this time, the reminder frequency increases, with a voice reminder every 10-15 minutes. When severe fatigue is reached, the lenses are switched to a completely blue light protection mode to almost completely filter out blue light. At this time, the scenery seen by the user will appear yellowish in tone, and the diffuse reflection filter function is activated. Through special optical materials, the light is diffusely reflected in the lenses, making the light entering the eyes more uniform and soft, similar to forming a soft light curtain in front of the eyes, reducing the direct stimulation of light to the eyes. At this time, a small current is sent from time to time to the part of the temple 110 that contacts the human body (this current does not harm the human body, but it is perceptible, similar to static electricity) to remind the user that they need to rest.

[0044] This embodiment provides an induction control method for human body signal recognition, which has the following effects: The glasses are awakened by generating an electrical signal when the temple 110 contacts the skin, without the need for manual operation of the switch, which improves the convenience and experience of the user. This touch switch design combined with the application of conductive materials further enhances the automation function of smart glasses, reduces the use of traditional physical buttons or manual switches, and makes the opening and use of glasses more intuitive and natural; through the signal capture module 200 (such as an accelerometer, an image sensor, and a temperature sensor, etc.), the system can monitor various physiological signals of the human body in real time, and can identify the user's movement state (such as walking, going up and down stairs, standing still, etc.), eye state (such as blinking) and environmental changes (such as temperature changes), and can also perform secondary unlocking. This precise recognition enables the glasses to automatically adjust their functions to adapt to different usage scenarios and needs, greatly improving the user's comfort and experience. The induction control glasses with human signal recognition realize more intelligent, convenient and personalized human-computer interaction, and can automatically adjust the functions of the glasses according to the user's physiological state and activity, thereby maximizing the user's wearing experience and usage efficiency.

[0045] Embodiment 2: Combination Figure 2 , shows a flow chart of human activity recognition of a human signal recognition induction control method according to an embodiment of the present invention, wherein step S3 includes the following steps: Step S31, filtering and invalid data processing; Step S32: linear interpolation; Step S33, filtering and denoising; Step S34: Standardization processing.

[0046] It should be added that the research process of human activity recognition based on wearable devices is as follows: Sensor data collection. Wearable devices (sensors) are used to collect the subject’s motion data, including acceleration, angular velocity, heart rate, and air pressure, which are real-time data from sensors such as accelerometers, gyroscopes, and heart rate monitors embedded in wearable devices.

[0047] Data preprocessing: There may be noise, missing values, and outliers in the raw data collected by the sensor, so the data needs to be preprocessed to improve the generalization ability of the model.

[0048] Feature extraction and selection. The goal of feature extraction is to convert the original time series sensor data (such as acceleration, angular velocity) into meaningful feature representations so that the classification model can efficiently learn patterns. It should be added that in the process of human activity recognition, data is usually continuously collected by accelerometers, gyroscopes and other sensors to form time series signals. Since these signals are continuous time series data, directly inputting the classification model will result in high computational complexity and difficulty in identifying patterns. Therefore, we use sliding windows for data segmentation, and each window is used as a sample for feature extraction. For example, the moment of time 0 is a window, and its corresponding acceleration and angular velocity are the features corresponding to the window. Similarly, the moment of time 20ms is also a window. The size of the window (that is, the duration of the window) can be set according to actual needs, and 2s is more preferred. Then, time domain feature extraction is performed, such as mean, standard deviation, root mean square and energy features. The calculation formula is the existing technology, so it will not be elaborated here. Then, frequency domain features are extracted. The time series signal is converted from the time domain to the frequency domain through Fourier transform (FFT), and the following features are extracted: main frequency, power spectrum entropy and frequency mean (Fourier transform and feature extraction are well-known technologies for people in this industry, so I won’t go into details here). After extracting features, PCA (Principal Component Analysis) and correlation analysis are used for feature selection to reduce the amount of calculation, select effective features, and improve the classification effect. If the correlation between feature A and feature B is 0.95, it means that their change patterns are almost the same. You can consider removing one of the features because they are redundant to some extent. PCA can help us convert multi-dimensional data into several key principal components, thereby reducing the dimension of the data, retaining the main information, and reducing redundant features. Before using PCA, you can first remove those highly correlated features through correlation analysis to further reduce redundant information and make the data more concise. In summary, the selected feature information can be obtained.

[0049] Classifier. Use the selected classifier (such as SVM, random forest, deep learning model, etc.) for training, and practice with a large amount of data to enable the classifier to have the ability to identify human activities. Preferably, for SVM (Support Vector Machine), select a suitable kernel function and adjust the C parameter; for random forest, adjust the number and depth of trees; for deep learning models, use cross-validation to adjust hyperparameters (such as learning rate, number of layers, number of units, etc.).

[0050] Activity recognition. Finally, the trained model is installed on the wearable device and used in the actual activity recognition task to classify and predict new unknown data, and to identify and label different human activities. The conventional process can basically complete the task of human activity recognition relatively well, but the details of each process module require more extensive and in-depth research so that the human behavior recognition technology based on wearable devices can be better applied in actual scenarios. The collected raw data comes directly from the sensor and may contain noise, outliers or invalid data. Generally, preprocessing operations are required, including data cleaning, noise filtering, data calibration, etc., to remove data points that do not meet the standards or interfere with the results, and then the data is standardized to ensure the accuracy and reliability of the data.

[0051] Furthermore, the purpose of filtering and invalid data processing is to remove the data collected within a certain period of time before and after the activity of each subject, as well as the invalid data recorded. The subject may be in an unstable state before and after the activity, such as preparing to start the activity or the activity has ended but the body is still in the process of gradually calming down. The data during these transition periods may not be representative and may contain more noise or signals that are not related to normal activities. Therefore, filtering out the data during these transition periods can improve the stability and accuracy of the training model. At the same time, the start and end times of the activities of different subjects may be different. If the data during the transition period before and after are not removed, the data of different subjects will be inconsistent when training the model. This may cause the model to be too dependent on the data during the transition period and fail to truly learn the characteristics of different activities themselves. By removing the data during the transition period before and after, the data of different subjects can be kept consistent, making the training model more generalizable. Moreover, in the task of human activity recognition, it is usually necessary to annotate the data and associate each time period with the corresponding activity category. The data during the transition period before and after may be difficult to accurately annotate because the start and end boundaries of the activity are not clear. Removing the data during these transition periods can simplify the annotation process and improve the consistency and accuracy of the annotation.

[0052] In summary, removing the data within a certain period of time before and after each subject can improve the stability, consistency and accuracy of the training model, ensure the quality and reliability of the training data, and thus improve the performance and accuracy of human activity recognition. This is a very necessary operation, including but not limited to removing the data within 2 seconds before and after the test.

[0053] Furthermore, linear interpolation is a commonly used data interpolation method that can be used to fill missing values ​​in data or perform data smoothing. It is a commonly used interpolation method in human behavior recognition. The data collected by the sensor may have some null values ​​or excessively large abnormal values ​​due to factors such as the sensor itself or the environment. It is necessary to first find these abnormal data based on the threshold and delete these abnormal values. Finally, these null values ​​will be filled in by interpolation based on their previous and subsequent values. The following is a simple linear interpolation method: Suppose you want to perform linear interpolation on one-dimensional data, where some data points are missing or need to be smoothed. First, determine the location of the missing values ​​in the data, that is, the data points that need to be interpolated. Second, find adjacent data points; for each missing value, find its adjacent known data points. Usually, the nearest known data points on the left and right are selected.

[0054] Furthermore, the linear interpolation formula used in step S32 is as follows: ; in, : The data point to be interpolated, the known data point on its left is , the known data points on the right are ; : The missing value after interpolation, the known data point on the left is , the known data points on the right are .

[0055] It should be noted that the previous step is repeated for all missing values, and interpolation is performed one by one until all data are traversed. It should also be noted that the linear interpolation method assumes that the data has a linear relationship near the missing values, which is suitable for data interpolation in some cases, but not in all cases. In cases where some data has nonlinear trends or outliers exist, other interpolation methods may be more appropriate. In addition, for linear interpolation of multidimensional data, the above method can be extended to multidimensional cases by finding nearby data points and using the multidimensional linear interpolation formula for calculation.

[0056] Furthermore, after the previous steps, the data collected by the sensor has been free of outliers, but still contains noise signals, so the data needs to be denoised. The noise signal and the real useful signal are mixed together in an additive or multiplicative manner (corresponding to additive noise and multiplicative noise, respectively), which will have a certain impact on the performance of the entire human activity recognition task. These noises come from the activities of the subjects or from the test environment, and almost inevitably enter the data collected by the terminal. Although some wearable devices have some filters themselves, their effects are still limited. In general conventional human behavior recognition research, researchers usually use traditional filters to process the collected signals, such as low-pass filtering, median filtering, sliding average filtering, and wavelet transform. The Butterworth low-pass filter, which is used more frequently, is used as an example. The Butterworth low-pass filter is used to remove high-frequency components in the signal and retain low-frequency components. It is a non-attenuation filter, that is, it has a flat frequency response in the passband. The design of the Butterworth low-pass filter is based on the Butterworth filter function, and the principle is as follows: The formula used for filtering and denoising in step S33 is as follows: ; It has the following features: 1. Flat passband response. The Butterworth low-pass filter has a flat frequency response within the passband, that is, it does not introduce additional amplitude-frequency response fluctuations.

[0057] 2. Smooth cutoff frequency transition. The cutoff frequency of the Butterworth low-pass filter refers to the frequency at which the filter begins to reduce the amplitude-frequency response. It has a smooth transition characteristic without steep edges.

[0058] 3. No attenuation slope: The attenuation rate of the Butterworth low-pass filter at the cutoff frequency is infinite, which means that after the cutoff frequency, the filter completely blocks high-frequency signals.

[0059] Furthermore, the collected signals have been processed through invalid data processing, outlier removal, interpolation, and denoising to obtain basically usable data. In sensor-based human activity recognition, in order for the classification and recognition algorithm model to better learn the characteristics of sample data, data standardization is an important preprocessing step. The purpose of standardization is to eliminate scale differences in the data, make data from different sensors comparable, and help improve the accuracy and reliability of activity recognition.

[0060] Furthermore, the standardization in step S34 is mean-variance standardization, and its formula is as follows: ; in, is the mean of the input sample x, is the standard deviation of the input sample x. Data standardization can improve the performance and reliability of sensor-based human activity recognition systems, enable the model to better understand and distinguish different activity patterns, and improve its adaptability to unknown data.

[0061] Furthermore, the sensor data after the data preprocessing process can be used for research, but in the task of human activity recognition, action segmentation is a necessary step. It generally occurs after data preprocessing and before algorithm model training. When the sensor collects the activity data of the subject, the data is continuously generated according to the rated minimum timestamp of the sensor, and the receiving end obtains continuous activity data with a certain interval according to the set sampling frequency. Taking the sampling frequency of 50Hz as an example, the receiving end obtains a piece of data in 1 / 50 seconds (that is, 50 pieces of data in 1 second); when preparing the human activity data set, each of these data is labeled with the corresponding action, and each piece of data is part of the action (according to human activity habits, the general action will last for 2~3 seconds), but cannot represent the entire action. The goal of action segmentation is to segment these continuous motion data into action samples in the form of windows according to human activity habits.

[0062] Action segmentation divides the continuous sensor data sequence into discrete action segments for better feature extraction and model training, and helps improve the accuracy and generalization ability of the activity recognition model. Action segmentation is generally completed in a window manner, setting a certain repetition rate (the purpose of this is to ensure the continuity of the action), and a fixed window (determined according to human activity habits) slides back continuously on the data, and extracts the data in the window as a sample until the entire data sequence is segmented. This method is also called sliding window segmentation. Taking the sampling frequency of 50Hz as an example, the approximate duration for a person to complete the entire action is 2.56s, 50Hz×2.56s = 128, that is, a single action sample contains 128 data; 128×50% = 64, the repetition rate is 64 data, that is, a fixed window with a length of 128 slides back once every 64 data to segment a sample. It can divide the continuous sensor data sequence into multiple independent samples, so that the model can better capture and identify action patterns in different time periods. By adjusting the window length and interval, the time resolution and feature extraction accuracy can be balanced, thereby improving the performance of activity recognition.

[0063] Furthermore, step S4 also includes the following steps: S41, judging the state of the preprocessed signal according to a preset rule; S42: Automatically adjust the glasses function according to the user's current state.

[0064] It should be noted that setting the current time The acceleration of the sensing control glasses is , the next moment The acceleration is The acceleration rate is - , the preset threshold is: Stillness threshold , Walking acceleration threshold ; Motion acceleration threshold ; Rate of change threshold ; The default rules are: when , and lasts for more than 3 seconds, the human body is judged to be in a static state; when , and adjacent If the sign changes alternately (such as positive to negative to positive), and at least three complete cycle fluctuations (such as peak-trough-peak) are detected within 5 seconds, the human body is judged to be in a walking state; when , , , and continuous The same symbol changes continuously; if the above state lasts for more than 2 seconds, the human body is determined to be in motion.

[0065] It should be added that since most human activities are similar, they can be roughly classified into the above three categories. When in the above three states, the glasses will also perform corresponding operations, including but not limited to: In static state: turn off the blue light protection mode to reduce power consumption; Walking status: Enable AR navigation prompts to enhance safety; Sports status: start the sports mode, display the heart rate and step frequency data in real time, and control the center 300 to drive the temple 110 to bend toward the human body (to achieve clamping).

[0066] Furthermore, the state determination rules can be optimized according to user behavior habits (such as movement frequency and gait characteristics) to further improve the accuracy.

[0067] Through parameters such as the acceleration change rate and the symbol alternation frequency, the static, walking and moving states can be distinguished more accurately than the judgment of a single threshold; low latency and low misjudgment rate, the judgment of a state requires the symbol change to last for at least two seconds, avoiding the interference of short pauses and sudden movements; the functions of traditional smart glasses are often "one size fits all" (such as turning on blue light protection all day), and this solution is dynamically adjusted according to the state to turn off high-power functions when static, extend battery life, and enable clamping and health monitoring during exercise, improve practicality, and realize on-demand activation of functional modules; secondly, the mechanical and electronic co-design, such as driving the temple 110 to bend to achieve dynamic clamping, adapt to different head shapes and exercise intensities, dynamically match the user state, and improve functional practicality and comfort. Compared with the traditional fixed function or single threshold judgment solution, this design is more in line with the actual use scenario, representing an important evolution of smart wearable devices towards scene intelligence and human-machine integration.

[0068] Embodiment three: The present embodiment provides a human body signal recognition induction control glasses, the human body signal recognition induction control glasses include a glasses frame 100 detachably connected by temples 110 and frames 120, a control center 300 and a signal capture module 200, the control center 300 also includes: A repository for pre-inputting scleral feature information of more than one group of users as unlocking keys and setting corresponding permissions; and for setting environmental information corresponding to instructions for mobilizing various structures of the glasses; A data processing center, used to process information from the signal capture module 200; A verification module is used to make a similarity match judgment based on the collected information and the information in the repository; The authorization module is used to open the corresponding permissions of the glasses and make corresponding adjustments based on the judgment results and the information recorded in the database.

[0069] Reference Figure 3 , is a schematic diagram of sclera information collection according to a preferred embodiment of the present invention; the sclera image collection equipment collects the eyeball of the object to be verified and presents the corresponding image information on the display module. A typical sclera recognition system includes: interception, feature enhancement, feature extraction and feature matching. First, the sclera part is intercepted from the original eye image, and the background other than the vascular texture is filtered out by the feature enhancement algorithm. According to the design of the descriptor, the feature descriptor is extracted from the enhanced image, and the feature descriptor set is used as the sclera feature to compare with other existing sclera features to achieve the purpose of biometric recognition.

[0070] Further, see Figure 4As a specific embodiment, an inductive control method for inductive control glasses for human signal recognition is provided, wherein the inductive control glasses for human signal recognition include a glasses frame 100, a signal capture module 200, a control center 300 and a battery 400; the glasses frame 100 includes temples 110 with touch areas and a frame 120 for loading lenses; the temples 110 and the frame 120 are detachably connected; the temples 110 are made of conductive material and are electrically connected to the control center 300; the signal capture module 200 is disposed on the glasses frame 100; the battery 400 is disposed in the temples 110 and is electrically connected to a charging port disposed on the surface of the temples 110; the signal capture module 200 and the battery 400 are electrically connected to the control center 300, respectively.

[0071] It should be appreciated that embodiments of the present invention may be implemented or enforced by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer-readable memory. The method may be implemented in a computer program using standard programming techniques - including a non-transitory computer-readable storage medium configured with a computer program, wherein the storage medium so configured causes the computer to operate in a specific and predefined manner - according to the methods and drawings described in the specific embodiments. Each program may be implemented in a high-level procedural or object-oriented programming language to communicate with a computer system. However, if desired, the program may be implemented in an assembly or machine language. In any case, the language may be a compiled or interpreted language. In addition, the program may be run on a programmed ASIC for this purpose.

[0072] In addition, the operations of the processes described herein may be performed in any suitable order unless otherwise indicated herein or otherwise clearly contradicted by context. The processes described herein (or variations and / or combinations thereof) may be performed under the control of one or more computer systems configured with executable instructions, and may be implemented as code (e.g., executable instructions, one or more computer programs, or one or more applications) that is executed collectively on one or more processors, by hardware, or a combination thereof. The computer program includes a plurality of instructions that may be executed by one or more processors.

[0073] Various aspects of the present invention may be implemented in machine-readable code stored on a non-transitory storage medium or device, whether removable or integrated into a computing platform, such as a hard disk, optical read and / or write storage medium, RAM, ROM, etc., so that it can be read by a programmable computer, and when the storage medium or device is read by the computer, it can be used to configure and operate the computer to perform the processes described herein. In addition, the machine-readable code, or portions thereof, may be transmitted over a wired or wireless network. When such media includes instructions or programs that implement the steps described above in conjunction with a microprocessor or other data processor, the invention described herein includes these and other different types of non-transitory computer-readable storage media. When programmed according to the methods and techniques of the present invention, the present invention also includes the computer itself.

[0074] The computer program can be applied to input data to perform the functions described herein, thereby converting the input data to generate output data stored in a non-volatile memory. The output information can also be applied to one or more output devices such as a display. In a preferred embodiment of the present invention, the converted data represents physical and tangible objects, including specific visual depictions of physical and tangible objects produced on the display.

[0075] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for induction control of human body signal recognition, characterized in that: The induction control glasses for human signal recognition include a glasses frame detachably connected by temples and frames, a control center, and a signal capture module. The induction control method includes: Step S1, the temples touch the skin, generating an electrical signal and waking up the glasses; Step S2: The signal capture module captures human body signals and transmits the signals to the control center; Step S3, the control center pre-processes the human body signal; Step S4: determine the status of the preprocessed signal and automatically adjust the glasses function according to the current status of the user.

2. The induction control method for human body signal recognition according to claim 1, characterized in that: The step S1 also includes the following steps: Step S0: pre-input one or more sets of user's sclera feature information as an unlocking key, and set corresponding permissions.

3. The induction control method for human body signal recognition according to claim 1, characterized in that: The signal capture module includes an acceleration sensor and an image sensor, and the step S2 further includes the following steps: Step S21, capturing a human face signal, a motion signal, and a blinking signal.

4. The induction control method for human body signal recognition according to claim 3, characterized in that: The step S2 further comprises the following steps: Step S211: The image sensor captures the user's sclera signal and transmits it to the control center; Step S212: The control center extracts characteristic information from the sclera signal according to the image processing algorithm, and performs data processing on the characteristic information to obtain a first signal; Step S213, performing feature comparison on the first signal; Step S214: If the calculation result is lower than the set threshold, it is determined to be a match, and it is considered that this feature corresponds to an identity in the database, and the glasses are unlocked at the same time; Step S215: At predetermined time intervals, the image sensor and the acceleration sensor continuously capture human body signals and transmit the signals to the control center.

5. The induction control method for human body signal recognition according to claim 1, characterized in that: The step S3 comprises the following steps: Step S31, filtering and invalid data processing; Step S32: linear interpolation; Step S33, filtering and denoising; Step S34: Standardization processing.

6. The induction control method for human body signal recognition according to claim 5, characterized in that: The linear interpolation formula used in step S32 is as follows: ； in, : The data point to be interpolated, the known data point on its left is , the known data points on the right are ; : The missing value after interpolation, the known data point on the left is , the known data points on the right are .

7. The induction control method for human body signal recognition according to claim 6, characterized in that: The formula used for filtering and denoising in step S33 is as follows: ； in: is the cut-off frequency, is the passband edge frequency.

8. The induction control method for human body signal recognition according to claim 7, characterized in that: The standardization in step S34 is mean-variance standardization, and its formula is as follows: ； in, is the mean of the input sample x, is the standard deviation of the input sample x.

9. A sensor-controlled glasses for human signal recognition, characterized in that: The human body signal recognition induction control glasses include a glasses frame detachably connected by temples and frames, a control center and a signal capture module, and the control center also includes: A repository for pre-inputting scleral feature information of more than one group of users as unlocking keys and setting corresponding permissions; and for setting environmental information corresponding to instructions for mobilizing various structures of the glasses; A data processing center, used for processing the information from the signal capture module; A verification module is used to make a similarity match judgment based on the collected information and the information in the repository; The authorization module is used to open the corresponding permissions of the glasses and make corresponding adjustments based on the judgment results and the information recorded in the database.

10. A sensor-controlled eyeglasses for human signal recognition, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program implements the steps of the control method according to any one of claims 1 to 8 when executed by the processor.

Images