Portable detection equipment and detection method applied to visual acuity identification

Abstract

The invention discloses portable detection equipment applied to visual acuity identification and a detection method. The equipment comprises an eyeshade type spectacle frame, a telescopically connected head band, a power supply module, a control module, an acquisition module, a wireless module and a display module, the display module is a lens type display screen and is mounted at the front end of the spectacle frame; the acquisition module comprises an acquisition unit and an analog-to-digital conversion unit, the acquisition unit is composed of an acquisition electrode and a reference electrode, during wearing, the acquisition electrode clings to the forehead of the brain, and the reference electrode is clamped at the earlobe; the power supply module, the control module and the wireless module are embedded in the spectacle frame; the control module is respectively connected with the acquisition module, the power supply module, the wireless module and the display module; two ends of the head band are respectively connected with two ends of the spectacle frame. According to the invention, data acquisition, analysis processing and result output are integrated, functions are integrated, and the visual acuity identification result can be obtained only by using detection equipment. Moreover, the integrated design can improve the portability of the detection equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of vision detection, and particularly to a portable detection device and a detection method applied to visual acuity recognition. Background Art

[0002] In the past few decades, the vision conditions of Chinese people have changed significantly, and the myopia rate has been showing a continuous growth trend. The increasing learning pressure and the long-term use of electronic devices are important reasons for eye fatigue and myopia. However, the root cause is that people do not pay attention to protection. Therefore, it is particularly important to prevent and control in a timely manner, detect and treat early, and strengthen vision protection.

[0003] Common vision detection methods include subjective vision detection methods and objective vision detection methods, such as the vision chart detection method and the computer optometry instrument, etc. In most cases, these detections can effectively evaluate a person's vision condition. However, they also have some limitations: they are easily affected by factors such as the psychological state, attention level, and cognitive ability of the person being detected, and have a certain degree of subjectivity. When it is required that the person being examined has the correct cognitive ability and can actively cooperate with the detection, for some special groups, such as young children, people with communication disorders, and hysterical patients, etc., effective feedback cannot be obtained, and the accuracy of the results cannot be ensured, having a certain population limitation. Traditional detections are easily affected by external factors, such as light intensity, distance, and spatial position, etc. The change of each condition may affect the detection result, having a certain environmental dependence, and need to be carried out in a specific environment. Due to the existing various deficiencies, an objective vision detection method that does not require participation at the conscious level is sought.

[0004] As a signal for recording the electrophysiological changes of brain activities, the electroencephalogram (EEG) signal contains a large amount of physiological and disease information, including information related to human vision. By analyzing the EEG signal that is not processed and fed back through subjective consciousness, the process of visual information processing can be understood more deeply, and the vision conditions of various populations can be detected more objectively. Therefore, it is particularly important to develop a portable visual acuity detection device based on the EEG signal that can meet the needs of the public. This device aims to solve the problems existing in current vision diagnosis and provide a novel, reliable, and objective visual acuity detection method. Summary of the Invention

[0005] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a portable detection device and a detection method applied to visual acuity recognition.

[0006] To achieve the above purpose, the technical solutions provided by the present invention are as follows:

[0007] A portable detection device applied to visual acuity recognition, comprising an eyeglass frame in the form of an eye mask, a headband with a telescopic connection, a power module, a control module, a collection module, a wireless module and a display module;

[0008] Among them,

[0009] The display module is a lens-type display screen, installed at the front end of the eyeglass frame, used to display the operation desktop and play visual stimulation videos after the detection starts;

[0010] The power module, control module and wireless module are all embedded in the eyeglass frame; the control module is respectively connected to the collection module, power module, wireless module and display module; the power module is used to provide operating power for the detection device; the control module issues instructions for the acquisition, transmission and processing of electroencephalogram signals and executes them; the wireless module transmits the detection results to the user terminal;

[0011] Both ends of the headband are respectively connected to both ends of the eyeglass frame.

[0012] Further, the control module includes a key button unit and a main control unit; the collection module includes a collection unit and an analog-to-digital conversion unit;

[0013] The collection unit includes a collection electrode and a reference electrode. When worn, the collection electrode is closely attached to the frontal part of the brain, and the reference electrode is clipped to the earlobe;

[0014] The main control unit is respectively connected to the key button unit and the analog-to-digital conversion unit.

[0015] Further, the main control unit uses an STM32F4 chip;

[0016] The analog-to-digital conversion unit uses an ADS1299 analog-to-digital conversion chip;

[0017] The power module is composed of a rechargeable lithium iron battery.

[0018] To achieve the above object, the present invention further provides a detection method applied to visual acuity recognition, which is implemented by using the above portable detection device, including:

[0019] Toggle the power-on button in the key button unit, initialize the system, and wear the portable detection device on the head after it starts; after wearing, the collection electrode is closely attached to the forehead, the reference electrode is clipped to the earlobe, the lens-type display screen is fixed directly in front of both eyes, and only the display desktop can be seen within the line of sight and there is no external light, that is, the wearing is completed;

[0020] Press the start button in the button unit. The user views a visual stimulation video on the lens display screen. At the same time, the acquisition unit starts to acquire EEG signals. When the video playback ends, the acquisition stops immediately. The control module transmits the acquired EEG signals to the analog-to-digital conversion unit for analog-to-digital conversion, and then the main control unit analyzes and processes the analog-to-digital converted EEG signals to obtain the inspection result, and finally sends the detection result to the client through the wireless module.

[0021] Furthermore, the process of analyzing and processing the analog-to-digital converted EEG signals by the main control unit includes preprocessing and segmentation, feature extraction and analysis, and detection model recognition.

[0022] In the preprocessing and segmentation stage, operations including denoising, artifact removal, segmentation, smoothing, and ensemble averaging are performed on the EEG signals.

[0023] In the feature extraction and analysis stage, time-domain features, frequency-domain features, and non-linear features are extracted and analyzed to extract features with strong correlation.

[0024] In the detection model recognition stage, the detection model associates the extracted features with categories to obtain the final detection result.

[0025] Furthermore, the specific process of the preprocessing and segmentation stage includes:

[0026] Perform denoising processing on the signal. By performing empirical mode decomposition on the EEG signal, it is adaptively decomposed into components of different frequencies, obtaining the original signal and eight intrinsic mode function components IMF1 - 8 and a residue after decomposition. Observe its corresponding frequency-domain expression, and filter out the components with frequencies higher than 50Hz, thereby successfully suppressing high-frequency noise while retaining most of the low-frequency effective signals.

[0027] To ensure that the data volumes of different categories are approximately the same and avoid the subsequent training process focusing on the learning of a certain category, the data is augmented accordingly. During the processing, the signal is randomly selected with a fixed-length time series window, segmented, and then the selected data is stretched or compressed to enhance and reconstruct the data. Among them, the data of the category with the largest data volume is used as the target number, and the data of other categories is augmented to achieve relative data balance. To facilitate observing the waveform, ensemble averaging is performed on the signal, and the signal is averaged 5 times, 10 times, or 20 times in sequence until the change in waveform components can be clearly observed and then stopped.

[0028] Furthermore, the specific process of feature extraction includes:

[0029] In the time-domain features, the mean absolute value MAV is extracted. The average value of the absolute value of the signal is calculated for the preprocessed signal under different time windows, where N is the number of points of the time window signal, x(i) is the value of the i-th point, and M is the logarithmic conversion value;

[0030] In the frequency domain features, the extracted one is the power spectral density PSD. The signal is divided into multiple overlapping windows, and each window is subjected to Fourier transform and then averaged, that is P(w) is the PSD value, L is the window length, C is the number of windows, and F k (w) is the Fourier transform of the k-th window;

[0031] In the non-linear features, the extracted one is the differential entropy DE, and its calculation formula is

[0032] The specific process of the analysis stage includes:

[0033] Perform Pearson analysis and Spearman rank correlation analysis on the extracted features to judge whether the extracted features have a strong correlation with visual acuity. Combine the features with correlation for subsequent classification.

[0034] Furthermore, the feature extraction stage also includes extracting features including peak-to-peak value and response area.

[0035] Furthermore, the detection model recognition stage includes:

[0036] Assign class labels to the extracted features, where one label represents one visual acuity level; then input them into the Softmax classifier to calculate the probability that it belongs to a certain class, where Then measure the difference between the prediction result and the true label through the cross-entropy loss function, where the cross-entropy expression is x i is the sample input, and y i is the sample label; when the value of the cross-entropy loss function is smaller, the prediction result is closer to the true label; at the same time, use the extracted features for the training of AdaBoost, SVM, and RF classifiers to build a composite model, and pass the test set data through the trained model to obtain the final visual acuity result through a voting mechanism.

[0037] Furthermore, the detection model recognition stage also includes: If there is new data later, put it into the constructed model for training and testing to expand the model data volume and improve reliability.

[0038] Compared with the prior art, the principle and advantages of this technical solution are as follows:

[0039] 1. Integrate the telescopically connectable headband, power module, control module, acquisition module, wireless module, and display module onto the eyeglass frame in the form of an eye mask, that is, integrate data acquisition, analysis and processing, and result output into one, with integrated functions. The visual acuity recognition result can be obtained only by using the detection device. Moreover, the integrated design can improve the portability of the detection device.

[0040] 2. The device adopts an integrated design and is not easily affected by external environmental factors. It can be used immediately after wearing and is suitable for different groups of people.

[0041] 3. Equipped with a power module, it can be detected anytime and anywhere in various environments when the power is sufficient. When the power is insufficient, the lithium iron phosphate battery can be replaced, or the battery can be charged.

[0042] 4. In the preprocessing process, high-frequency noise is removed by segmenting the EEG signal, and then data augmentation is performed on the minority class data to balance the dataset. After superposition and smoothing, it is easier to observe the change trend of the waveform for subsequent extraction of accurate and effective signal features.

[0043] 5. In the feature extraction process, time-domain, frequency-domain, and non-linear features are extracted, and signal features are extracted from different angles, making full use of the value contained in the signal waveform. Correlation analysis is performed on multiple types of features in combination, and strongly correlated features are screened.

[0044] 6. In the model construction process, a multi-classification task of visual acuity is performed by jointly comparing the Softmax classifier and classifiers such as AdaBoost, and a composite model is constructed. Then, the visual acuity is judged through a voting mechanism. The method of using a composite model improves the discrimination accuracy of the model for different visual acuity levels.

[0045] 7. The detection data and results are transmitted to the user terminal through the wireless module, which is convenient for the user to follow up the changes in personal vision in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the services required in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0047] Figure 1 It is a schematic diagram of a portable detection device for visual acuity recognition according to an embodiment of the present invention;

[0048] Figure 2 It is a connection block diagram of each module in a portable detection device for visual acuity recognition according to an embodiment of the present invention;

[0049] Figure 3 This is a schematic block diagram of the internal structure of a portable detection device for visual acuity recognition according to an embodiment of the present invention;

[0050] Figure 4 This is a schematic flow chart of the principle of a detection method for visual acuity recognition according to an embodiment of the present invention;

[0051] Figure 5 This is a schematic flow chart of the principle of analyzing and processing the electroencephalogram signal after analog-to-digital conversion in a detection method for visual acuity recognition according to an embodiment of the present invention by the main control unit;

[0052] Figure 6 This is a flow chart of the preprocessing and segmentation stage in a detection method for visual acuity recognition according to an embodiment of the present invention;

[0053] Figure 7 This is a flow chart of the feature extraction and analysis stage in a detection method for visual acuity recognition according to an embodiment of the present invention;

[0054] Figure 8 This is a flow chart of the detection model recognition stage (including the model construction part) in a detection method for visual acuity recognition according to an embodiment of the present invention.

[0055] Reference numerals:

[0056] 1 - spectacle frame; 2 - headband; 3 - power supply module; 4 - control module; 5 - acquisition module; 6 - wireless module; 7 - display module; 8 - acquisition electrode; 9 - reference electrode; 10 - button unit; 11 - main control unit; 12 - analog-to-digital conversion unit; 13 - acquisition unit. Detailed implementation manners

[0057] The present invention will be further described below with reference to specific embodiments:

[0058] As Figures 1 to 3 shown, a portable detection device for visual acuity recognition described in this embodiment includes an eyeglass frame 1 in the form of an eye mask, a telescopic headband 2, a power supply module 3, a control module 4, an acquisition module 5, a wireless module 6, and a display module 7;

[0059] Among them,

[0060] The display module 7 is a lens-type display screen installed at the front end of the spectacle frame 1 for displaying an operation desktop and playing a visual stimulation video after the detection starts; the acquisition module 5 includes an analog-to-digital conversion unit 12 and an acquisition unit 13. The acquisition unit 13 consists of an acquisition electrode 8 and a reference electrode 9. When worn, the acquisition electrode 8 is closely attached to the forehead of the brain, and the reference electrode 9 is clipped to the earlobe; the power module 3, the control module 4, and the wireless module 6 are all embedded in the spectacle frame 1; the control module 4 is respectively connected to the acquisition module 5, the power module 3, the wireless module 6, and the display module 7; the power module 3 is used to provide operating power for the detection device; the control module 4 issues instructions for the acquisition, transmission, and processing of electroencephalogram signals and executes them; the wireless module 6 transmits the detection results to the user terminal; both ends of the headband 2 are respectively connected to both ends of the spectacle frame 1 and can be appropriately adjusted according to the head circumference of the human body when worn to meet the needs of different users.

[0061] Specifically, in this embodiment, the control module 4 includes a key button unit 10 and a main control unit 11; the main control unit 11 is respectively connected to the key button unit 10 and the analog-to-digital conversion unit 12.

[0062] The main control unit 11 uses an STM32F4 chip; the analog-to-digital conversion unit 12 uses an ADS1299 analog-to-digital conversion chip; the power module 3 consists of a rechargeable lithium-iron battery.

[0063] As Figure 4 shown, the working principle of this embodiment is as follows:

[0064] Toggle the power-on button in the key button unit 10 to initialize the system (including electroencephalogram acquisition initialization, video playback initialization, and wireless (transmission) module initialization). After the portable detection device is started, wear it on the head; after wearing, the acquisition electrode 8 is closely attached to the forehead, the reference electrode 9 is clipped to the earlobe, and the lens-type display screen is fixed directly in front of both eyes. Only the display desktop can be seen within the line of sight and there is no external light, that is, the wearing is completed;

[0065] Press the start button in the key button unit 10. The user can view the visual stimulation video on the lens-type display screen. At the same time, the acquisition unit 13 starts to acquire electroencephalogram signals; when the video playback ends, the acquisition stops immediately; the control module 4 transmits the acquired electroencephalogram signals to the analog-to-digital conversion unit 12 for analog-to-digital conversion, and then analyzes and processes the analog-to-digital converted electroencephalogram signals through the main control unit 11 to obtain the inspection results, and finally sends the detection results to the client through the wireless module 6.

[0066] As Figure 5 shown, in the above, the process of analyzing and processing the analog-to-digital converted electroencephalogram signals through the main control unit 11 includes preprocessing and segmentation, feature extraction and analysis, and detection model recognition;

[0067] In the preprocessing and segmentation stage, operations including denoising, artifact removal, segmentation, smoothing, and ensemble averaging are performed on the EEG signals;

[0068] In the feature extraction and analysis stage, time-domain features, frequency-domain features, and non-linear features are extracted and analyzed to extract features with strong correlation;

[0069] In the detection model recognition stage, the detection model associates the extracted features with categories to obtain the final detection result.

[0070] As Figure 6 shown, the specific process of the preprocessing and segmentation stage includes:

[0071] The signal is denoised. By performing empirical mode decomposition on the EEG signal, it is adaptively decomposed into components of different frequencies, obtaining the original signal, eight intrinsic mode function components IMF1 - 8 after decomposition, and a residue; observing their corresponding frequency-domain expressions, the components with frequencies higher than 50 Hz are filtered out, thereby successfully suppressing high-frequency noise while retaining most of the low-frequency effective signals;

[0072] To ensure that the data volumes of different categories are approximately the same and avoid subsequent training processes focusing on the learning of a certain category, the data is augmented accordingly; during processing, a fixed-length time-series window is randomly selected for the signal, the signal is segmented, and then the selected data is stretched or compressed to enhance and reconstruct the data; among them, the data of the category with the largest data volume is used as the target number, and the data of other categories is augmented to achieve relative data balance; to facilitate observing the waveform, ensemble averaging is performed on the signal, and the signal is successively averaged 5 times, 10 times, or 20 times until the change in waveform components can be clearly observed and then stopped.

[0073] As Figure 7 shown, the specific process of feature extraction includes:

[0074] In time-domain features, the mean absolute value MAV is extracted. The average value of the absolute value of the signal is calculated for the preprocessed signal under different time windows, where N is the number of points in the time-window signal, x(i) is the value of the i-th point, and M is the logarithmic conversion value;

[0075] In frequency-domain features, the power spectral density PSD is extracted. The signal is divided into multiple overlapping windows, and the Fourier transform is performed on each window and then averaged, that is P(w) is the PSD value, L is the window length, C is the number of windows, F k (w) is the Fourier transform of the k-th window;

[0076] In non-linear features, the differential entropy DE is extracted, and its calculation formula is

[0077] The specific process in the analysis stage includes:

[0078] Perform Pearson analysis and Spearman rank correlation analysis on the extracted features to determine whether the extracted features have a strong correlation with visual acuity (i.e., whether the P - value is less than 0.05), and combine the correlated features for subsequent classification.

[0079] In addition, the feature extraction stage also includes extracting features such as peak - to - peak value and response area.

[0080] Such as Figure 8 As shown, the detection model recognition stage includes:

[0081] Assign class labels to the extracted features, where one label represents one level of visual acuity; then input them into a Softmax classifier to calculate the probability of belonging to a certain class, where Furthermore, measure the difference between the prediction result and the true label through the cross - entropy loss function, where the cross - entropy expression is x i is the sample input, and y i is the sample label; when the value of the cross - entropy loss function is smaller, the prediction result is closer to the true label; at the same time, use the extracted features for training AdaBoost, SVM, and RF classifiers to build a composite model, and pass the test set data through the trained model to obtain the final visual acuity result through a voting mechanism.

[0082] Subsequently, if there is new data, put it into the pre - constructed model for training and testing to expand the model data volume and improve reliability.

[0083] In this embodiment,

[0084] Integrate the stretchable - connected headband, power module, control module, acquisition module, wireless module, and display module on the eyeglass - type frame, that is, integrate data acquisition, analysis and processing, and result output into one, with functional integration. Only using the detection device can obtain the visual acuity recognition result. Moreover, the integrated design can improve the portability of the detection device.

[0085] The device adopts an integrated design and is not easily affected by external environmental factors. It can be used immediately upon wearing and is suitable for different groups of people.

[0086] Equipped with a power module, it can detect anytime and anywhere in various environments when the power is sufficient. When the power is insufficient, the lithium - iron battery can be replaced, or the battery can be charged.

[0087] During the preprocessing process, high-frequency noise is removed by splitting the EEG signals, and then data augmentation is performed on the minority-class data to balance the dataset. After superposition and smoothing, it is easier to observe the change trend of the waveform for subsequent extraction of accurate and effective signal features.

[0088] During the feature extraction process, time-domain, frequency-domain, and non-linear features are extracted, extracting signal features from different perspectives, making full use of the value contained in the signal waveform, and performing correlation analysis on multiple types of features in combination to screen strongly correlated features.

[0089] During the model construction process, a multi-classification task for visual acuity is performed by jointly comparing a Softmax classifier and classifiers such as AdaBoost, constructing a composite model, and then discriminating visual acuity through a voting mechanism. The method of using a composite model is adopted to improve the discrimination accuracy of the model for different visual acuity levels.

[0090] The detection data and detection results are transmitted to the user terminal through a wireless module, facilitating the user to keep track of their personal vision change status in real time.

[0091] The above-described embodiments are only the preferred embodiments of the present invention and do not limit the scope of implementation of the present invention. Therefore, any changes made according to the shape and principle of the present invention should be covered within the protection scope of the present invention.

Claims

1. A portable detection device for visual acuity identification, characterized in that: It includes an eye mask type glasses frame, a retractable headband, a power module, a control module, a collection module, a wireless module and a display module; in, The display module is a lens-type display screen installed at the front end of the glasses frame, used to display the operating desktop and play the visual stimulation video after the detection starts; The power module, control module and wireless module are all embedded in the glasses frame; the control module is connected to the acquisition module, power module, wireless module and display module respectively; the power module is used to provide operating power for the detection equipment; the control module issues and executes instructions for the acquisition, transmission and processing of EEG signals; the wireless module transmits the detection results to the user terminal; The two ends of the headband are respectively connected to the two ends of the glasses frame.

2. A portable detection device for visual acuity identification according to claim 1, characterized in that: The control module includes a key button unit and a main control unit; the acquisition module includes an acquisition unit and an analog-to-digital conversion unit; the acquisition unit includes an acquisition electrode and a reference electrode. When worn, the acquisition electrode is close to the forehead of the brain, and the reference electrode is clamped at the earlobe; The main control unit is connected to the key button unit and the analog-to-digital conversion unit respectively.

3. A portable detection device for visual acuity identification according to claim 2, characterized in that: The main control unit adopts STM32F4 chip; The analog-to-digital conversion unit adopts the ADS1299 analog-to-digital conversion chip; The power module is composed of a rechargeable lithium iron battery.

4. A detection method for visual acuity recognition, characterized in that: The portable detection device according to claim 2 or 3 is used for implementation, comprising: Press the power button in the key button unit to initialize the system. After the portable detection device is started, wear it on the head. After wearing it, the collection electrode is close to the forehead, the reference electrode is clamped on the earlobe, and the lens-type display screen is fixed in front of the eyes. Only the display desktop can be seen within the field of vision and there is no external light. The wearing is complete. By pressing the start button in the button unit, the user watches the visual stimulation video on the lens-type display screen, and at the same time the acquisition unit starts to collect EEG signals; when the video playback ends, the acquisition stops immediately; the control module transmits the collected EEG signals to the analog-to-digital conversion unit for analog-to-digital conversion, and then the main control unit analyzes and processes the EEG signals after analog-to-digital conversion to obtain the inspection results, and finally sends the inspection results to the client through the wireless module.

5. A detection method for visual acuity recognition according to claim 4, characterized in that: The process of analyzing and processing the EEG signal after analog-to-digital conversion by the main control unit includes preprocessing and segmentation, feature extraction and analysis, and detection model recognition; In the preprocessing and segmentation stage, the EEG signals are subjected to operations including denoising, artifact removal, segmentation, smoothing, and stacking and averaging; In the feature extraction and analysis stage, the time domain features, frequency domain features and nonlinear features are extracted and analyzed to extract the features with strong correlation; In the detection model recognition stage, the detection model is used to associate the extracted features with categories to obtain the final detection results.

6. A detection method for visual acuity recognition according to claim 5, characterized in that: The specific process of preprocessing and segmentation includes: The signal is denoised by performing classical modal decomposition on the EEG signal, adaptively decomposing it into components of different frequencies, and obtaining the original signal and the eight intrinsic mode function components IMF1 to 8 after decomposition and a residual. The corresponding frequency domain expression is observed, and the components with a frequency higher than 50Hz are filtered out, thereby successfully suppressing high-frequency noise while retaining most of the low-frequency effective signals. In order to ensure that the amount of data in different categories is similar and to avoid the subsequent training process focusing on the learning of a certain category, the data is expanded accordingly; in the processing, the signal is randomly selected with a time series window of fixed length, the signal is segmented, and then the selected data is stretched or compressed to enhance and reconstruct the data; the category with the largest amount of data is used as the target number, and the data of other categories are expanded to achieve relative data balance; in order to facilitate the observation of the waveform, the signal is superimposed and averaged, and the signal is superimposed 5 times, 10 times or 20 times in turn until the change of the waveform component can be obviously observed and then stop.

7. A detection method for visual acuity recognition according to claim 5, characterized in that: The specific process of feature extraction includes: In the time domain features, the mean absolute value MAV is extracted, and the average value of the absolute value of the signal is calculated in different time windows for the preprocessed signal. N is the number of points in the time window signal, x(i) is the value of the i-th point, and M is the logarithmic transformation value; In the frequency domain, the power spectral density PSD is extracted. The signal is divided into multiple overlapping windows, and each window is Fourier transformed and then averaged, that is, P(w) is the PSD value, L is the window length, C is the number of windows, F k (w) is the Fourier transform of the kth window; Among the nonlinear features, the differential entropy DE is extracted, and its calculation formula is The specific process of the analysis phase includes: Pearson analysis and Spearman rank correlation analysis were performed on the extracted features to determine whether the extracted features were strongly correlated with visual acuity, and the features with correlation were combined for subsequent classification.

8. A detection method for visual acuity recognition according to claim 7, characterized in that: The feature extraction stage also includes extracting features including peak-to-peak value and response area.

9. A detection method for visual acuity recognition according to claim 5, characterized in that: The detection model identification stage includes: The extracted features are assigned category labels, one label represents a visual acuity level; then they are input into the Softmax classifier to calculate the probability of belonging to a certain category, where The cross entropy loss function is then used to measure the difference between the predicted result and the true label, where the cross entropy expression is x i is the sample input, y i is the sample label; when the value of the cross entropy loss function is smaller, the predicted result is closer to the true label; at the same time, the extracted features are used for AdaBoost, SVM, and RF classifier training to build a composite model, and the test set data is passed through the trained model to obtain the final visual acuity result through a voting mechanism.

10. A detection method for visual acuity recognition according to claim 9, characterized in that: The detection model identification stage also includes: if there is new data in the future, it will be put into the constructed model for training and testing, expanding the model data volume and improving reliability.

Images