Vision tester

Abstract

The invention discloses a vision tester, and relates to the field of visual function measurement. The vision tester comprises a display, a computer and a selector, every two of the display, the computer and the selector are in communication connection; during vision examination, a testee is positioned right opposite to the display; the display is used for displaying sighting marks with different sizes and shapes; the selector is worn by a testee and is used for collecting a sighting mark identification result of the testee; and the computer is used for determining the vision examination result of the testee according to the identification result of the testee on the sighting mark. According to the visual acuity tester, visual acuity screening is not limited by scenes, and convenience and accuracy of visual acuity screening can be improved.

Description

Technical Field

[0001] This application relates to the field of visual function measurement technology, and in particular to a vision testing instrument. Background Technology

[0002] Vision problems are a global public health issue. High myopia can lead to a series of serious eye complications, such as retinal detachment and macular degeneration, which are major causes of irreversible vision loss. Besides myopia, there are also other vision problems such as hyperopia, astigmatism, and amblyopia. Some vision problems can be effectively controlled or cured through early intervention. Vision screening can detect potential vision problems in a timely manner, providing a basis for subsequent diagnosis and treatment. Currently, medical institutions conducting vision screenings in communities and schools mostly use common near and far vision charts for manual screening. However, because the optotypes on the chart are of fixed size, testing needs to be conducted under specific lighting and at a fixed distance, thus placing high demands on the testing environment. Summary of the Invention

[0003] In view of the problems mentioned in the background art, this application provides a vision testing instrument that enables vision screening to be carried out without being limited by the scene, thereby improving the convenience and accuracy of vision screening.

[0004] To achieve the above objectives, this application provides the following solution.

[0005] This application provides a vision testing device, including: a display, a computer, and a selector; the display, computer, and selector are communicatively connected to each other; during the vision test, the test subject is positioned directly opposite the display; the display is used to display optotypes of different sizes and shapes; the selector is worn by the test subject and is used to collect the test subject's identification results of the optotypes; the computer is used to determine the test subject's vision test results based on the test subject's identification results of the optotypes.

[0006] Optionally, the selector is equipped with a first distance measuring device; the first distance measuring device is used to measure the current distance d between the subject and the display and send it to the computer; a second distance measuring device is installed at the bottom of the display; the second distance measuring device is used to measure the current height h of the display and send it to the computer; the computer is used to calculate the target height H of the display based on the height L of the subject and the current distance d between the subject and the display; the display has a height adjustment device; the height adjustment device is used to adjust the current height h of the display to the target height H.

[0007] Optionally, the display further includes a moving device; the moving device is used to adjust the current distance d between the subject and the display to a target distance D; the display is also used to display the current distance d between the subject and the display and the current height h of the display.

[0008] Optionally, the first and second rangefinders are infrared rangefinders or ultrasonic rangefinders.

[0009] Optionally, the computer is also used to calculate the size of the target displayed on the monitor based on the height L of the subject and the distance d between the subject and the monitor.

[0010] Optionally, the display also includes a brightness adjustment module; the brightness adjustment module includes a photosensitive element, a signal conditioning circuit, and a microcontroller; the photosensitive element generates a voltage signal internally by acquiring ambient light intensity; the signal conditioning circuit amplifies and converts the voltage signal to generate a digital signal; the microcontroller receives the digital signal and calculates the target screen brightness; the microcontroller controls the display to adjust the current screen brightness to the target screen brightness.

[0011] Optionally, the computer is equipped with a data management module; the data management module stores classified information to be tested for vision testing and basic information of the test subject; the information to be tested includes optotypes of different sizes and shapes; the basic information of the test subject includes the test subject's name, age, gender, height and vision information.

[0012] Optionally, the computer also includes a user management module; the user management module includes a log recording function to record user login, logout, addition, modification, and deletion operations.

[0013] Optionally, the computer is further equipped with a business module; the business module is used to determine that the subject's identification result of the visual target is incorrect if the subject's identification result of the visual target is not collected by the selector within three seconds after the visual target is displayed on the display.

[0014] Optionally, the business module is also used to calculate the accuracy rate of the test subject in identifying visual targets of the same size but different shapes; if the accuracy rate is between 40% and 80%, then the visual acuity value corresponding to the visual target of that size is the test subject's visual acuity test result.

[0015] Based on the specific embodiments provided in this application, the following technical effects are disclosed.

[0016] Using the vision testing device provided in this application, during vision testing, the test subject only needs to wear a selector and stand directly opposite the monitor. The monitor displays optotypes of different sizes and shapes, and the selector collects the test subject's identification results of the optotypes and transmits them to a computer. The computer can then determine the test subject's vision test result based on the identification results. This vision testing device of this application does not require vision testing and analysis in a specific location, allowing people to test their vision without being restricted by the scene. It enables medical institutions to conduct accurate, convenient, and scene-independent vision screenings outside of hospitals. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the vision testing instrument of this application. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] The purpose of this application is to propose a vision testing device that enables vision screening to be conducted without being limited by the setting, thereby improving the convenience and accuracy of vision screening.

[0021] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] In one exemplary embodiment, this application provides a vision testing device, such as... Figure 1 As shown, the vision testing device includes: a display 1, a computer 2, and a selector 3. The display 1, computer 2, and selector 3 are interconnected. During the vision test, the subject 4 is positioned directly opposite the display 1. The display 1 is used to display optotypes of different sizes and shapes. The selector 3 is worn by the subject 4 and is used to collect the subject 4's identification results of the optotypes. The computer 2 is used to determine the subject 4's vision test result based on the subject's identification results of the optotypes.

[0023] In one exemplary embodiment, the selector 3 is designed to be portable and can be worn on the subject 4. The selector 3 is equipped with a first distance measuring device, which is communicatively connected to the computer 2. The first distance measuring device is used to measure the current distance d between the subject 4 and the display 1 and transmit this measurement to the computer 2.

[0024] The display 1 is typically an LCD screen, with a screw base for attaching a bracket and mounting holes on the back for suspension. A second distance measuring device is mounted on the bottom of the display 1; this device communicates with the computer 2. The second distance measuring device measures the current height h of the display 1 and sends the measurement to the computer 2. The computer 2 calculates the target height H of the display 1 based on the height L of the subject 4 and the current distance d between the subject 4 and the display 1, using the following formula:

[0025]

[0026] Where, k e The eye height coefficient, fitted to the Asian ergonomics statistical database, is 0.92-0.94; the typical eye height coefficient k for adults is... e It is 0.936. c The screen height of monitor 1 is 15°, which is the medical standard downward viewing angle to avoid eye strain. ∈ represents the natural line of sight angle of the subject in a relaxed head and neck state, calculated using the following formula:

[0027]

[0028] The height L of the subject 4 can be input by the user or measured by a millimeter-wave radar module installed on the top of the display 1. That is, in an exemplary embodiment, a millimeter-wave radar module is also installed on the top of the display 1 for non-contact measurement of the height L of the subject 4, thereby automatically adapting to the height of children (<100cm) and the sitting height detection of wheelchair users.

[0029] The display 1 has a height adjustment device; the height adjustment device is used to adjust the current height h of the display 1 to a target height H. In an exemplary embodiment, the height adjustment device may be a servo lifting mechanism, using a worm gear drive + encoder feedback, with a stroke of 80cm and a positioning accuracy of ±0.5mm.

[0030] In one exemplary embodiment, the display 1 further includes a moving device, which can be in the form of a servo motor driving a belt and rollers. The moving device is used to adjust the current distance *d* between the subject 4 and the display 1 to a target distance *D*. For example, according to the national standard GB-11533-2011, for distance vision testing, the target distance *D* is typically 5m; for near vision testing, the target distance *D* is typically 25cm. However, due to limited space in some application scenarios, the moving device is not typically used to adjust the current distance *d* during distance vision testing. Instead, the target height *H* and the size of the displayed optotype are adjusted directly based on the subject 4's height *L* and the current distance *d*. In other words, once the subject 4 is stationary, no further movement is required; all adjustments are automatically made by the vision testing instrument, making it very convenient to use.

[0031] The display 1 and computer 2 are also used to display the current distance d between the subject 4 and the display 1, as well as the current height h of the display 1, so as to facilitate the adjustment of different current distances d between the subject 4 and the display 1 during near and far vision tests. Of course, in addition to automatic adjustment, the tester can also manually adjust the height adjustment device and the movement device of the display 1 through the computer 2.

[0032] In an exemplary embodiment, the first and second rangefinders can be infrared rangefinders or ultrasonic rangefinders, employing the time difference method for distance measurement. Calculating the distance by measuring the time difference between transmitting and receiving infrared / ultrasonic rays offers advantages such as high measurement accuracy, wide measurement range, and fast response speed.

[0033] In an exemplary embodiment, the first and second rangefinders are infrared rangefinders. The working principle of infrared ranging is based on the reflection and reception of light. An infrared rangefinder consists of an infrared transmitter and an infrared receiver. The infrared transmitter emits a beam of infrared light, which illuminates the display 1. The display 1 reflects a portion of the infrared light back to the infrared receiver. The receiver receives the reflected infrared light and converts it into an electrical signal. This electrical signal is converted into a digital signal by a converter and transmitted to the processor of the infrared rangefinder. The processor calculates the current distance d between the subject 4 and the display 1 based on the received infrared signal. The processor is communicatively connected to the display 1. During distance adjustment, the display 1 displays the current distance d between the infrared rangefinder and the subject 4 on the screen in real time, ensuring the subject 4 maintains the correct detection distance and avoiding interference with the vision test results.

[0034] Specifically, the current distance d between the subject 4 and the display 1 is calculated using the time difference between the transmitter sending the signal and the receiver receiving the signal. This is because the speed of infrared light in the air is known. By writing the time difference T between the transmitter sending the signal and the receiver receiving the signal into the processor, the current distance d between the subject 4 and the display 1 is calculated using the distance calculation formula (3):

[0035] d=(c*T) / 2 (3)

[0036] In the formula, c is the speed of light, which is 3 × 10⁻⁶. 8 m / s.

[0037] Infrared ranging is a non-contact ranging technology that measures the distance to an object by emitting and receiving infrared rays. It has advantages such as high measurement accuracy, wide measurement range, and fast response speed.

[0038] This application uses a display 1 to display optotypes of different sizes and shapes. The standard logarithmic visual acuity chart (E-chart) is currently the most commonly used visual acuity chart in China. The optotypes on the chart are in the shape of the letter "E," with the notch in the "E" pointing in four directions (up, down, left, and right). The optotype sizes are arranged according to a logarithmic law, with each row of optotypes differing by one logarithmic unit (10 to the power of 0.1). This arrangement makes the visual acuity test results more accurate and better reflects subtle changes in visual acuity. The international standard visual acuity chart (Snellen chart) is widely used internationally. The optotypes on this chart are mainly letters (such as C, D, E, F, L, N, O, P, T, Z, etc.), and the optotype sizes are arranged in fractional form. For example, 20 / 20 indicates that at a distance of 20 feet, one can clearly see the optotypes that normal vision should be able to see at 20 feet. The size changes of the optotypes on this chart do not follow a logarithmic law, but rather a fractional proportion. Considering that children may not yet recognize letters or numbers, optotypes on children's vision charts are usually graphic, such as animal figures (cats, dogs, ducks, etc.), patterns (triangles, circles, squares, etc.), or pointing symbols. These graphics are simple and easy to understand, making them suitable for children's vision checks. The vision test method is usually as follows: the subject stands at a specified distance (e.g., 5 meters), one eye is covered, and the subject uses the other eye to identify the optotypes such as the letter "E," letters, or figures on the vision chart. The optotypes are checked in descending order of size, and the visual acuity value corresponding to the smallest optotype that can be accurately identified is recorded as the vision test result.

[0039] In an exemplary embodiment, the computer 2 is further configured to calculate the size of the target displayed on the display 1 based on the height L of the subject 4 and the distance d between the subject and the display 1. The size of the target is measured according to the side length of the circumscribed square of the target, and the formula for calculating the size of the target displayed on the display 1 is as follows:

[0040]

[0041] Where S0 is the reference optotype size; for example, for a distance visual acuity chart, a visual acuity of 1.0 at a distance of 5m corresponds to an optotype side length of 8.73mm. S is the actual side length of the optotype displayed on monitor 1, in mm. d0 is the standard test distance, which is 5m for a distance visual acuity chart and 25cm for a near visual acuity chart. V ref This represents the current visual acuity level, such as 0.8 / 1.0 / 1.2.

[0042] V t This is the threshold visual acuity value, determined based on the subject's age (age). For example, when age < 10, as a developmental compensation for children, V... t =0.67*(1+0.1*(10-age)). When age>45, as presbyopia compensation, V t = 1.0 * (1 + 0.02 * (age - 45)). When age is between 10 and 45, V t =1.0.

[0043] Γ is the pupil-target matching coefficient, which is calculated using the following formula:

[0044]

[0045] In one exemplary embodiment, the display 1 further includes a brightness adjustment module. The brightness adjustment module includes a photosensitive element, a signal conditioning circuit, and a microcontroller. The photosensitive element generates a voltage signal internally by acquiring ambient light intensity. The signal conditioning circuit amplifies and converts the voltage signal to generate a digital signal. The microcontroller receives the digital signal and calculates the target screen brightness. The microcontroller controls the display to adjust the current screen brightness to the target screen brightness.

[0046] The photosensitive element can be a four-quadrant silicon photodiode (BPW34). The signal conditioning circuit uses a logarithmic amplifier (LOG114) instead of a linear amplifier. When ambient light shines on the surface of the photosensitive element, the light energy is absorbed, causing electrons in the element to jump to a higher energy level, forming electron-hole pairs. Under the influence of an electric field, these pairs separate, forming a current or voltage signal. This signal is amplified and processed by the signal conditioning circuit. The processed signal is then transmitted to the microcontroller via the output interface. By analyzing the signal, the microcontroller tests the intensity of the ambient light and automatically adjusts the brightness of the electronic screen to ensure that the screen's light intensity remains within the measured range.

[0047] In one exemplary embodiment, the formula for calculating the target screen brightness is as follows:

[0048]

[0049] Among them, L target The target screen brightness is calculated in cd / m². 2 L max The maximum brightness of monitor 1 is expressed in cd / m². 2 .

[0050] E v E0 represents the ambient illuminance measured in real time by the photosensitive element, in lux. E0 is the comfort illuminance benchmark, also in lux. This application employs a sliding time window adaptive algorithm to dynamically calculate the comfort illuminance benchmark.

[0051] E0(t)=α·E v (t)+(1-α)·E0(t-1) (7)

[0052] Where α is the forgetting factor, ranging from 0.05 to 0.2, and automatically increases when there are sudden environmental changes; t represents time, then E0(t) and E v (t) represent E0 and E at time t, respectively. v .

[0053] k is the luminance response coefficient, measured in lux. -1 The calculation formula is as follows:

[0054]

[0055] β is the pupil compensation factor, and its calculation formula is as follows:

[0056]

[0057] Where D p The pupil diameter can be measured by a binocular camera or estimated by age. When using a binocular camera, the camera is usually positioned at the top center of the monitor 1, facing the subject 4. It can track eye position in real time, dynamically compensate for head movements, monitor the pupil diameter, and collect data at a sampling rate of 60Hz. When using a binocular camera, head movement compensation can be appropriately added to formulas (1) and (4). When estimated by age, the female pupil diameter D... p = 5.40 + 0.285 × age - 0.0109 × age 2 Male pupil diameter D p = 5.83 + 0.181 × age - 0.0053 × age 2 .

[0058] ΔL pupil For pupil brightness compensation, the unit is cd / m2; specifically,

[0059]

[0060] Among them, D ref This is a pupil reference value. The pupil reference model used in this application is:

[0061] D ref =7.0-0.03·(age-20)+0.002·(age-20) 2 (11)

[0062] Age can be entered via a user interface or retrieved from data stored on a computer. A pupil reference model can be used to address differences in brightness perception caused by pupil constriction in older adults.

[0063] γ is the viewing angle attenuation coefficient, calculated using the following formula:

[0064]

[0065] Where θ is the ambient light incident angle, in degrees, calculated by the photosensitive element array.

[0066] In an exemplary embodiment, a four-quadrant photosensitive array is used to calculate the angle of the main light source based on the light intensity differences in the four directions, resulting in:

[0067]

[0068] in, This represents the ambient illuminance measured in real time by the i-th photosensitive array; Let be the normal vector of the i-th photosensitive array.

[0069] The ambient light adaptive brightness adjustment algorithm designed in this application can meet two core requirements: 1) Ensure that the screen brightness and background contrast meet vision test standards under different ambient lighting conditions, typically requiring a background brightness of 80-320 cd / m². 2 1) The contrast between the target and the background brightness is >90%; 2) Avoid drastic changes in screen brightness caused by sudden changes in ambient light, which may affect the comfort of the test.

[0070] In one exemplary embodiment, the computer is preferably a portable computer such as a tablet or laptop, but a desktop computer can also be used. The computer includes a data management module, a user management module, and a business module.

[0071] The data management module stores categorized information for vision testing, including target information and basic information of the test subject. The target information includes optotypes of different sizes and shapes. The basic information of the test subject includes their name, age, gender, height (L), and vision information (including historical vision test values).

[0072] The user management module supports user (typically inspectors) login, logout, addition, modification, and deletion operations. In one exemplary embodiment, the user management module includes a log system to record these operations. The system backend can view the operation logs to understand user activity and promptly identify and handle anomalies or risk events.

[0073] The business module is used to determine that if the test subject's identification result of the visual target is not collected by the selector within three seconds after the visual target is displayed on the display, the identification result is incorrect. The business module is also used to calculate the test subject's accuracy rate in identifying visual targets of the same size but different shapes; if the accuracy rate is between 40% and 80%, then the visual acuity value (visual acuity score) corresponding to that size visual target is the test subject's visual acuity test result.

[0074] By using the vision testing device of this application, the identification results of the vision test performed by the subject 4 can be obtained on the selector 3 and transmitted to the computer 2. The computer 2 simultaneously integrates the identification results and the subject's basic information to generate the subject's vision test information, enabling the subject to conveniently use the vision testing device for vision testing anywhere, without needing to go to a specific location for vision testing and analysis, thus making vision testing unrestricted by the scene. A rangefinder is provided on the display 1 and the selector 3 to measure the distance between the subject and the display screen, thereby determining the size of the visual target displayed on the screen. The rangefinder uses the time difference method for distance measurement, that is, it calculates the distance by measuring the time difference between transmitting and receiving infrared / ultrasound rays, which has the advantages of high measurement accuracy, wide measurement range, and fast response speed.

[0075] The laptop 2 is equipped with vision testing software. Through data management and business modules, it collects basic user information and identification results, and then the business module generates the test subject's vision test results, enabling accurate and scientific vision testing. Users can also query test information at any time to compare and analyze recent changes in vision data (this function can also be extended by the software) and take timely vision protection measures. The user management module includes a log recording system for user logins, additions, modifications, and deletions. The system backend can view the operation logs to understand user activity and promptly identify and handle abnormal or risky events. The business module limits the user's reaction time during vision testing to no more than 3 seconds, ensuring that users can complete the test within a normal reaction time. This guarantees that the test data is closer to real-world data and improves the accuracy of the vision test results.

[0076] The selector 3 of this application is designed to be portable and worn by the subject 4. It can monitor the distance between the subject and the display screen, and also collect the subject 4's recognition results of visual targets. In an exemplary embodiment, the selector 3 incorporates a microphone, a microcontroller, a wireless communication module, and a memory processor to implement voice acquisition, processing, and communication functions.

[0077] The microphone is used to capture the subject's voice signal. It needs to be equipped with a wind noise reduction design to minimize environmental interference. The microcontroller provides sufficient computing power to run the real-time speech recognition algorithm, requiring a processing power of 400MHz and support for floating-point operations. The wired communication module communicates with computer 2 to send speech recognition results, including recognized words and confidence levels. The memory processor stores the speech recognition model and temporary audio data. The selector 3 uses an ergonomic plastic shell, is easily worn (approximately 10cm x 4cm x 2cm), and has the microphone positioned close to the top to minimize obstruction.

[0078] The algorithm built into computer 2 is related to the algorithm of the microcontroller built into selector 3. When computer 2 displays the visual target on monitor 1, it wirelessly sends the "expected word" for the current visual target (e.g., "up" when the E-shaped notch is facing upwards). Selector 3 dynamically adjusts its recognition vocabulary based on this message, greatly improving accuracy. When selector 3 receives the "expected word" for the current visual target from computer 2, it wakes up the microphone to start collecting the voice signal of the subject 4. The business module is set to limit the subject's reaction time during vision testing to no more than 3 seconds. If no voice signal is collected from subject 4 within 3 seconds, the current visual target selection is determined to be incorrect, and computer 2 is notified to proceed to the next visual target recognition. Setting the reaction time for vision testing in the business module to no more than 3 seconds ensures that the subject can perform vision testing within a normal reaction time, guaranteeing that the test data is closer to real-life data and improving the accuracy and authenticity of the subject's vision test data.

[0079] For vision testing scenarios, the speech recognition algorithm integrated into the microcontroller must accurately identify the subject's descriptions of visual targets (such as "up," "down," "C," "cat," "triangle," etc.). Standard speech recognition (such as end-to-end models based on deep learning) may be resource-intensive and overfit to small vocabularies (20-30 words). Therefore, this application designs a lightweight context-aware speech recognition model (LCASR Model) with the following advantages: 1) Dynamic vocabulary constraint: During recognition, only the "expected vocabulary" sent by computer 2 is considered, rather than the entire vocabulary, significantly reducing computational complexity and error rate; for example, if the current visual target is an "E" shape, the vocabulary only includes {"up," "down," "left," and "right"}. 2) Multi-scale feature fusion: Combining traditional features (MFCCs) and deep learning features, an attention mechanism is used to focus on key speech segments. 3) Real-time adaptation: During the examination, the model fine-tunes acoustic features (such as accent compensation) based on the subject's previous responses, improving personalized accuracy. Verification has shown that the LCASR Model of this application is lightweight, occupies less than 100KB of memory, making it suitable for embedded devices; has a recognition latency of less than 500ms; and an accuracy of more than 95% (more than 90% in noisy environments).

[0080] The speech recognition algorithm integrated in the microcontroller includes the following steps S1 to S3.

[0081] S1. Preprocessing (real-time execution): Spectral subtraction is used to reduce noise in the acquired speech signal, minimizing environmental noise. The formula is as follows:

[0082]

[0083] Where Y(f) is the spectrum of the noisy speech, which is obtained by calculating the original acquired speech signal through FFT (Fourier Transform); N(f) is the noise spectrum, which is obtained through pre-calibration or real-time estimation; λ is the over-subtraction factor, which is 1.5 by default; This is the spectrum after noise reduction.

[0084] S2. Endpoint detection: Based on short-time energy and zero-crossing rate, determine the start and end points of speech, and process only valid speech segments.

[0085] S3. Feature Extraction: Calculate MFCCs (Mel-Frequency Cepstral Coefficients) as the basic features. First, analyze the spectral density of the denoised speech signal. The process is framed, with a frame length of 25ms and a frame shift of 10ms. The spectrum X(k) is obtained from each frame using FFT. A Mel filter bank is applied to obtain the filtering results M from m Mel filters. m :

[0086]

[0087] Where N is the number of FFT points (e.g., 512); H m (k) represents the weights of m Mel filters, where m ranges from 20 to 40, covering the frequency band sensitive to human hearing.

[0088] Taking the logarithm, we get:

[0089] S m =ln(M m (16)

[0090] The result obtained after DCT transformation is:

[0091]

[0092] Where n is the cepstral coefficient index, and the first 13 are taken as MFCCs. M is the number of Mel filters. n This is the extracted 13-dimensional MFCC feature vector for each frame.

[0093] The core recognition model of this application, LCASRModel, employs a lightweight neural network, combining CNN (for extracting local features), RNN (for processing temporal sequences), and an attention mechanism (for focusing on keyframes). The input is an MFCC sequence, i.e., c n The sequence is constructed; the output is the probability distribution of words.

[0094] The multi-scale CNN layer uses convolutional kernels of different sizes (1x3, 1x5) to process MFCC feature vectors in parallel, capturing both short-term and long-term features. For each convolutional kernel size k, the output feature map is:

[0095]

[0096] Among them W k It is the weight matrix, b k X represents the bias, X represents the input feature, * represents convolution, and ReLU() represents the ReLU activation function. t represents the time step.

[0097] The Bi-LSTM layer uses a bidirectional LSTM to process the time sequence and outputs the hidden state sequence h. t :

[0098] h t =LSTM(x t ,h t-1 (19)

[0099] Where x t It is a feature of the current frame.

[0100] The Attention layer dynamically assigns weights, focusing on keywords in the speech (such as visual marker descriptors). The Attention weights are as follows:

[0101]

[0102] Where v, W h W s b is a trainable parameter; s is the embedding vector of the current “expected word”, obtained from computer 2.

[0103] Context vector:

[0104] c = ∑ t a t h t (twenty one)

[0105] The output layer uses a fully connected layer and a softmax function to output word probabilities:

[0106]

[0107] Where V i d i It is the word w i The parameters of P(w) i |X) represents the vocabulary w when the input feature is X. i The probability, which is P(w) in the following text. i ).

[0108] The model has approximately 50,000 parameters to ensure embedded system compatibility. Cross-entropy loss is used during training.

[0109] L=-∑ i y i log P(w i ) (twenty three)

[0110] Where y i is the true label (one-hot encoded); L is the loss value.

[0111] During the vision test, the audio features of the subject's previous N responses (N=3-5) are stored. An online learning algorithm (such as incremental SGD) is then used to fine-tune the parameters of the attention layer to adapt to accents or speech rates. Fine-tuning loss:

[0112] L adapt =L+β·||θ-θ0|| 2 (twenty four)

[0113] Where θ is the current model parameter, θ0 is the initial model parameter, and β is the regularization coefficient (to prevent overfitting).

[0114] During LCASR model training, diverse speech datasets were collected, covering test subjects of different ages and accents (e.g., recording 5000 samples, with at least 200 samples per sight word). Background noise (such as white noise, dialogue), speed variation, and pitch shifting were added. The model was trained on a computer using Python / TensorFlow, then quantized (converted to 8-bit integers) and deployed to a selector. Compared to traditional speech recognition (such as HMM or end-to-end RNN), the LCASR model in this application reduces the error rate by 30-50% with a limited vocabulary, and is particularly adept at handling unclear pronunciation in children or the elderly. On an ARM Cortex-M7 microcontroller, model inference time is <50ms and memory usage is <100KB.

[0115] After processing by the LCASR Model, the probability P(w) is taken. i The highest vocabulary i The recognition result is the test subject 4's identification of the current visual target. A confidence score is also calculated.

[0116] confidence = max(P(w) i (25)

[0117] If the confidence level is less than the threshold (e.g., 0.8), prompt "Please repeat" via microphone; otherwise, send the identification result to the computer. If the identification result matches the target, record it as correct; otherwise, record it as incorrect. If the microphone prompts "Please repeat," reset the timer by 3 seconds.

[0118] In another exemplary embodiment, if the optotype uses only the "E" optotype from the E-chart visual acuity chart, the selector 3 can collect the recognition results without using voice recognition, instead using remote control buttons. In this exemplary embodiment, the surface of the selector 3 is provided with a remote control system, which includes buttons, a transmitter, a converter, and an integrated circuit. The buttons on the remote control system are used to allow the test subject to select the direction of the optotype notch, specifically including up, down, left, right, and a confirmation button. Each button on the remote control system corresponds to a different circuit inside the system, and each internal circuit uses a specific encoding method to correspond to the button. When a specific button is pressed, the corresponding circuit in the integrated circuit is connected, and the chip can detect which circuit is connected and determine which button has been pressed. Then, the chip emits an encoded sequence signal corresponding to that button. This signal is amplified and modulated before being sent to the light-emitting diode in the transmitter, where it is converted into an infrared signal and radiated outward by the converter. The computer 2 receives the infrared signal, demodulates it, and recovers the control signal, thus obtaining the test subject's selection result for the four directions: up, down, left, and right.

[0119] Computer 2 is equipped with comprehensive vision testing software. According to the pre-set test program, test targets will randomly appear on the screen for the test subject to perform vision testing. Finally, the test subject's vision is calculated based on the accuracy of the test subject's answers.

[0120] The data management module manages all data within the software. Users can register an account to gain access to the software and, after logging in, perform functions such as vision testing, vision information inquiry, and vision report printing. The user management module can grant or revoke user permissions based on their roles, enabling the rational allocation and management of system resources. Administrators can add, delete, modify, and query user information, while regular users can query basic information and vision test data.

[0121] The user management module can also verify user logins. When a user logs in, the module verifies whether the username and password entered by the user match, and decides whether to allow the user to log in based on the verification result. If the entered username and password match completely, the module will indicate that the verification is successful and allow the user to log in. If the entered username and password do not match, the module will indicate that the verification has failed and refuse the user to log in.

[0122] In this preferred embodiment, the user management module can also record user operation logs, including user login, addition, modification, deletion and other operations. Administrators can view the operation logs to understand the user's operation status and promptly detect and handle abnormal or risk events.

[0123] The business module is used to manage business processes. Similar to a paper vision chart, the data management module stores "E" optotypes of different sizes, and each optotype of different sizes has a corresponding visual acuity value. During the vision test, an "E" optotype of the same size but different orientation is randomly generated and displayed on the monitor 1. The "E" optotypes are displayed line by line in different sizes, and their actual display size on the screen can be calculated using formula (4). The test subject 4 needs to select the orientation of the notch of the "E" optotype in sequence using the selector 3, and find the best recognition line by checking line by line.

[0124] When the number of correctly identified optotypes in a row exceeds 80%, the test proceeds to a row with smaller optotypes. When the number of correctly identified optotypes in a row is less than 40%, the test proceeds to a row with larger optotypes. When the number of correctly identified optotypes in a row is between 40% and 80%, a row of optotypes of the same size is randomly generated for retesting. If the number of correctly identified optotypes is between 40% and 80%, the visual acuity represented by the optotypes in that row is the subject's visual acuity. If the number of correctly identified optotypes is outside the 40% to 80% range, the above testing logic continues. The subject is required to select the direction of the notch in the optotype within 3 seconds; otherwise, the test automatically proceeds to the next optotype, and the previous optotype selection is considered incorrect.

[0125] The data management module manages the data within the software. The business module, by accessing the test subject's basic information from the data management module, can adjust the monitor height and calculate the optotype size. Finally, combined with the test subject's identification results, it generates complete vision data, including uncorrected visual acuity, myopia, astigmatism, etc. Administrators can delete, modify, and query test subject vision test data through the user management module. Users can access the data management module through the user management module to query their personal vision data.

[0126] The vision testing instrument of this application is designed based on the national standard GB 11533-2011. A computer 2 with software installed controls the display content of a monitor 1 via Wi-Fi, Bluetooth, or wired connection. The monitor 1 automatically adjusts the brightness using a photosensitive device. A selector 3 measures the actual distance between the subject 4 and the screen and transmits it to the computer software. The computer 2 transmits the test information (optotypes of different sizes and shapes) to the monitor 1 via Wi-Fi and displays it synchronously on the screen. The subject 4 provides identification results based on the screen content, and the selector 3 feeds back the results to the computer 2, which are then recorded in the software. The computer 2 stores the test information and the basic information of the subject 4, categorized by type. It also integrates the tested information (the subject's identification results of the optotypes) and the basic information to generate the subject 4's vision test results. The generated vision test results are stored in a data management module, enabling users to conduct accurate and scientific vision tests. Users can also query the test information at any time, facilitating comparison and analysis of recent vision data changes and timely implementation of corresponding vision protection measures.

[0127] The vision testing instrument of this application does not require a specific location to test and analyze vision, so that people can test their vision without being restricted by the scene. It can be used conveniently anywhere to accurately test vision, which is convenient for medical institutions to conduct vision screening in remote towns and communities.

[0128] It should be noted that the terms "comprising," "including," or any other variations thereof used in this application are intended to cover non-exclusive inclusion, such that a product or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a product or system. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in a product or system that includes said element.

[0129] The foregoing description illustrates and describes several preferred embodiments of this application. However, as mentioned above, it should be understood that this application is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of this application's conception through the foregoing teachings or related technical or knowledge. Any modifications and variations made by those skilled in the art that do not depart from the spirit and scope of this application should be within the protection scope of the appended claims.

[0130] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0131] It should be noted that the information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with relevant regulations.

[0132] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A vision testing instrument, characterized in that, include: Monitors, computers, and selectors; The monitor, computer, and selector are connected in pairs for communication; during a vision test, the subject is positioned directly opposite the monitor. The display is used to show visual targets of different sizes and shapes; the selector is worn by the test subject and is used to collect the test subject's recognition results of the visual targets. The computer is used to determine the visual acuity test results of the test subject based on the subject's identification of visual targets.

2. The vision testing instrument according to claim 1, characterized in that, The selector is equipped with a first distance measuring device; the first distance measuring device is used to measure the current distance d between the subject and the display and send it to the computer; a second distance measuring device is installed at the bottom of the display; the second distance measuring device is used to measure the current height h of the display and send it to the computer; the computer is used to calculate the target height H of the display based on the height L of the subject and the current distance d between the subject and the display; the display has a height adjustment device; the height adjustment device is used to adjust the current height h of the display to the target height H.

3. The vision testing instrument according to claim 2, characterized in that, The display also has a moving device; the moving device is used to adjust the current distance d between the subject and the display to a target distance D; the display is also used to display the current distance d between the subject and the display and the current height h of the display.

4. The vision testing instrument according to claim 2, characterized in that, The first and second rangefinders are either infrared rangefinders or ultrasonic rangefinders.

5. The vision testing instrument according to claim 2, characterized in that, The computer is also used to calculate the size of the target displayed on the monitor based on the subject's height L and the distance d between the subject and the monitor.

6. The vision testing instrument according to claim 1, characterized in that, The display is also equipped with a brightness adjustment module; the brightness adjustment module includes a photosensitive element, a signal conditioning circuit and a microcontroller; the photosensitive element generates a voltage signal by acquiring the ambient light intensity; the signal conditioning circuit amplifies and converts the voltage signal to generate a digital signal; the microcontroller receives the digital signal and calculates the target screen brightness; the microcontroller controls the display to adjust the current screen brightness to the target screen brightness.

7. The vision testing instrument according to claim 1, characterized in that, The computer is equipped with a data management module; the data management module stores the test information and the basic information of the test subject in categories; the test information includes optotypes of different sizes and shapes; the basic information of the test subject includes the test subject's name, age, gender, height and vision information.

8. The vision testing instrument according to claim 7, characterized in that, The computer also includes a user management module; the user management module contains log records for recording user login, logout, addition, modification, and deletion operations.

9. The vision testing instrument according to claim 1, characterized in that, The computer is also equipped with a business module; the business module is used to determine that the subject's identification result of the visual target is incorrect if the subject's identification result is not collected by the selector within three seconds after the visual target is displayed on the display.

10. The vision testing instrument according to claim 1, characterized in that, The business module is also used to calculate the accuracy rate of the test subject in identifying visual targets of the same size but different shapes; if the accuracy rate is between 40% and 80%, then the visual acuity value corresponding to the visual target of that size is the test subject's vision test result.

Images