Personalized visual training device of split vision equipment

Abstract

The invention relates to the technical field of ocular vision optics, in particular to a personalized visual training device of split vision equipment. Comprising an information import module, a computer processing module, a training execution module, a local storage module and a cloud server module. The computer processing module can verify training scheme information in the local storage module according to the simultaneous visual index, the fusion function vector and the stereoscopic visual index; when the verification result is incomplete, the local storage module can download the training data in the cloud server module according to the verification result so as to supplement training scheme information; and the computer processing module can control the training execution module to execute the training process according to the training scheme information. In the prior art, asthenopia conditions of different individuals are different, so that the universality and effectiveness of training are insufficient. Compared with the prior art, personalized customization can be carried out according to user conditions, so that universality and effectiveness are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of optometry technology, and more particularly to a personalized vision training device for a split vision device. Background Technology

[0002] With changes in people's lifestyles, the incidence of myopia is constantly rising, posing a threat to the physical health of children and adolescents as well as the sustainable development of the economy and society.

[0003] Visual function and eye strain are both important indicators for evaluating visual health, with visual function training being a crucial foundational indicator. Many factors contribute to eye strain, including eye habits, screen time, study duration, and outdoor activity time, all of which influence the rate and severity of its accumulation. Currently, relatively mature training programs exist for improving visual function and alleviating eye strain. However, in practice, individual differences in eye strain levels limit the universality and effectiveness of these training programs. Summary of the Invention

[0004] To address the technical problems of existing technologies, this invention provides a personalized vision training device for split vision equipment.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0006] A personalized vision training device for split-view equipment includes: an information import module, a computer processing module, a training execution module, a local storage module, and a cloud server module. The computer processing module receives visual function detection information input from the information import module. This visual function detection information includes simultaneous vision detection information, fusion vision detection information, and stereoscopic vision detection information. The computer processing module generates a simultaneous vision index based on the simultaneous vision detection information. It also generates a fusion function vector based on the fusion vision detection information and a stereoscopic vision index based on the stereoscopic vision detection information. The computer processing module verifies the training scheme information in the local storage module based on the simultaneous vision index, fusion function vector, and stereoscopic vision index. When the verification result is incomplete, the local storage module downloads training data from the cloud server module to supplement the training scheme information. Finally, the computer processing module controls the training execution module to perform the training process based on the training scheme information.

[0007] Furthermore, the simultaneous vision detection information includes the identification contrast threshold and reaction time; the computer processing module is able to calculate the simultaneous vision index based on the contrast threshold, reaction time, and the simultaneous vision index calculation formula; the simultaneous vision index calculation formula is: SVI = 100 * [ w_c * N(C_th) + w_t * N(T_react) ]; SVI is the simultaneous vision index; w_c is the contrast threshold weight; N(C_th) is the normalized contrast threshold; w_t is the reaction time weight; N(T_react) is the normalized reaction time.

[0008] Furthermore, the computer processing module can process the contrast threshold according to the contrast threshold processing formula to obtain the normalized contrast threshold; the contrast threshold processing formula is: N(C_th) = (C_max - C_th) / (C_max - C_min); C_max is the preset upper limit of the contrast threshold; C_th is the contrast threshold; C_min is the preset lower limit of the contrast threshold.

[0009] Furthermore, the computer processing module can process the reaction time according to the reaction time processing formula to obtain the normalized reaction time; the reaction time processing formula is: N(T_react) = (T_max - T_react) / (T_max - T_min); T_max is the preset upper limit of the reaction time; T_react is the reaction time; T_min is the preset lower limit of the reaction time.

[0010] Furthermore, the fusion vision detection information includes fusion range and fusion stability; the computer processing module can calculate the fusion function vector based on the fusion range, fusion stability, and fusion function vector calculation formula; the fusion function vector calculation formula is: FFV=[FFI,S_Category]; FFI is the fusion function index; S_Category is the stability level classification; the fusion function index calculation formula is: FFI = FR_Base × S_Modifier × (1 - imbalance factor); FR_Base is the fusion range base score; S_Modifier is the stability adjustment factor.

[0011] Furthermore, the fusion range includes the horizontal fusion range and the vertical fusion range; the computer processing module can calculate the fusion range base score according to the fusion range base score calculation formula; the fusion range base score calculation formula is: FR_Base = 100 * [ (FR_H + FR_V) / (FR_H_max + FR_V_max) ]; FR_H is the horizontal fusion range; FR_V is the vertical fusion range; FR_H_max is the theoretical maximum value of the horizontal fusion range; FR_V_max is the theoretical maximum value of the vertical fusion range.

[0012] Furthermore, the imbalance factor is: min(0.3, |FR_H / FR_H_max - FR_V / FR_V_max| / 2).

[0013] Furthermore, the stereo vision detection information includes stereo vision sharpness and depth motion perception sensitivity; the computer processing module can calculate the stereo vision index based on the stereo vision sharpness, depth motion perception sensitivity, and the stereo vision index calculation formula; the stereo vision index calculation formula is: (SI) = 100 × [log_comp(SA) × (1 + α × N(DMS))]; SI is the stereo vision index; log_comp(SA) is the logarithmic compensation function of stereo vision sharpness; N(DMS) is the normalized fraction of depth motion perception sensitivity; α is the compensation coefficient.

[0014] Furthermore, the logarithmic compensation function is as follows:

[0015] SA represents stereoscopic acuity; SA_min represents the theoretical minimum value of stereoscopic acuity; SA_max represents the theoretical maximum value of stereoscopic acuity.

[0016] Furthermore, the normalized scores are as follows:

[0017] DMS_raw is the depth motion sensing sensitivity; DMS_mid is the median depth motion sensing sensitivity; k is the adjustment factor.

[0018] Compared with the prior art, the present invention has the following advantages:

[0019] This invention allows for customized training plans to be developed for users based on information obtained from the information import module, specifically the user's actual functional situation. This enhances the universality and effectiveness of the training.

[0020] Traditional methods can only make fuzzy judgments of "presence" or "good, average, poor." This system transforms the raw data (arcseconds, prism diopters, reaction time) from visual function detection into high-precision continuous quantification indices by constructing multiple dedicated calculation processes. This not only achieves accurate measurement of visual function but also captures complex subtypes such as "acceptable fusion range but extremely unstable," providing an unprecedentedly accurate data foundation for subsequent personalized training.

[0021] This invention abandons simple linear weighted scoring and designs unique algorithms for each visual function based on its physiological mechanism. For example, the fusion function uses a "product model" to reflect that "stability is a limiting factor for the effectiveness of range"; stereo vision uses a "logarithmic compensation model" to reflect the principle that "improvement in low-value areas is more difficult and more important" in depth perception. This makes the evaluation results more accurately reflect the user's visual perception level and makes the training objectives more scientific and reasonable.

[0022] Analysis dimensions such as "imbalance factors" can automatically detect imbalances in the development of a user's visual functions, such as excessive differences between horizontal and vertical fusion abilities, or a severe disconnect between static stereoscopic vision and dynamic sensitivity. This allows the system to issue early warnings and automatically introduce balanced training content to ensure the coordinated development of various visual functions and effectively prevent overall bottlenecks caused by overtraining in a single dimension.

[0023] By combining a core local storage module with dynamic cloud server updates, the system ensures both the stability and privacy of basic functions while achieving unlimited scalability and update capabilities. New training algorithms and content templates can be deployed via the cloud at any time, ensuring that all users continuously have access to the most cutting-edge and effective training solutions, thus solving the pain points of slow software updates and stagnant content in traditional devices. Attached Figure Description

[0024] Figure 1 Overall structure diagram. Detailed Implementation

[0025] The following are specific embodiments of the present invention, which are described in conjunction with the accompanying drawings. However, the present invention is not limited to these embodiments.

[0026] A personalized vision training device for split-view equipment includes: an information import module, a computer processing module, a training execution module, a local storage module, and a cloud server module. The computer processing module receives visual function detection information input from the information import module. This visual function detection information includes simultaneous vision detection information, fusion vision detection information, and stereoscopic vision detection information. The computer processing module generates a simultaneous vision index based on the simultaneous vision detection information. It also generates a fusion function vector based on the fusion vision detection information and a stereoscopic vision index based on the stereoscopic vision detection information. The computer processing module verifies the training scheme information in the local storage module based on the simultaneous vision index, fusion function vector, and stereoscopic vision index. When the verification result is incomplete, the local storage module downloads training data from the cloud server module to supplement the training scheme information. Finally, the computer processing module controls the training execution module to perform the training process based on the training scheme information.

[0027] The simultaneous vision index aims to intuitively reflect the efficiency and ability of a user's eyes to simultaneously receive and process information; a higher value indicates better simultaneous vision function. The simultaneous vision detection information includes the contrast threshold and reaction time. Specifically, the computer processing module processes the simultaneous vision index as follows:

[0028] The computer processing module calculates the simultaneous viewing index (SPI) based on the contrast threshold, response time, and the formula for calculating the SPI. The formula for calculating the SPI is:

[0029] SVI=100*[w_c*N(C_th)+w_t*N(T_react)].

[0030] SVI stands for Simultaneous Vision Index, which is the quantized value of the final output, typically ranging from 0 to 100 (theoretically it may exceed this, but it can be constrained by parameter settings). A higher score indicates better simultaneous vision functionality.

[0031] In the above formula, 100 is the scaling factor. The purpose of setting the scaling factor is to expand the result of the weighted sum to a percentage range that is easy to understand and compare.

[0032] w_c represents the contrast threshold weight. It indicates the proportion of importance of the contrast threshold in the overall evaluation. w_c is a constant between 0 and 1, and usually satisfies w_c + w_t = 1. For example, if both are considered equally important, w_c can be set to 0.5.

[0033] N(C_th) is the normalized contrast threshold. It is the value obtained by standardizing the recognition contrast threshold C_th obtained by the information import module, and its range is between 0 and 1.

[0034] $w_t$ is the reaction time weight, representing the proportion of the importance of reaction time in the comprehensive evaluation. Similar to $w_c$, $0 < w_t < 1$, and $w_c + w_t = 1$.

[0035] $N(T_{react})$ is the normalized reaction time, which is the value obtained after standardizing the reaction time $T_{react}$ acquired by the information import module, and its range is between 0 and 1.

[0036] Among them, the computer processing module can process the contrast threshold according to the contrast threshold processing formula to obtain the normalized contrast threshold;

[0037] The contrast threshold processing formula is:

[0038] $N(C_{th})=(C_{max}-C_{th}) / (C_{max}-C_{min})$;

[0039] $C_{max}$ is the preset upper limit of the contrast threshold. An empirical value, representing the critical point considered clinically for extremely poor or absent simultaneous vision function. For example, it can be set to 50% (that is, a very large contrast difference is required to perceive).

[0040] $C_{th}$ is the contrast threshold, which is specifically acquired by the information import module. It is the minimum contrast difference (usually expressed as a percentage, such as 10%) between the background and the visual target when the user can just simultaneously perceive two test images under the binocular split vision condition. The lower the value, the better the function.

[0041] $C_{min}$ is the preset lower limit of the contrast threshold. An empirical value, representing the best (lowest) threshold that can be achieved by normal or excellent simultaneous vision function. For example, it can be set to 1%.

[0042] The meaning of the contrast threshold processing formula is as follows: $(C_{max}-C_{th})$ makes the original threshold $C_{th}$ lower (better function), and this difference is larger. Dividing by $(C_{max}-C_{min})$ normalizes it to the 0 - 1 interval. The lower $C_{th}$ is, the closer $N(C_{th})$ is to 1, and the greater the contribution.

[0043] On the other hand, the computer processing module can process the reaction time according to the reaction time processing formula to obtain the normalized reaction time;

[0044] The reaction time processing formula is:

[0045] $N(T_{react})=(T_{max}-T_{react}) / (T_{max}-T_{min})$;

[0046] N(T_react) is the normalized reaction time. It is the value of the reaction time T_react obtained by the information import module after normalization, and its range is between 0 and 1.

[0047] T_max is a preset upper limit for the response time. It represents a critical value for extremely slow response times. For example, it can be set to 2000 milliseconds (2 seconds).

[0048] T_react represents the response time. It's the average time it takes for a user to make a correct recognition / judgment from the presentation of the test image, typically measured in milliseconds (ms). A lower value indicates better performance.

[0049] T_min is a preset lower limit for reaction time. It represents an ideal value for a very rapid reaction. For example, it can be set to 300 milliseconds (close to the natural physiological limit for simple visual tasks).

[0050] The meaning of the reaction time processing formula is as follows: the lower T_react (the better the function), the larger (T_max-T_react), and the closer N(T_react) is to 1.

[0051] An example of calculating the simultaneous viewing index is as follows:

[0052] Assume the system parameters are set as follows:

[0053] C_min=1%, C_max=50%, T_min=300ms, T_max=2000ms, w_c=0.6, w_t=0.4 (assuming that clinical experience suggests that contrast sensitivity contributes slightly more to the baseline of simultaneous vision than reaction speed).

[0054] Now assume there is a user with poor functionality. The detection results of the information import module are: recognition contrast threshold C_th = 30%, average reaction time T_react = 1500ms. Then the calculation process is as follows: N(C_th) = (50-30) / (50-1) = 20 / 49 ≈ 0.408;

[0055] N(T_react)=(2000-1500) / (2000-300)=500 / 1700≈0.294;

[0056] SVI=100*[0.6*0.408+0.4*0.294]=100*[0.2448+0.1176]=100*0.3624≈36.2.

[0057] Meanwhile, assuming a fully functional user exists. The detection results of the information import module are: recognition contrast threshold C_th = 5%, average reaction time T_react = 500ms. The calculation process is as follows: N(C_th) = (50-5) / (50-1) = 45 / 49 ≈ 0.918;

[0058] N(T_react)=(2000-500) / (2000-300)=1500 / 1700≈0.882;

[0059] SVI=100*[0.6*0.918+0.4*0.882]=100*[0.5508+0.3528]=100*0.9036≈90.4.

[0060] This index allows the computer processing module to quickly and quantitatively assess a user's simultaneous visual acuity (SVI) level, serving as a key criterion for matching and validating the training content's difficulty level. For example, the system can be configured to only allow access to advanced training programs requiring precise binocular coordination when SVI > 70.

[0061] The fusion function vector aims to comprehensively describe a user's fusion vision capabilities, encompassing two core dimensions: "range" and "stability." This vector not only provides a comprehensive assessment but also identifies specific capability gaps, offering precise guidance for personalized training. The fusion vision detection information includes fusion range and fusion stability. Specifically, the computer processing module processes the fusion function vector as follows:

[0062] The computer processing module calculates the fusion function vector based on the fusion range, fusion stability, and fusion function vector calculation formula.

[0063] The formula for calculating the fusion function vector is:

[0064] FFV=[FFI,S_Category];

[0065] FFI stands for Convergence Functionality Index (0-100 points).

[0066] S_Category is a stability level classification, specifically divided into four levels: A, B, C, and D.

[0067] The formula for calculating the integration function index is:

[0068] FFI = FR_Base × S_Modifier × (1 - Imbalance Factor);

[0069] FR_Base is the base score for the fusion range;

[0070] S_Modifier is a stability adjustment factor.

[0071] Specifically, the scope of integration includes horizontal integration scope and vertical integration scope;

[0072] The computer processing module can calculate the basic score of the fusion range according to the formula for calculating the basic score of the fusion range;

[0073] The formula for calculating the basic score for the fusion range is:

[0074] FR_Base=100*[(FR_H+FR_V) / (FR_H_max+FR_V_max)];

[0075] FR_H represents the horizontal blending range. The unit is prism diopters (Δ). It is calculated as: FR_H = |aggregation value| + |divergence value|. For example, if aggregation is +20Δ and divergence is -10Δ, then FR_H = 30Δ.

[0076] FR_V represents the vertical blending range. Unit: prism diopters (Δ). Calculated as above.

[0077] FR_H_max is the theoretical maximum value of the horizontal fusion range. It is the upper limit of the normal value matched based on age, for example: 40Δ for adults.

[0078] FR_V_max is the theoretical maximum value of the vertical fusion range. For example: 8Δ for adults.

[0079] On the other hand, S_Modifier sets the step size based on the value of FS_raw. Specifically, the calculation process of FS_raw is as follows: FS_raw = T_fused_ratio * 100 - 2 * N_breaks.

[0080] Specifically, FS_raw is defined as follows: At a set medium-difficulty fusion point (such as using a prism volume of 50% of the patient's fusion range), maintain for 60 seconds and record the "fusion holding time percentage" (%) and "rupture-recovery count".

[0081] Where T_fused_ratio is the ratio of the fusion holding time to the total mission time (60 seconds). N_breaks is the number of times the fusion needs to be restored after a break.

[0082] The meaning of FS_raw is: the longer the holding time and the fewer the number of breaks, the higher the FS_raw value and the better the stability.

[0083] S_Modifier is divided into four stages based on the value of FS_raw. Specifically:

[0084] If FS_raw ≥ 80 (excellent stability, no discount), then the value of S_Modifier is 1.0.

[0085] If 60 ≤ FS_raw < 80 (good stability, slight linear discount), then S_Modifier = 0.9 + 0.1 * (FS_raw - 60) / 20.

[0086] If 40 ≤ FS_raw < 60 (moderate stability, moderate discount), then S_Modifier = 0.7 + 0.2 * (FS_raw - 40) / 20.

[0087] If FS_raw < 40 (poor stability, severely discounted, down to 0.5), then S_Modifier = 0.5 * (1 + (FS_raw / 40)).

[0088] The formula for calculating the imbalance factor is as follows: Imbalance factor = min(0.3,|FR_H / FR_H_max-FR_V / FR_V_max| / 2).

[0089] The imbalance factor serves to penalize situations where the difference between the level of integration and the range of vertical integration is too large. This means that if the ratio of horizontal integration to the maximum horizontal value differs significantly from the ratio of vertical integration to the maximum vertical value, it indicates a clear imbalance in ability development, requiring a slight penalty to the total score (maximum deduction of 30%).

[0090] S_Category is also divided based on the value of FS_raw. The specific levels are as follows:

[0091] If FS_raw ≥ 80, it is grade A (excellent).

[0092] If 60 ≤ FS_raw < 80, it is grade B (good).

[0093] If 40 ≤ FS_raw < 60, then it is grade C (medium).

[0094] If FS_raw < 40, it is grade D (poor).

[0095] An example of calculating the fusion function vector is as follows:

[0096] Preset FR_H_max=40Δ, FR_V_max=8Δ.

[0097] Define a user group with a moderate range and good stability (balanced development). The data obtained through the information import module is as follows: Horizontal fusion: Aggregation +18Δ, Divergence -9Δ. Therefore, FR_H = 27Δ. Vertical fusion: Upward transfer +2Δ, Downward transfer -1Δ. Therefore, FR_V = 3Δ. Fusion maintenance task: T_fused_ratio = 0.7, N_breaks = 3, therefore FS_raw = 70 - 6 = 64.

[0098] The calculation process is as follows:

[0099] FR_Base=100×[(27+3) / (40+8)]=100×(30 / 48)=62.5.

[0100] Stability adjustment factor: FS_raw=64, which is grade B (good).

[0101] S_Modifier=0.9+0.1×(64-60) / 20=0.9+0.02=0.92.

[0102] Imbalance factor: Level ratio: 27 / 40 = 0.675.

[0103] Vertical ratio: 3 / 8 = 0.375.

[0104] Difference: |0.675-0.375|=0.3.

[0105] Imbalance factor = min(0.3, 0.3 / 2) = min(0.3, 0.15) = 0.15.

[0106] The final FFI = 62.5 × 0.92 × (1 - 0.15) = 62.5 × 0.92 × 0.85 ≈ 48.9.

[0107] Stability rating: S_Category="B".

[0108] Final fusion function vector: FFV=[48.9,"B"].

[0109] We define a user with a large range but extremely poor stability (a clear weakness). The information obtained through the information import module is as follows: Horizontal fusion: Aggregation +30Δ, Divergence -20Δ, therefore FR_H = 50Δ (exceeding FR_H_max, calculated as 40Δ). Vertical fusion: Upward rotation +3Δ, Downward rotation -2Δ, therefore FR_V = 5Δ. Fusion maintenance task: T_fused_ratio = 0.4, N_breaks = 8, therefore FS_raw = 40 - 16 = 24.

[0110] The calculation process is as follows:

[0111] FR_Base=100×[(40+5) / (40+8)]=100×(45 / 48)=93.75 (Horizontal blending is calculated based on the upper limit of 40).

[0112] Stability adjustment factor: FS_raw=24, belonging to grade D (poor).

[0113] S_Modifier=0.5×(1+24 / 40)=0.5×1.6=0.8.

[0114] Imbalance factors:

[0115] Horizontal ratio: 40 / 40 = 1.0 (up to the upper limit).

[0116] Vertical ratio: 5 / 8 = 0.625.

[0117] Difference: |1.0-0.625|=0.375.

[0118] Imbalance factor = min(0.3, 0.375 / 2) = min(0.3, 0.1875) = 0.1875.

[0119] The final FFI = 93.75 × 0.8 × (1 - 0.1875) = 93.75 × 0.8 × 0.8125 ≈ 60.9.

[0120] Final fusion function vector: FFV=[60.9,"D"].

[0121] When the computer processing module performs verification using FFV values, the selection is made based on the S_Category level. Specifically: S_Category="A": Allows entry into high-difficulty, high-dynamic fusion tasks. S_Category="B / C": Focuses on consolidation and anti-interference training. S_Category="D": Must start with basic stability training. Simultaneously, fine adjustments are made based on the imbalance factor and FR_Base values. The magnitude of the imbalance factor determines whether vertical fusion-specific training needs to be added. The FR_Base value determines the initial prism amount or disparity size for training. The cloud server can then precisely push content packages based on FFV=[FFI,S_Category]. For example, requesting a "large-range - low-stability correction package" for [60.9,"D"] contains numerous low-difficulty, long-duration fusion maintenance tasks. This composite computational logic better aligns with the physiological mechanisms of fusion vision, providing more precise guidance for the generation and selection of personalized training content.

[0122] Stereoscopic detection information includes stereoscopic sharpness and depth motion perception sensitivity. The computer processing module can calculate the stereoscopic index based on stereoscopic sharpness, depth motion perception sensitivity, and the stereoscopic index calculation formula. The stereoscopic index calculation formula is:

[0123] (SI)=100×[log_comp(SA)×(1+α×N(DMS))];

[0124] SI stands for Stereo Index;

[0125] log_comp(SA) is the logarithmic compensation function for stereo sharpness.

[0126] N(DMS) is the normalized score of depth motion sensing sensitivity.

[0127] α is the compensation coefficient, a constant between 0 and 0.3, for example, set to 0.2. It represents the maximum compensation ratio of depth motion perception sensitivity to the overall stereoscopic index. Setting it to 0.2 means that the sensitivity can improve the static stereoscopic score by up to 20%.

[0128] Specifically, the calculation process of log_comp(SA) is as follows:

[0129] SA stands for Stereo sharpness, measured in arcseconds (arcsec). A smaller value indicates better static stereo vision.

[0130] SA_min is the theoretical minimum value of stereoscopic acuity, usually set to 20 arcseconds (the lower limit of excellent level for normal adults).

[0131] SA_max is the theoretical maximum value of stereoscopic acuity, usually set to 3000 arcseconds (the threshold of extremely poor stereoscopic function, close to monocular vision).

[0132] log_comp(SA) has the following characteristics:

[0133] When SA=SA_min, log_comp(SA)=1.

[0134] When SA = SA_max, log_comp(SA) = 0.

[0135] The function decreases linearly on the logarithmic scale of SA, reflecting its high discriminative power in the low-value region.

[0136] The calculation process for N(DMS) is as follows:

[0137] DMS_raw represents the depth motion perception sensitivity, obtained from the information import module. This value indicates the minimum motion speed threshold (degrees per second) at which a user can correctly judge changes in the depth direction during a depth motion perception test. A lower value indicates higher sensitivity.

[0138] DMS_mid is the median depth motion sensing sensitivity, an empirical parameter, for example, set to 4 degrees / second (medium sensitivity).

[0139] k is an adjustment factor that controls the steepness of the logic function; for example, it can be set to 1.0.

[0140] The meaning of N(DMS) is to map the sensitivity threshold to the 0-1 range. When the threshold is lower than the median (high sensitivity), the score quickly approaches 1; when the threshold is higher than the median (low sensitivity), the score quickly approaches 0.

[0141] The calculation example is as follows:

[0142] Specifically, the preset system parameters are: SA_min=20 arcsec, SA_max=3000 arcsec, DMS_mid=4deg / sec, k=1.0, α=0.2. Depth motion sensing measurement: uses the minimum discernible depth motion velocity, in degrees per second. The smaller the value, the higher the sensitivity.

[0143] We now pre-select a user with excellent static stereo vision and good dynamic sensitivity.

[0144] The detection results obtained by the information import module are as follows:

[0145] Stereo sharpness SA = 40 arcsec.

[0146] Depth motion sensing sensitivity threshold DMS_raw=2deg / sec (Excellent).

[0147] The calculation process is as follows:

[0148] log10(SA)=log10(40)≈1.602.

[0149] log10(SA_min)=log10(20)≈1.301.

[0150] log10(SA_max)=log10(3000)≈3.477.

[0151] log_comp(40)=1-(1.602-1.301) / (3.477-1.301)=1-0.301 / 2.176≈1-0.138≈0.862.

[0152] N(DMS)≈0.881.

[0153] SI=100×[0.862×(1+0.2×0.881)]=100×[0.862×(1+0.176)]=100×[0.862×1.176]≈100×1.014=101.4.

[0154] We now pre-define a user with poor static stereo vision and poor dynamic sensitivity.

[0155] The data obtained by the information import module is:

[0156] SA=200arcsec.

[0157] DMS_raw=8deg / sec (poor sensitivity).

[0158] The calculation process is as follows:

[0159] Calculate log_comp(SA):

[0160] log10(200)≈2.301.

[0161] log_comp(200)=1-(2.301-1.301) / 2.176=1-1.000 / 2.176≈1-0.459≈0.541.

[0162] N(DMS)≈0.018.

[0163] SI=100×[0.541×(1+0.2×0.018)]=100×[0.541×(1+0.0036)]=100×[0.541×1.0036]≈100×0.543≈54.3.

[0164] When the computer processing module performs verification using the stereo index (SI):

[0165] The division is based on the SI value, as follows:

[0166] SI≥80: Fine stereo vision, capable of performing high-precision depth discrimination tasks.

[0167] 40≤SI<80: Moderate stereopsis, focus on conventional stereopsis enhancement.

[0168] SI<40: Crude stereopsis, requires starting with basic stereopsis stimulation.

[0169] Simultaneously, compare the relative magnitudes of log_comp(SA) and N(DMS): If log_comp(SA) > 2 × N(DMS): static performance is superior to dynamic performance, emphasizing dynamic sensitivity training. If N(DMS) > 2 × log_comp(SA): dynamic performance is superior to static performance, emphasizing static sharpness training. Otherwise: balanced training.

[0170] The cloud server then pushes targeted content packages based on the SI value and the static / dynamic ratio, such as "low SI - static-based training package" or "medium SI - dynamic reinforcement package".

[0171] In summary, this invention enables customized training programs for users based on information obtained from the information import module, specifically the user's actual functional situation. This enhances the universality and effectiveness of the training.

[0172] Meanwhile, traditional methods can only make fuzzy judgments of "presence" or "good, average, poor." This system transforms the raw data (arcseconds, prism diopters, reaction time) from visual function detection into high-precision continuous quantification indices by constructing multiple dedicated calculation processes. This not only achieves accurate measurement of visual function but also captures complex subtypes such as "acceptable fusion range but extremely unstable," providing an unprecedentedly accurate data foundation for subsequent personalized training.

[0173] On the other hand, the system of this invention abandons simple linear weighted scoring and designs unique algorithms for the physiological mechanisms of each visual function. For example, the fusion function adopts a "product model" to reflect that "stability is a limiting factor for the effectiveness of range"; stereo vision adopts a "logarithmic compensation model" to reflect the rule that "improvement in low-value areas is more difficult and more important" in depth perception. This makes the evaluation results more realistically reflect the user's visual perception level and makes the setting of training objectives more scientific and reasonable.

[0174] Meanwhile, analytical dimensions such as "imbalance factors" can automatically detect imbalances in the development of a user's visual functions, such as excessive differences between horizontal and vertical fusion capabilities, or a severe disconnect between static stereoscopic vision and dynamic sensitivity. This allows the system to issue early warnings and automatically introduce balanced training content, ensuring the coordinated development of various visual functions and effectively preventing overall bottlenecks caused by overtraining in a single dimension.

[0175] Secondly, through an architecture that combines a local storage core module with dynamic supplementation via cloud servers, the system ensures both the stability and privacy of basic functions while achieving unlimited scalability and update capabilities. New training algorithms and content templates can be deployed via the cloud at any time, ensuring that all users can continuously access the most cutting-edge and effective training solutions, thus solving the pain points of slow software updates and stagnant content in traditional devices.

[0176] The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which this invention pertains may make various modifications or additions to the described specific embodiments or use similar methods to substitute them, without departing from the spirit of the invention or exceeding the scope defined by the appended claims.

Claims

1. A personalized training device for a split-view device, characterized in that: include: Information import module, computer processing module, training execution module, local storage module, cloud server module; The computer processing module is capable of receiving visual function detection information input by the information import module; The visual function detection information includes simultaneous visual detection information, fused visual detection information, and stereoscopic visual detection information; The computer processing module is able to generate a simultaneous vision index based on the simultaneous vision detection information; The computer processing module is able to generate a fusion function vector based on the fusion view detection information; The computer processing module is capable of generating a stereo vision index based on the stereo vision detection information; The computer processing module can verify the training scheme information in the local storage module based on the simultaneous view index, the fusion function vector, and the stereo view index. When the verification result is incomplete, the local storage module can download the training data in the cloud server module according to the verification result to supplement the training scheme information; The computer processing module can control the training execution module to perform the training process based on the training scheme information.

2. The personalized training device for a split-view device according to claim 1, characterized in that: The simultaneous visual detection information includes the recognition contrast threshold and reaction time; The computer processing module is able to calculate the simultaneous viewing index based on the contrast threshold, the reaction time, and the simultaneous viewing index calculation formula. The formula for calculating the simultaneous viewing index is as follows: SVI = 100 * [ w_c * N(C_th) + w_t * N(T_react) ]; The SVI is the Simultaneous Vision Index; w_c is the contrast threshold weight; N(C_th) is the normalized contrast threshold; w_t is the reaction time weight; N(T_react) is the normalized reaction time.

3. The personalized training device for a split-view device according to claim 2, characterized in that: The computer processing module is able to process the contrast threshold according to the contrast threshold processing formula to obtain the normalized contrast threshold. The contrast threshold processing formula is as follows: N(C_th) = (C_max - C_th) / (C_max - C_min); C_max is the preset upper limit of the contrast threshold; C_th is the contrast threshold; C_min is a preset lower limit of the contrast threshold.

4. The personalized training device for a split-view device according to claim 2, characterized in that: The computer processing module is able to process the reaction time according to the reaction time processing formula to obtain the normalized reaction time. The reaction time processing formula is as follows: N(T_react) = (T_max - T_react) / (T_max - T_min); T_max is a preset upper limit for the reaction time; T_react is the reaction time; T_min is a preset lower limit for the reaction time.

5. The personalized training device for a split-view device according to claim 1, characterized in that: The fusion visual detection information includes fusion range and fusion stability; The computer processing module can calculate the fusion function vector based on the fusion range, fusion stability, and fusion function vector calculation formula. The formula for calculating the fusion function vector is as follows: FFV=[FFI,S_Category]; The FFI stands for Convergence Function Index. The S_Category is a stability level classification; The formula for calculating the fusion function index is as follows: FFI = FR_Base × S_Modifier × (1 - Imbalance Factor); FR_Base is the base score for the fusion range; S_Modifier is a stability adjustment factor.

6. The personalized training device for a split-view device according to claim 5, characterized in that: The fusion range includes the horizontal fusion range and the vertical fusion range; The computer processing module can calculate the basic score of the fusion range according to the formula for calculating the basic score of the fusion range. The formula for calculating the basic score of the fusion range is: FR_Base = 100 * [ (FR_H + FR_V) / (FR_H_max + FR_V_max) ]; FR_H refers to the horizontal fusion range; FR_V refers to the vertical fusion range; FR_H_max is the theoretical maximum value of the horizontal fusion range; FR_V_max is the theoretical maximum value of the vertical fusion range.

7. A personalized training device for a split-view device according to claim 6, characterized in that: The imbalance factor is: min(0.3, |FR_H / FR_H_max - FR_V / FR_V_max| / 2).

8. The personalized training device for a split-view device according to claim 1, characterized in that: The stereo vision detection information includes stereo vision sharpness and depth motion perception sensitivity. The computer processing module can calculate the stereo vision index based on the stereo vision sharpness, the depth motion perception sensitivity, and the stereo vision index calculation formula. The formula for calculating the stereoscopic vision index is as follows: (SI) = 100 × [log_comp(SA) × (1 + α × N(DMS))]; SI refers to the stereoscopic index; The log_comp(SA) is the logarithmic compensation function for the stereoscopic acuity; The N(DMS) is the normalized fraction of the depth motion sensing sensitivity; α is the compensation coefficient.

9. A personalized training device for a split-view device according to claim 8, characterized in that: The logarithmic compensation function is as follows: SA refers to the stereoscopic visual acuity. SA_min is the theoretical minimum value of stereoscopic acuity; SA_max is the theoretical maximum value of stereoscopic sharpness.

10. A personalized training device for a split-view device according to claim 8, characterized in that: The normalized scores are as follows: The DMS_raw is the depth motion sensing sensitivity; The DMS_mid is the median value of the depth motion sensing sensitivity; k is an adjustment factor.

Images