A stereoscopic vision training supervision method and supervision system in polarization mode

Abstract

This application relates to the field of training supervision technology, and in particular to a method and system for supervising stereoscopic vision training in polarization mode. The method includes acquiring real-time visual training images and real-time eye-tracking parameters of a patient during vision training; determining corresponding standard eye-tracking parameters based on the real-time visual training images; determining a first training score based on the real-time and standard eye-tracking parameters; determining a simulated fusion image based on preset left and right eye disparities and real-time eye-tracking parameters; calculating the standard deviation of fixation points in the simulated fusion image; and determining a second training score based on the standard deviation of fixation points; determining a total training score based on the first and second training scores; and when the total training score is lower than a preset threshold, identifying visual training weaknesses and their degree based on the real-time visual training images and the simulated fusion image, and optimizing the corresponding visual training images for the patient in the next training stage based on both. This application has the effect of improving the effectiveness of vision training.

Description

Technical Field

[0001] This application relates to the field of training supervision technology, and in particular to a stereoscopic vision training supervision method and supervision system in polarization mode. Background Technology

[0002] With the profound changes in modern lifestyles, the human visual system is facing unprecedented challenges. The frequent use of intelligent monitoring systems, the normalization of close-range work, and the homogenization of visual stimuli in the process of urbanization have led to an explosive increase in vision problems. Issues such as accommodative lag, eye movement control defects, and stereoscopic vision disorders caused by visual dysfunction are becoming key factors restricting the quality of life for the public. Against this backdrop, vision health training, as a core means of prevention, intervention, and rehabilitation, is experiencing exponential growth in market demand.

[0003] Traditional vision training methods often rely on static stimulation patterns, a typical example being the alternating fixation method with three colored beads. This method guides patients to switch fixation points through the spatial arrangement of red, yellow, and blue beads, aiming to enhance their dynamic depth perception. However, traditional training methods may have limitations such as a single approach and limited stimulation effects. These limitations may lead to some patients being able to pass static testing but still exhibiting visual processing deficiencies in dynamic scenarios. The limitations of traditional technical approaches are becoming increasingly apparent, potentially failing to meet the deeper needs of dynamic visual function reconstruction, thus resulting in poor training outcomes. Summary of the Invention

[0004] To improve the effectiveness of vision training, this application provides a stereoscopic vision training supervision method and supervision system in polarization mode.

[0005] Firstly, this application provides a stereoscopic vision training supervision method in polarization mode, employing the following technical solution:

[0006] A method for supervising stereoscopic vision training in a polarization mode, comprising:

[0007] Acquire real-time visual training images and corresponding real-time eye movement parameters of the patient during the visual training process. The real-time eye movement parameters include fixation point coordinates, saccade rate, and pupil diameter.

[0008] Based on the real-time visual training image, the corresponding standard eye-tracking parameters are determined, and based on the real-time eye-tracking parameters and the standard eye-tracking parameters, the first training score is determined.

[0009] The simulated fusion image is determined based on the preset left and right eye parallax and the real-time eye movement parameters. The standard deviation of the fixation point in the simulated fusion image is calculated, and the second training score is determined based on the standard deviation of the fixation point.

[0010] The total training score is determined based on the first training score and the second training score. When the total training score is lower than a preset threshold, the visual training weaknesses and their degree of weakness are determined based on the real-time visual training image and the simulated fused image.

[0011] Based on the weaknesses and degree of weakness in the visual training, the visual training images corresponding to the patient in the next training stage are optimized.

[0012] By adopting the above technical solution, and comparing real-time eye-tracking parameters with preset standard eye-tracking parameters, it is convenient to intuitively assess the accuracy and efficiency of patients in the vision training process. By calculating the standard deviation of fixation points in the simulated fusion images, it is convenient to evaluate the performance of patients in complex visual tasks. This dynamic evaluation mechanism facilitates the timely detection of deviations and deficiencies in patients' vision training. By quantifying the deviations and deficiencies in patients' vision training as weak points and degrees of weakness, it is convenient to intelligently optimize the vision training images for the next training stage. Through timely and targeted optimization, it is convenient to ensure that subsequent vision training images better meet the actual needs of patients, thereby helping to improve the vision training effect.

[0013] In one possible implementation, determining the visual training weaknesses and their severity based on the real-time visual training images and the simulated fused images includes:

[0014] The offset point is determined based on the real-time visual training image and the simulated fused image, and the initial offset degree corresponding to the offset point is determined based on the offset result. The offset point is the gaze point with coordinate deviation.

[0015] The offset stereometry and offset depth corresponding to each offset point are identified from the real-time visual training image. Based on the offset stereometry and offset depth of each offset point, the delay influence range of each offset point is determined. Based on the delay influence range of all offset points, the offset influence value corresponding to each offset point is determined.

[0016] The initial offset degree is adjusted based on the offset influence value corresponding to each offset point to obtain the optimized offset degree. Offset points with optimized offset degrees higher than the preset offset degree threshold are identified as weak points in visual training, and the corresponding optimized offset degree is identified as the degree of weakness.

[0017] By employing the aforementioned technical solution, and combining real-time visual training images with simulated fused images to accurately locate offset points, it is possible to capture subtle deviations that occur during visual training. By analyzing the offset stereoscopicity and offset depth of each offset point, it is possible to further quantify the impact of the patient's recognition or tracking deviation at the offset point on subsequent visual training. Since there may be mutual influence between adjacent or close offset points, by comprehensively considering the delay influence range of all offset points, it is possible to quantify the potential mutual influence between offset points, thereby facilitating a more comprehensive assessment of the offset degree of each offset point. Finally, based on a preset offset degree threshold, the final visual training weak points are selected from all offset points. By analyzing the mutual influence between adjacent or close offset points, all offset points are screened, rather than identifying all fixation points with coordinate deviations as visual training weak points. Screening offset points improves the accuracy of identifying visual training weak points.

[0018] In one possible implementation, determining the offset impact value for each offset point based on the delay impact range corresponding to all offset points includes:

[0019] Based on the delay influence range corresponding to each offset point, construct the corresponding offset layer, and overlay all offset layers to determine the overlapping area and the overlapping pixel value corresponding to each overlapping area;

[0020] Identify the path interval between each offset point and each overlapping region, and determine the offset influence value corresponding to each offset point based on each path interval and the corresponding overlapping pixel value.

[0021] By adopting the above technical solution and constructing an offset layer corresponding to each offset point, it is easy to visualize the range of delay impact of each offset point. This allows relevant supervisors to intuitively view the range and extent of impact after an anomaly occurs at each offset point. By overlaying all offset layers, the overlapping area and the corresponding overlapping pixel value are determined, making it easy to intuitively view the common impact between offset points and the intensity of the common impact. By analyzing the path interval and corresponding overlapping pixel value between each offset point and the overlapping area, the accuracy of determining the offset impact of the overlapping area on each offset point is improved.

[0022] In one possible implementation, after identifying the offset stereo and offset depth corresponding to each offset point from the real-time visual training image, the method further includes:

[0023] Integrate the fixed time and duration of each offset point within a preset time period, and determine the first screening score for each offset point based on the fixed time and duration of each offset point;

[0024] Based on the offset solidity and offset depth corresponding to each first screening offset point, determine the second screening score for each offset point;

[0025] Based on the first and second screening scores for each offset point, all offset points within the preset time period are filtered to obtain the final offset points.

[0026] By adopting the above technical solution, since the patient's concentration during vision training also affects the patient's tracking and recognition of fixation points, after determining the offset points, not all offset points are used as the basis for screening weak points in vision training. Instead, the determination time and duration of each offset point are analyzed to facilitate the screening of offset points. During the screening process, the complexity of each offset point is analyzed by analyzing the offset stereoscopicity and offset depth. By comprehensively considering the determination time, duration of continuous offset, and complexity of each offset point, the accuracy of the screening results can be improved.

[0027] In one possible implementation, optimizing the visual training images for the patient in the next training stage based on the visual training weaknesses and their severity includes:

[0028] Acquire historical training data and identify the sensitive features corresponding to each visual training weakness from the historical training data;

[0029] Based on the aforementioned visual training weaknesses and corresponding sensitive features, optimize the training images;

[0030] Based on the degree of weakness, an optimized training type is determined. Based on the optimized training image and the optimized training type, the visual training image corresponding to the patient in the next training stage is optimized.

[0031] By adopting the above technical solution, historical training data is acquired, and sensitive features corresponding to each visual training weakness are identified. This facilitates in-depth analysis of training weaknesses. Since historical training data contains the patient's training performance and reaction patterns over a period of time, analyzing and mining this data makes it easier to accurately locate sensitive features that significantly affect visual training weaknesses. Based on visual training weaknesses and sensitive features, optimized training images are determined, which improves the fit between subsequent visual training images and the patient's actual situation. In addition, timely adjustment and optimization of training types based on the degree of weakness helps improve the patient's concentration during visual training and reduces fatigue caused by using fixed training types.

[0032] In one possible implementation, the method further includes:

[0033] Obtain the total training score for each observation time within the observation period, and fit the total training score for each observation time to obtain the score trend.

[0034] The selection type for the patient is determined based on the score trend, and the selection type includes passive selection and active selection.

[0035] When the selection type is active selection, the selection item is determined based on the eye movement parameters corresponding to each observation time, and the training mode type corresponding to each selection item is different.

[0036] A selection list is generated based on the selection options, and the selection list is superimposed on the visual training image corresponding to the next training time. A prompt is generated to remind the patient to choose the training mode type from the selection list.

[0037] Obtain the selected eye-tracking parameters corresponding to the selected time period, determine the target selection item corresponding to the patient based on the selected eye-tracking parameters, and adjust the visual training image for the next training stage based on the training mode type corresponding to the target selection item.

[0038] By adopting the above technical solution, the total training score corresponding to each observation moment within the observation period is obtained, and the score trend is fitted to these scores. This facilitates the phased evaluation of the patient's training effect. Based on the phased evaluation effect, the patient can be provided with a choice type. When the patient can choose the active choice type, the selection options can be determined by analyzing the observation eye movement parameters corresponding to each observation moment for the patient to choose independently. Based on the patient's independent choice, the visual training images in the subsequent training stages are adjusted, which helps to improve the patient's enthusiasm and autonomy in the vision training process. By guiding the patient to actively participate in the training process, the training effect is improved.

[0039] Secondly, this application provides a monitoring system, which adopts the following technical solution:

[0040] A monitoring system comprising:

[0041] At least one processor;

[0042] Memory;

[0043] At least one application, wherein the at least one application is stored in memory and configured to be executed by at least one processor, the at least one application being configured to: execute the stereoscopic vision training supervision method described above in the polarization mode.

[0044] Thirdly, this application provides a computer-readable storage medium, which adopts the following technical solution:

[0045] A computer-readable storage medium includes: a computer program storing a method for supervising stereoscopic vision training under the aforementioned polarization mode that can be loaded by a processor and executed.

[0046] Fourthly, this application provides a computer program product, which adopts the following technical solution:

[0047] A computer program product includes a computer program that, when executed by a processor, implements the stereoscopic vision training supervision method under the aforementioned polarization mode.

[0048] In summary, this application includes at least one of the following beneficial technical effects:

[0049] By comparing real-time eye-tracking parameters with preset standard eye-tracking parameters, the accuracy and efficiency of patients during vision training can be intuitively assessed. By calculating the standard deviation of fixation points in simulated fusion images, the performance of patients in complex visual tasks can be evaluated. This dynamic assessment mechanism facilitates the timely detection of deviations and deficiencies in patients' vision training. By quantifying these deviations and deficiencies into weak points and degrees of weakness in vision training, the visual training images for the next training stage can be intelligently optimized. Timely and targeted optimization ensures that subsequent visual training images better meet the actual needs of patients, thereby helping to improve the effectiveness of vision training.

[0050] By combining real-time visual training images with simulated fused images to accurately locate offset points, it is possible to capture subtle deviations that occur in patients during visual training. By analyzing the offset stereometry and offset depth of each offset point, it is possible to further quantify the impact of recognition or tracking deviations at offset points on subsequent visual training. Since adjacent or nearby offset points may influence each other, by comprehensively considering the delay influence range of all offset points, it is possible to quantify the potential mutual influence between offset points, thereby facilitating a more comprehensive assessment of the offset degree of each offset point. Finally, based on a preset offset degree threshold, the final visual training weak points are selected from all offset points. By analyzing the mutual influence between adjacent or nearby offset points, all offset points are screened, rather than identifying all fixation points with coordinate deviations as visual training weak points. Screening offset points improves the accuracy of identifying visual training weak points. Attached Figure Description

[0051] Figure 1 This is a flowchart illustrating a stereoscopic vision training supervision method under polarization mode in an embodiment of this application.

[0052] Figure 2 This is a schematic diagram of a process for determining vulnerability information in an embodiment of this application;

[0053] Figure 3 This is a schematic diagram of the structure of a monitoring system according to an embodiment of this application. Detailed Implementation

[0054] The following is in conjunction with the appendix Figures 1 to 3 This application will be described in further detail.

[0055] After reading this specification, those skilled in the art may make modifications to this embodiment without contributing any inventive step, but such modifications are protected by patent law as long as they fall within the scope of the claims of this application.

[0056] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, 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, 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.

[0057] It should be noted that, in the optional embodiments of this application, the data related to object information, when applied to specific products or technologies, requires the permission or consent of the object. Furthermore, the collection, use, and processing of this data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. In other words, if the embodiments of this application involve data related to an object, it must be obtained with the object's authorization and consent, the authorization and consent of relevant departments, and in accordance with the relevant laws, regulations, and standards of the country and region. If the embodiments involve personal information, the acquisition of all personal information requires the individual's consent. If sensitive information is involved, the separate consent of the information subject is required. The embodiments also need to be implemented with the object's authorization and consent.

[0058] Specifically, this application provides a stereoscopic vision training supervision method in polarization mode, executed by a supervision system. This supervision system can be a server or a terminal device. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. The terminal device can be a smartphone, tablet, laptop, desktop computer, etc., but is not limited to these. The terminal device and the server can be directly or indirectly connected via wired or wireless communication, and this application does not impose any limitations on this.

[0059] refer to Figure 1 , Figure 1This is a flowchart illustrating a stereoscopic vision training supervision method in polarization mode according to an embodiment of this application. The method includes steps S110-S150, wherein:

[0060] Step S110: Acquire real-time visual training images and corresponding real-time eye movement parameters of the patient during the visual training process. The real-time eye movement parameters include real-time fixation coordinates, real-time saccade rate, and real-time pupil diameter.

[0061] Specifically, in this application, patients need to wear polarized glasses for visual training. These glasses can be used in conjunction with polarized 3D display technology to provide independent visual training images for the patient's left and right eyes. This technology is fundamental to stereoscopic vision training and helps train the coordinated working ability of the left and right eyes. Polarized glasses utilize the principle of polarized light, filtering out light from certain directions through specific lenses, allowing only light from specific directions to pass through. This helps reduce glare during visual training, providing patients with clearer visual input and enabling them to more accurately capture and analyze information in the visual training images. The real-time visual training image is the image displayed in the polarized glasses at the current moment. In the initial stage of visual training, the visual training images in the polarized glasses are generally played or switched according to a preset training plan. The preset training plan is related to the patient's actual vision, training needs, and training stage. It can be set by relevant diagnostic personnel based on the patient's actual situation and uploaded to the monitoring system. The monitoring system then determines the appropriate training image and controls the polarized glasses to play or switch it. Therefore, the real-time visual training image of the polarized glasses can be determined through the monitoring system.

[0062] Real-time eye-tracking parameters are various indicators or data related to eye movements that are monitored or recorded in real time by professional acquisition equipment such as eye trackers during the visual training process. These include, but are not limited to, fixation point coordinates, saccade rate, and pupil diameter. Real-time eye-tracking parameters can reflect the patient's eye movement state, attention allocation, and visual cognitive process during visual training. Among these, real-time fixation coordinates refer to the coordinates of the patient's eye position in the visual training image at the current moment. Analyzing fixation coordinates helps to understand the patient's focus area during visual training, thereby facilitating the analysis of their visual attention and information processing patterns. Real-time saccade rate refers to the speed at which the patient's eye moves from one fixation point to another. Saccade rate reflects the flexibility and speed of the patient's eye movements and is an important indicator for assessing the patient's eye movement ability. Real-time pupil diameter refers to the size of the patient's pupil at the current moment, usually measured in millimeters. Changes in pupil diameter reflect the patient's emotional state, cognitive load, and the intensity of visual stimuli at the current moment. In visual training, changes in pupil diameter help to understand the patient's response to and adaptation to the training content. Therefore, it is necessary to examine and analyze the patient's fixation coordinates, saccade rate, and pupil diameter during visual training supervision.

[0063] Step S120: Determine the corresponding standard eye-tracking parameters based on the real-time visual training images, and determine the first training score based on the real-time eye-tracking parameters and the standard eye-tracking parameters.

[0064] Specifically, each real-time visual training image corresponds to standard eye-tracking parameters, including but not limited to standard fixation point coordinates, standard saccade rate, and standard pupil diameter. These parameters can be uploaded to the supervision system in advance by relevant staff based on historical vision training experience. The standard eye-tracking parameters and the corresponding visual training images are mutually bound. Based on the binding identifiers contained in the real-time visual training images, the corresponding standard eye-tracking parameters can be obtained quickly and accurately.

[0065] By comparing real-time eye-tracking parameters with standard eye-tracking parameters, the differences between the two can be calculated. For example, the coordinate difference between real-time fixation point coordinates and standard fixation point coordinates, the rate difference between real-time saccade rate and standard saccade rate, and the diameter difference between real-time pupil diameter and standard pupil diameter can be calculated. After calculating the coordinate difference, rate difference, and diameter difference, the three are normalized to ensure they are at the same quantization level. Then, the quantization results are weighted based on preset difference weights to obtain the corresponding first training score. The smaller the coordinate difference, rate difference, and diameter difference, the higher the corresponding first training score. The specific preset difference weights are not specifically limited in this embodiment and can be determined by relevant personnel based on historical visual training data and uploaded to the supervision system.

[0066] Step S130: Determine the simulated fusion image based on the preset left and right eye disparity and real-time eye movement parameters, calculate the standard deviation of the fixation point in the simulated fusion image, and determine the second training score based on the standard deviation of the fixation point.

[0067] Specifically, the preset left-right eye disparity is the foundation for stereoscopic vision formation. It refers to the slight difference in the horizontal direction between the images seen by the left and right eyes during visual training. The specific preset left-right eye disparity is not limited in this embodiment and can be set according to the patient's actual training needs and visual condition. When determining the simulated fusion image, the patient's current gaze area can be located based on the gaze point coordinates in the real-time eye-tracking parameters. Then, based on the preset left-right eye disparity, the corresponding position and disparity value of the gaze area in the patient's left and right eye images are calculated. During this process, a stereo matching algorithm can be used to generate a disparity map to obtain more accurate disparity information. The stereo matching algorithm can be SGBM or AD-Census; the specific stereo matching algorithm is not specifically limited in this embodiment. After determining the disparity information, the left and right eye images can be deformed based on the disparity information to generate virtual viewpoint images. Specifically, the left eye image can be shifted to the right by the disparity value, and the right eye image can be shifted to the left by the disparity value to simulate the difference in visual perspective between the two eyes. Finally, the deformed left and right eye images can be fused based on a preset fusion algorithm to obtain a simulated fused image. The preset fusion algorithm can be a linear weighted fusion algorithm, a deep learning fusion algorithm, etc. The specific preset fusion algorithm is not specifically limited in this embodiment.

[0068] The standard deviation of fixation points reflects a patient's adaptation to the current visual training image. A large standard deviation may indicate that the patient may have difficulty fusing binocular images or is not adapted to the difficulty of the current visual training image. Therefore, calculating the standard deviation of fixation points corresponding to the simulated fusion image can help analyze the actual situation faced by the patient during vision training, facilitating subsequent adjustments or optimizations to the visual training content. The calculation of the standard deviation of fixation points corresponding to the simulated fusion image involves the following steps: 1. Calculating the mean: Calculate the average values ​​of all fixation point coordinates along the X and Y axes. These means characterize the center position of the fixation point during multiple training sessions. 2. Calculating the variance: For each fixation point coordinate, calculate the square of the difference between it and the corresponding axis mean, then calculate the average of these squared differences to obtain the variance along the X and Y axes. 3. Calculating the standard deviation of fixation points: Take the square root of the variance to obtain the standard deviation of fixation points along the X and Y axes. The standard deviation of fixation points reflects the dispersion of fixation points in the simulated fusion image.

[0069] Finally, the second training score can be determined based on the preset score mapping relationship and the fixation point standard deviation. The preset score mapping relationship is the correspondence between the fixation point standard deviation and the second training score. The smaller the fixation point standard deviation, the more stable the patient's attention is during the visual training process, and the higher the corresponding second training score. The specific content of the preset score mapping relationship is not specifically limited in this application embodiment, and can be determined by relevant personnel based on historical experimental data and then uploaded to the detection system.

[0070] Step S140: Determine the total training score based on the first training score and the second training score. When the total training score is lower than a preset threshold, determine the weak points and degree of weakness in visual training based on the real-time visual training image and the simulated fused image.

[0071] Specifically, the first training score and the second training score are summed to obtain the total training score. A higher total training score indicates higher quality vision training for the patient. When the total training score is lower than a preset threshold, it indicates poor tracking and recognition of fixation points in the visual training images. In this case, the weaknesses and degrees of weakness in the patient's vision training during fixation point recognition or tracking can be determined by analyzing real-time visual training images and simulated fused images, so as to facilitate targeted adjustments to the visual training images in the future. The specific preset threshold is not specifically limited in this embodiment and can be determined by relevant personnel based on historical experimental data and uploaded to the monitoring system. During vision training, patients need to complete a series of visual tasks, such as fixation tracking, stereo perception, and dynamic tracking. By recording and analyzing the patient's performance in completing these visual tasks, deficiencies in their visual function can be identified, namely, fixations with recognition or tracking deviations. To determine weak points in vision training, the standard fixation coordinates corresponding to planar and stereo fixations in real-time training images and simulated fused images can be directly identified. These standard fixation coordinates are then matched with the patient's real-time fixation coordinates during vision training. Based on the matching results, the weak points and their severity in vision training are determined. Furthermore, to improve the accuracy of determining weak points in vision training, the technical solution provided in this application, based on real-time vision training images and simulated fused images, determines the weak points and their severity in vision training, specifically including steps S210-S230, such as... Figure 2 As shown, where:

[0072] Step S210: Determine the offset point based on the real-time visual training image and the simulated fused image, and determine the initial offset degree corresponding to the offset point based on the offset result. The offset point is the gaze point with coordinate deviation.

[0073] Specifically, since the real-time visual training image is the image displayed in polarized glasses, the standard planar gaze point in the real-time visual training image can be identified through feature recognition. After image fusion processing, the standard fused image corresponding to the real-time visual training image can also be obtained. The method for determining the standard fused image can refer to the method for determining the simulated fused image in the above embodiments, and will not be repeated here. Similarly, the standard stereo gaze point in the standard fused image can also be obtained through feature recognition. The specific feature recognition method is not specifically limited in the embodiments of this application.

[0074] Based on the patient's real-time eye movement parameters and simulated fusion images, the actual planar fixation point and the actual stereo fixation point can be identified. By comparing the coordinates of the actual planar fixation point with the coordinates of the standard planar fixation point, the actual planar fixation point with planar coordinate deviation can be determined. Similarly, by comparing the coordinates of the actual stereo fixation point with the coordinates of the standard stereo fixation point, the actual stereo fixation point with stereo coordinate deviation can be determined, i.e., the offset point. The larger the coordinate deviation, the greater the degree of offset.

[0075] Step S220: Identify the offset stereo and offset depth corresponding to each offset point from the real-time visual training image, and determine the delay influence range of each offset point based on the offset stereo and offset depth of each offset point, and determine the offset influence value corresponding to each offset point based on the delay influence range of all offset points.

[0076] Specifically, when the offset point is a stereo fixation point, the final degree of offset cannot be assessed solely using coordinate deviation. Instead, it is necessary to further analyze the complexity of the offset point and adjust or optimize the offset degree of the corresponding stereo fixation point based on this complexity. In this embodiment, the offset stereometry and offset depth of the offset point can be used to characterize its complexity. The offset stereometry characterizes the stereoscopic nature of the corresponding offset point in the simulated fused image and can be identified from the simulated fused image based on block matching algorithms, semi-global matching algorithms, or deep learning models. The offset depth is the distance between the offset point and the patient's pupil. It can be determined by converting the simulated fused image into a depth map and then identifying the depth value corresponding to the offset point in the depth map. The specific methods for determining the offset stereometry and offset depth are not specifically limited in this embodiment.

[0077] For any offset point, because the patient's recognition deviation when identifying or tracking fixation points with strong stereo perception and strong depth perception may affect the patient's recognition and tracking of subsequent fixation points, it is necessary to determine the potential range caused by the offset point based on the offset stereometry and offset depth. The influence radius corresponding to the offset stereometry and offset depth can be determined first according to a preset parameter mapping relationship. The preset parameter mapping relationship is the correspondence between the parameter combination of offset stereometry and offset depth and the influence radius. Specific details are not limited in this embodiment. The delay influence range corresponding to the offset point is a circle centered on the offset point with the influence radius as its radius. There is a conversion relationship between the influence radius and the offset influence value. Based on this conversion relationship, the offset influence value corresponding to any influence radius can be determined. The larger the influence radius, the larger the corresponding offset influence value. Using the above method, the delay influence range and offset influence value corresponding to each offset point can be determined.

[0078] After determining the delay influence range corresponding to each offset point, the impact of each delay influence range on other offset points can be analyzed. For example, in visual training, point a needs to be tracked first, and then point b needs to be tracked. After determining the delay influence range corresponding to point a, it is necessary to first determine whether point b is within the delay influence range of point a. If so, the offset of point b needs to be reduced based on the delay influence range of point a, that is, to eliminate the impact on the recognition or tracking of point b caused by the patient's offset in recognizing or tracking point a. Generally, the offset degree of point b can be reduced based on the relative distance between the delay influence range corresponding to point a and point b. Furthermore, in order to improve the accuracy of determining the offset influence value corresponding to each offset point, the technical method provided in this application embodiment, when determining the offset influence value corresponding to each offset point based on the delay influence range corresponding to all offset points, may specifically include:

[0079] An offset layer is constructed based on the delay influence range corresponding to each offset point, and all offset layers are superimposed to determine the overlapping area and the overlapping pixel value corresponding to each overlapping area; the path interval between each offset point and each overlapping area is identified, and the offset influence value corresponding to each offset point is determined based on each path interval and the corresponding overlapping pixel value.

[0080] Specifically, for any offset point, a region containing the delay-affected area can be identified from the visual training image corresponding to the offset point based on a preset feature recognition algorithm. Combining this region image with a preset layer yields the offset layer corresponding to the offset point. To facilitate subsequent overlay processing between offset layers, the same preset layer can be used when processing different offset points. Using the above method, the offset layer corresponding to each offset point can be obtained. A linear weighted sum can be used for layer overlay processing to ensure that each overlapping pixel value in the overlapping region can represent the combined effect of multiple offset layers. The overlapping pixel value in the overlapping region can be the average pixel value of all pixels in the overlapping region.

[0081] The center point of each overlapping region is identified, and the path interval between the center point and the offset point is calculated. The specific method for determining the path interval between two points is not detailed in this embodiment. Since the same offset point may be affected by multiple overlapping regions, it is necessary to calculate the path interval between each offset point and each preceding overlapping region according to the visual training sequence. Finally, a preset weight and weighted summation method are used to calculate the offset influence value of each path interval and the corresponding delay influence range corresponding to the overlapping pixel value on the offset point. The specific preset weights can be set by relevant personnel according to actual needs.

[0082] Step S230: Adjust the initial offset degree based on the offset influence value corresponding to each offset point to obtain the optimized offset degree. Identify the offset points with optimized offset degrees higher than the preset offset degree threshold as weak points in visual training, and determine the corresponding optimized offset degree as the weakness degree.

[0083] Specifically, each initial offset degree can be converted into an offset value. After calculating the offset influence value corresponding to each offset point, the offset value corresponding to each initial offset degree is calculated based on the offset influence value to obtain the optimized offset value. Finally, the optimized offset value is converted back into the optimized offset degree. A higher optimized offset degree indicates a more severe deviation in the patient's visual training process. Therefore, to improve the effectiveness of identifying weak points in visual training, offset points with optimized offset degrees higher than a preset offset degree threshold can be considered as weak points in visual training, and the optimized offset degree is determined as the weakness level of the weak point. Instead of identifying all fixation points with coordinate deviations as weak points in visual training, all offset points are screened by analyzing the mutual influence between adjacent or close offset points. This screening of offset points improves the accuracy of identifying weak points in visual training.

[0084] Step S150: Based on the weaknesses and degree of weakness in visual training, optimize the visual training images corresponding to the patient in the next training stage.

[0085] Specifically, after identifying weaknesses in visual training, target images can be selected based on the degree of weakness. These target images replace the corresponding visual training images for the next training stage, facilitating the updating of the patient's visual training content in the next stage. The monitoring system stores suitable target images for each visual training weakness at different degrees of weakness. These images can be determined in advance by relevant personnel based on historical experimental data and uploaded to the monitoring system. Furthermore, to improve the fit between subsequent visual training images and the patient's actual situation, this embodiment of the application, when optimizing the patient's visual training images for the next training stage based on the visual training weaknesses and their degree, may specifically include:

[0086] Acquire historical training data and identify the sensitive features corresponding to each visual training weakness from the historical training data; determine the optimized training images based on the visual training weaknesses and corresponding sensitive features; determine the optimized training type based on the degree of weakness; and optimize the visual training images for the patient in the next training stage based on the optimized training images and optimized training type.

[0087] Specifically, the historical training data includes training records of patients during visual training within a historical time period. These records contain sensitive features corresponding to each fixation point—features that demonstrate optimal training effectiveness during vision training. These features include, but are not limited to, target velocity, background complexity, and disparity gradient. A pre-defined feature recognition algorithm can determine the total training score corresponding to different features when a patient undergoes visual training based on their visual training weaknesses. Features contained in the visual training images corresponding to visual training weaknesses with total training scores exceeding a pre-defined threshold are identified as sensitive features. After identifying the sensitive features corresponding to visual training weaknesses, optimized training images containing these sensitive features can be traversed from the supervised system. The supervised system contains optimized training images with different sensitive features, which are uploaded in advance by relevant personnel.

[0088] Different levels of weakness correspond to different optimization training types. For example, when the weakness is level one, the corresponding optimization training type is stereo vision training; when the weakness is level three, the corresponding optimization training type is level two dynamic tracking training. Different levels of weakness are suitable for different optimization training types. The optimization training type corresponding to any level of weakness can be determined based on a preset type mapping relationship. The specific content of the preset type mapping relationship is not specifically limited in this application embodiment.

[0089] Since historical training data contains patients' training performance and response patterns over a period of time, analyzing and mining this data makes it easier to accurately identify sensitive features that significantly affect weaknesses in visual training. Based on these weaknesses and sensitive features, optimized training images can be determined, which helps improve the fit between subsequent visual training images and the patient's actual situation. In addition, timely adjustments to the training type based on the degree of weakness can improve the patient's focus during visual training and reduce fatigue caused by using fixed training types.

[0090] In this embodiment of the application, by comparing real-time eye-tracking parameters with preset standard eye-tracking parameters, it is convenient to intuitively evaluate the accuracy and efficiency of the patient's vision training process. By calculating the standard deviation of fixation points in the simulated fusion image, it is convenient to evaluate the patient's performance in complex visual tasks. This dynamic evaluation mechanism facilitates the timely detection of deviations and deficiencies in the patient's vision training process. By quantifying the deviations and deficiencies in the patient's vision training process into weak points and degrees of weakness in vision training, it is convenient to intelligently optimize the vision training images for the next training stage. Through timely and targeted optimization, it is convenient to ensure that subsequent vision training images better meet the actual needs of the patient, thereby helping to improve the vision training effect.

[0091] Since a patient's concentration during vision training also affects their ability to track and recognize fixation points, the method does not use all offset points as the basis for screening weak points in vision training after identifying them. Instead, it analyzes the time of determination and duration of each offset point to facilitate screening. The technical solution provided in this application identifies the offset stereometry and offset depth corresponding to each offset point from real-time vision training images, and then further includes:

[0092] The system integrates the fixed time and duration of each offset point within a preset time period, and determines the first screening score for each offset point based on the fixed time and duration of each offset point. Based on the offset three-dimensionality and offset depth corresponding to each first screening offset point, the system determines the second screening score for each offset point. Based on the first and second screening scores of each offset point, the system filters all offset points within the preset time period to obtain the final offset points.

[0093] Specifically, the preset time period is a period of time after the offset stereometry and offset depth corresponding to each offset point are identified from the real-time visual training image. The duration of the preset time period can be 60 seconds or 120 seconds, and the specific duration is not specifically limited in this embodiment. Based on the eye movement parameters corresponding to the preset time period, the determination time of each offset point and the duration of continuous offset at each offset point can be determined. Based on the determination time of each offset point within the preset time period, the offset frequency corresponding to each offset point can be determined. Based on the duration of continuous offset at each offset point, the average offset duration of each offset point can be determined. Based on the mapping relationship of the first screening parameters, the first screening score corresponding to the parameter combination of offset frequency and average offset duration of each offset point can be determined. Based on the second screening parameter mapping relationship, the second screening score corresponding to the parameter combination of offset solidity and offset depth for each offset point can be determined. The first screening parameter mapping relationship is the correspondence between the parameter combination of offset frequency and average offset duration and the first screening score. The second screening parameter mapping relationship is the correspondence between the parameter combination of offset solidity and offset depth and the second screening score. The specific first screening parameter mapping relationship and the second screening parameter mapping relationship are not specifically limited in this application embodiment. They can be determined by relevant personnel based on historical experimental data and then uploaded to the supervision system.

[0094] The sum of the first and second screening scores for each offset point is calculated to obtain the total score for each offset point. Offset points with a total score lower than a preset score threshold are then removed to obtain the final offset points. The specific preset score threshold is not limited in this embodiment; however, considering the determination time, duration of the offset, and complexity of each offset point helps improve the accuracy of the screening results.

[0095] Furthermore, to enhance patients' motivation and autonomy during vision training, the method provided in this application embodiment also includes:

[0096] The system obtains the total training score for each observation moment within the observation period and fits the score trend for each observation moment. Based on the score trend, it determines the patient's selection type, which includes passive and active selection. When the selection type is active selection, it determines the selection item based on the observation eye-tracking parameters for each observation moment, with each selection item corresponding to a different training mode type. It generates a selection list based on the selection items and overlays it onto the visual training image for the next training moment, while generating a prompt to remind the patient to independently select the training mode type from the selection list. It obtains the selection eye-tracking parameters for the selection period, determines the patient's target selection item based on the selection eye-tracking parameters, and adjusts the visual training image for the next training stage based on the training mode type corresponding to the target selection item.

[0097] Specifically, the observation period is a time preceding the current moment. The duration of this observation period can be 10 minutes or 30 minutes; the specific duration is not specifically limited in this embodiment and can be determined by relevant personnel based on historical experimental data before being uploaded to the monitoring system. A sliding window linear regression can be used to calculate the slope of the score change. For example, if the slope is greater than 0.5 and persists for 3 windows, the score trend is determined to be upward; if the slope is not greater than 0.5, the score trend is determined to be downward. During this process, median filtering can also be used to remove instantaneous fluctuations, such as a sudden drop in score due to blinking.

[0098] When the score trend is upward, it can be determined that the patient's selection type is active selection. At this time, the selection options that the patient can actively choose can be determined based on the observation eye movement parameters at each observation time during the observation period. Each selection option corresponds to a different training mode type, which includes, but is not limited to, stereo vision training, annotation control training, dynamic tracking training, visual attention training, and visual memory training. Since the observation eye movement parameters at each observation time during the observation period can reflect the patient's follow-up training status during the observation period, different follow-up training statuses can select different training mode types. The observation eye movement parameters at each observation time can be quantified to obtain the patient's follow-up training status value. Then, the selection options that the patient can choose can be determined based on the preset mode type mapping relationship. The preset mode type mapping relationship is the correspondence between the follow-up training status value and the selection option.

[0099] After identifying at least one option, a selection list containing all options can be presented as a semi-transparent overlay below the screen of the polarized glasses in a non-disruptive layout. A prompt is generated to remind the patient to independently select the training model type from the list. The prompt can be text or an external voice reminder; the specific format is not limited in this embodiment. The selection period is a time interval after the selection list is generated. The duration of the selection period can be 3 seconds or 5 seconds; the specific duration is not limited in this embodiment. For example, after the selection list is presented, a 5-second selection period is opened. During this period, the patient's eye-tracking data is continuously recorded. If the patient gazes at an option for more than 3 seconds, selection confirmation is automatically triggered. Based on the patient's independent selection, the visual training images in subsequent training stages are adjusted to match the training mode type corresponding to the target selection. This improves the patient's enthusiasm and autonomy during vision training, thereby enhancing the training effect by guiding their active participation in the training process.

[0100] This application provides a monitoring system, such as... Figure 3 As shown, Figure 3The monitoring system 300 shown includes a processor 301 and a memory 303. The processor 301 and the memory 303 are connected, for example, via a bus 302. Optionally, the monitoring system 300 may also include a transceiver 304. It should be noted that in practical applications, the transceiver 304 is not limited to one, and the structure of this monitoring system 300 does not constitute a limitation on the embodiments of this application.

[0101] Processor 301 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 301 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0102] Bus 302 may include a pathway for transmitting information between the aforementioned components. Bus 302 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 302 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 3 The symbol is represented by only one line, but this does not mean that there is only one bus or one type of bus.

[0103] The memory 303 may be a ROM (Read Only Memory) or other type of static storage device capable of storing static information and instructions, RAM (Random Access Memory) or other type of dynamic storage device capable of storing information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0104] The memory 303 is used to store application code that executes the solution of this application, and its execution is controlled by the processor 301. The processor 301 is used to execute the application code stored in the memory 303 to implement the content shown in the foregoing method embodiments.

[0105] The monitoring system includes, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and in-vehicle terminals (such as in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. It can also include servers. Figure 3 The monitoring system shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0106] This application provides a computer-readable storage medium storing a computer program that, when run on a computer, enables the computer to execute the corresponding content in the aforementioned method embodiments.

[0107] This application provides a computer program product including a computer program that, when executed by a processor, implements the methods described in any of the above embodiments.

[0108] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0109] The above description is only a partial embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for supervising stereoscopic vision training in polarization mode, characterized in that, include: Acquire real-time visual training images and corresponding real-time eye movement parameters of the patient during the visual training process. The real-time eye movement parameters include fixation point coordinates, saccade rate, and pupil diameter. Based on the real-time visual training image, the corresponding standard eye-tracking parameters are determined, and based on the real-time eye-tracking parameters and the standard eye-tracking parameters, the first training score is determined. The simulated fusion image is determined based on the preset left and right eye parallax and the real-time eye movement parameters. The standard deviation of the fixation point in the simulated fusion image is calculated, and the second training score is determined based on the standard deviation of the fixation point. The total training score is determined based on the first training score and the second training score. When the total training score is lower than a preset threshold, the visual training weaknesses and their degree of weakness are determined based on the real-time visual training image and the simulated fused image. Based on the weaknesses and degree of weakness in the visual training, the visual training images corresponding to the patient in the next training stage are optimized.

2. The stereoscopic vision training supervision method in polarization mode according to claim 1, characterized in that, The step of determining the weaknesses and degree of weakness in visual training based on the real-time visual training images and the simulated fused images includes: The offset point is determined based on the real-time visual training image and the simulated fused image, and the initial offset degree corresponding to the offset point is determined based on the offset result. The offset point is the gaze point with coordinate deviation. The offset stereometry and offset depth corresponding to each offset point are identified from the real-time visual training image. Based on the offset stereometry and offset depth of each offset point, the delay influence range of each offset point is determined. Based on the delay influence range of all offset points, the offset influence value corresponding to each offset point is determined. The initial offset degree is adjusted based on the offset influence value corresponding to each offset point to obtain the optimized offset degree. Offset points with optimized offset degrees higher than the preset offset degree threshold are identified as weak points in visual training, and the corresponding optimized offset degree is identified as the degree of weakness.

3. The stereoscopic vision training supervision method in polarization mode according to claim 2, characterized in that, The step of determining the offset impact value for each offset point based on the delay impact range corresponding to all offset points includes: Based on the delay influence range corresponding to each offset point, construct the corresponding offset layer, and overlay all offset layers to determine the overlapping area and the overlapping pixel value corresponding to each overlapping area; Identify the path interval between each offset point and each overlapping region, and determine the offset influence value corresponding to each offset point based on each path interval and the corresponding overlapping pixel value.

4. The stereoscopic vision training supervision method in polarization mode according to claim 2, characterized in that, The step of identifying the offset stereo and offset depth corresponding to each offset point from the real-time visual training image further includes: Integrate the fixed time and duration of each offset point within a preset time period, and determine the first screening score for each offset point based on the fixed time and duration of each offset point; Based on the offset solidity and offset depth corresponding to each first screening offset point, determine the second screening score for each offset point; Based on the first and second screening scores for each offset point, all offset points within the preset time period are filtered to obtain the final offset points.

5. The stereoscopic vision training supervision method in polarization mode according to claim 2, characterized in that, The step of optimizing the visual training images for the patient in the next training stage based on the aforementioned visual training weaknesses and their severity includes: Acquire historical training data and identify the sensitive features corresponding to each visual training weakness from the historical training data; Based on the aforementioned visual training weaknesses and corresponding sensitive features, optimize the training images; Based on the degree of weakness, an optimized training type is determined. Based on the optimized training image and the optimized training type, the visual training image corresponding to the patient in the next training stage is optimized.

6. The stereoscopic vision training supervision method in polarization mode according to claim 1, characterized in that, Also includes: Obtain the total training score for each observation time within the observation period, and fit the total training score for each observation time to obtain the score trend. The selection type for the patient is determined based on the score trend, and the selection type includes passive selection and active selection. When the selection type is active selection, the selection item is determined based on the eye movement parameters corresponding to each observation time, and the training mode type corresponding to each selection item is different. A selection list is generated based on the selection options, and the selection list is superimposed on the visual training image corresponding to the next training time. A prompt is generated to remind the patient to choose the training mode type from the selection list. Obtain the selected eye-tracking parameters corresponding to the selected time period, determine the target selection item corresponding to the patient based on the selected eye-tracking parameters, and adjust the visual training image for the next training stage based on the training mode type corresponding to the target selection item.

7. A monitoring system, characterized in that, The monitoring system includes: At least one processor; Memory; At least one application, wherein the at least one application is stored in memory and configured to be executed by at least one processor, the at least one application being configured to: perform a stereoscopic vision training supervision method in a polarization mode according to any one of claims 1-6.

8. A computer-readable storage medium, characterized in that, include: The computer program stores a method for supervising stereoscopic vision training in a polarization mode that can be loaded by a processor and executed as described in any one of claims 1-6.

9. A computer program product, characterized in that, The method includes a computer program that, when executed by a processor, implements the steps of a stereoscopic vision training supervision method in a polarization mode according to any one of claims 1-6.

Images