A training system for improving the profile recognition function of a low vision population

Abstract

This invention specifically relates to a training system for improving contour recognition function in people with low vision, belonging to the field of visual function training technology. It includes: a training image core parameter configuration module; a training image generation module; a training process control module; and a data recording and analysis module. Based on the functional characteristics of people with low vision, this invention effectively helps them improve their ability to integrate multiple visual features into contour features through targeted training image design, rigorous training process control, and the simultaneous use of training aids. This reduces various visual difficulties they encounter in daily life and improves their quality of life.

Description

Technical Field

[0001] This invention relates to the field of visual function training technology, and in particular to a training system for improving the contour recognition function of people with low vision. Background Technology

[0002] Low vision is a visual impairment. According to the World Health Organization, low vision is defined as a visual acuity that is still below 0.3 after refractive correction (such as wearing glasses or contact lenses), drug treatment, or surgery.

[0003] Therefore, people with low vision cannot see the details of objects in daily life (such as the texture on a wooden tabletop or the face of a pedestrian walking towards them), but can only recognize the general outlines (such as a flat tabletop or the shadow of a person walking towards them).

[0004] Most importantly, unlike nearsighted patients whose vision is low when looking at distant objects but can reach the normal standard of 1.0 when observing nearby objects, low-vision patients do not experience an improvement in vision when observing objects up close. The visual perception quality is only slightly improved because the image of the object on the retina becomes larger when it gets closer.

[0005] Existing data indicate that visual perception learning (clinically also known as "visual training") can significantly improve visual function in various populations, including strabismus and amblyopia patients and the elderly, including visual acuity, spatial contrast sensitivity, stereopsis, and motion perception.

[0006] Practice has shown that visual perception learning can also help improve related visual functions (such as contour recognition) in people with low vision. However, there are currently almost no reports on training systems specifically for people with low vision.

[0007] Therefore, a training system for improving contour recognition in people with low vision is proposed to address the aforementioned problems. Summary of the Invention

[0008] The purpose of this invention is to provide a training system for improving the contour recognition function of people with low vision in order to solve the above-mentioned problems.

[0009] To achieve the above objectives, the present invention adopts the following technical solution:

[0010] A training system for improving contour recognition capabilities in visually impaired individuals includes:

[0011] The training image core parameter configuration module is configured to set the contrast segment and spatial frequency filtering segment of the main training image, as well as the spatial frequency of the auxiliary training image, based on the spatial contrast sensitivity curve of the training individual.

[0012] The training image generation module is configured to generate visual stimulus images for training based on the original training images in the database, the parameters determined by the core parameter configuration module of the training images, and the parameters passed by the flow control module.

[0013] The training process control module is configured to control the sequence between different trials during a single training session, and, based on the individual trainee's recognition of the training images, sends control signals to the training image generation module to adjust the training parameters in the next trial.

[0014] The data recording and analysis module is configured to record and analyze the current training data, and adjust the training parameters for the next training session based on the analysis results.

[0015] Preferably, the specific method for setting the contrast segment and spatial frequency segment of the main training image is as follows:

[0016] The lower limit of the contrast segment is set to the first value, and the upper limit is 1.0. The lower limit of the spatial frequency filtering segment is 0, and the upper limit is the x-coordinate of the point on the spatial contrast sensitivity curve of the training individual, where the vertical axis is the reciprocal of the lower limit of the contrast segment to be used.

[0017] Preferably, the method for determining the spatial frequency of the auxiliary training image is as follows:

[0018] S1-1. Calculate the difference between the spatial contrast sensitivity curve of the training individual and the preset spatial contrast sensitivity curve reference value, and perform standardization processing to convert the difference values ​​at all spatial frequencies into relative difference values ​​with a numerical range of [0,1].

[0019] S1-2. Use a preset model to fit the relative difference values ​​to the data, and determine the best model based on statistical analysis;

[0020] S1-3. Determine the spatial frequencies that should be used for auxiliary training images. .

[0021] Preferably, the method by which the training image generation module generates training images includes:

[0022] S2-1. The training image is set to a circle, consisting of an inner circular region and an outer annular region. The relationship between the diameter of the inner circular region and the width of the outer annular region is as follows:

[0023] ;

[0024] S2-2. Based on the original training images in the database, determine the main training image within the inner circular region according to the setting of the spatial frequency filtering segment to be used. ;

[0025] S2-3. Based on the information recorded in the training image database, determine the orientation information of the main content in the original training images used in S2-2. ;

[0026] S2-4. Determine the auxiliary training images within the outer annular region. .

[0027] Preferably, the method further includes:

[0028] The recognition difficulty of the training images is determined by the main recognition difficulty and the secondary recognition difficulty.

[0029] The main recognition difficulty corresponds to the main training image, and the parameters that adjust the overall contrast of the image are determined by the inner circular region. Decide, The initial value is the second value;

[0030] The sub-recognition difficulty corresponds to the auxiliary training image and is determined by the contrast of the grating in the auxiliary training image within the outer annular region. The initial contrast value of the grating is the third value.

[0031] Preferably, the process control method for the training process control module to determine the training parameters for the next trial in the current training is as follows:

[0032] S3-1. If the training process control module receives an abnormal situation prompt signal during the presentation of the training image in the previous trial, it determines that the training in the previous trial is "invalid" and sends a control signal to the training stimulus generation module to determine that the training image will not be adjusted in the next trial and will continue to use the training image from the previous trial; otherwise, it determines that it is "valid" and proceeds to S3-2.

[0033] S3-2. If the training individual correctly identifies the training image in the previous trial, the sub-identification difficulty increases by 1 minimum change unit in the next trial; if the identification result is incorrect in the previous trial, the sub-identification difficulty decreases by a first specified number of minimum change units in the next trial; where the first specified number is a positive integer greater than 1.

[0034] S3-3. If, in the most recently completed second specified number of trials, the sub-identification difficulty reaches the upper limit of the sub-identification difficulty in at least the third specified number of trials, then the main identification difficulty in the next trial is increased by 1 minimum change unit, and the sub-identification difficulty is reduced to the lower limit of the sub-identification difficulty; if, in the most recently completed second specified number of trials, the sub-identification difficulty reaches the lower limit of the sub-identification difficulty in at least the third specified number of trials, then the main identification difficulty in the next trial is reduced by 1 specified number of minimum change units, and the sub-identification difficulty is increased to the upper limit of the sub-identification difficulty.

[0035] Preferably, the data recording and analysis module can also analyze and adjust the upper limit of the spatial frequency filtering segment used in the main training image in the next training session based on the completion status of each training session, specifically:

[0036] If, during the current training, the difference between the primary recognition difficulty threshold and the upper limit of the primary recognition difficulty when the training individual identifies the training image is less than the minimum change unit specified by the fourth number, then in the next training, the upper limit of the spatial frequency filtering segment used in the primary training image will be increased.

[0037] Preferably, if, in the current training, the difference between the main recognition difficulty threshold and the lower limit of the main recognition difficulty when the training individual identifies the training image is less than the minimum change unit of the fourth specified number, then in the next training, the upper limit of the spatial frequency filtering segment used in the main training image is reduced.

[0038] Preferably, the orientation θ of the grating in the auxiliary training image is related to the orientation information of the main content in the original training image. The relationship can also be: ,and ;in, It is a normal distribution function.

[0039] Preferably, the method further includes:

[0040] The human-computer interaction module is configured to include a display device, an information input device, a sound prompt device, and a monitoring device; the refresh rate of the display device is not lower than the fourth value, enabling the training images to flicker at a maximum of 40Hz.

[0041] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0042] 1. Based on the functional characteristics of people with low vision, this invention effectively helps people with low vision improve their ability to integrate multiple visual features into contour features through targeted training image design, rigorous training process control, and synchronous use of training aids, thereby reducing various visual difficulties they encounter in daily life and improving their quality of life.

[0043] 2. By comparing the differences in sensitivity curves and model fitting, this invention accurately locates the spatial frequency channel with the strongest current function of an individual, and scientifically determines the applicable spatial frequency filtering range based on this. This enables the filtered main training image to effectively eliminate interference from other spatial frequencies, thereby significantly improving the efficiency and effect of integrating single visual features such as spatial frequency, contrast, and orientation into contour features during training. This avoids the blindness and efficiency uncertainty that inevitably occur when training with uniformly configured parameters.

[0044] 3. This invention cleverly matches the main training image and the auxiliary training image that guides the integration of spatial frequency and orientation features through a double-layer image structure of inner circle and outer ring, thereby achieving the goal of "providing simplified visual features from the main training image, guiding feature integration through the auxiliary training image, strengthening the ability to integrate multiple features through repeated training, and ultimately improving the contour recognition ability under complex conditions in real life".

[0045] 4. The training process of this invention offers excellent controllability. Through multi-dimensional monitoring of eye movement, distance, and illumination, abnormal situations can be corrected in a timely manner. The recognition difficulty can be dynamically adjusted based on the results of each trial to ensure the effectiveness of the training. It also exhibits good adaptability to long-term training. The data recording and analysis module can gradually adjust the spatial filtering range of the main training image based on data analysis, making the training image increasingly closer to real life. This adapts to the needs of different training stages and gradually improves the contour recognition ability under complex real-life conditions. Optional enhancement functions further improve the effect. Flashing stimulation and mid-frequency current stimulation enhance attention and neuroplasticity, while multimodal feedback further optimizes the training experience. It is suitable for people with low vision and helps improve their contour recognition ability. Attached Figure Description

[0046] Further details, features, and advantages of this application are disclosed in the following description of exemplary embodiments in conjunction with the accompanying drawings, in which:

[0047] Figure 1 This is a system structure diagram of the present invention;

[0048] Figure 2 A schematic diagram illustrating the working effect of different spatial frequency channels;

[0049] Figure 3 This is a schematic diagram of the training images. Detailed Implementation

[0050] Several embodiments of this application will now be described in more detail with reference to the accompanying drawings to enable those skilled in the art to implement this application. This application may be embodied in many different forms and for various purposes and should not be limited to the embodiments set forth herein. These embodiments are provided to make this application thorough and complete, and to fully convey the scope of this application to those skilled in the art. The embodiments described do not limit this application.

[0051] Unless otherwise defined, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It will be further understood that terms such as those defined in commonly used dictionaries shall be interpreted as having a meaning consistent with their meaning in the relevant field and / or the context of this specification, and shall not be interpreted in an idealized or overly formal sense unless expressly defined herein.

[0052] Example 1

[0053] Its specific implementation method is combined with the appendix Figure 1 To be continued Figure 3 Please provide a detailed explanation.

[0054] Appendix Figure 1 This invention provides a structural block diagram of a training system for improving contour recognition in visually impaired individuals. The diagram illustrates the connection between the training image core parameter configuration module and the data recording and analysis module, and marks the main functional interaction flow of each module.

[0055] In this embodiment, it includes:

[0056] The training image core parameter configuration module, other training parameter setting module, and training image generation module are used for the design and generation of training images, and are the core of the entire training system or for training contour recognition function in people with low vision. The human-computer interaction module, training process control module, and data recording and analysis module can assist in the implementation of the training plan and can gradually adjust the training parameters in an adaptive manner to achieve personalized training based on the individual trainee's situation. This design fully reflects modern science's understanding of the nature of vision.

[0057] Specifically, in order to recognize optical images projected onto the retina, the human visual nervous system needs to perform a series of complex operations.

[0058] The first stage involves extracting various perceptible and identifiable basic features (such as spatial frequency, orientation, and direction of motion) from optical images projected onto the retina. This work primarily involves the retina, thalamus, primary visual cortex, and parts of the extrastriate cortex.

[0059] The second stage requires the extraction of various basic features and their correct integration to gradually form a "perceptual image" in the mind that matches the real image. This stage is mainly completed in the extrastriate cortex. From this perspective, if either of these two stages goes wrong, the final visual perception or visual quality will be significantly affected.

[0060] Patients with strabismus and amblyopia encountered in ophthalmology clinics often require various treatments, including visual training. During the treatment process, basic visual functions such as visual acuity and spatial contrast sensitivity in these patients gradually improve to normal levels. Existing scientific research has shown that this treatment process is primarily accompanied by an improvement in the efficiency of the visual system's image analysis in the first stage.

[0061] Therefore, visual training for patients with strabismus and amblyopia often sets the training goal as "improving the perception ability of single visual features in the first stage", and designs training programs accordingly.

[0062] However, the situation is completely different for patients with low vision. The pathogenesis of low vision is diverse and it is almost untreatable in clinical practice, which is fundamentally different from strabismus and amblyopia.

[0063] This means that for patients with low vision, setting the training goal as "improving the efficiency of the first stage of image analysis of the visual system" and attempting to improve the perception ability of a single basic visual feature (such as spatial contrast sensitivity or visual acuity) is of very limited significance.

[0064] Therefore, the training goal for low vision patients should be set in the second stage of visual system image analysis, to improve the integration efficiency (i.e., the speed at which low vision patients can recognize blurred images) and the final integration effect (i.e., how much content, such as the outlines of various objects, can be distinguished from blurred images after sufficient recognition time) as much as possible when multiple visual features are reintegrated. The aim is to improve the ability to recognize object outlines and improve the quality of life of low vision patients as much as possible, even if it is impossible to improve various individual basic visual functions.

[0065] In real life, there are often examples of children who have poor basic visual functions in one area but strong ability to integrate and apply multiple functions. For instance, when searching for a specific target in a pile of blocks, the child who finds it the fastest is often not the one with the best eyesight; similarly, in badminton, better eyesight or faster running speed does not necessarily mean better performance.

[0066] In order to achieve the goal of "improving the efficiency and effectiveness of the second stage of image analysis in the vision system", the training system or training scheme needs to be carefully designed.

[0067] Unlike training for strabismus and amblyopia, which may rely solely on training images with single visual features (such as gratings) or simply combine multiple visual features, training for amblyopia requires a different approach. This is because such methods significantly increase the difficulty for the visual system of low-vision patients to extract various visual features from already blurred training images, leading to slow training progress and significantly reduced effectiveness. The correct approach is to design a training scheme that facilitates both the extraction of individual visual features (such as spatial frequency, orientation, and contrast) and the integration of various visual features.

[0068] Therefore, an ideal training system for patients with low vision should have the following characteristics:

[0069] (a) The training images presented should have real-world significance; otherwise, not only will the identification criteria be difficult to grasp, but it will also be difficult to play a role in improving the quality of life of low-vision patients.

[0070] (b) The training images at the beginning of training should be simplified as much as possible, and "redundant visual features should be eliminated and only visual features used for integration" to facilitate the first stage of visual system image analysis, that is, to facilitate the extraction of various visual features.

[0071] (c) Provide cues to assist in the reintegration of various visual features during training, and continuously adjust the strength of cues according to the training progress to facilitate and promote the second phase of image analysis of the visual system.

[0072] (d) As training progresses, the simplification of the training images is gradually reduced to make them more realistic, thus promoting the application of the improved visual system image analysis second phase in real-life scenarios.

[0073] The design of this training system is very close to the ideal training system described above.

[0074] This system includes a raw training image library, containing only real, natural images. The training image generation module uses these real, natural images and performs spatial frequency filtering. The filtered images remove most unnecessary information, retaining only the main object contour information. This setup ensures that the training images used by this system meet the characteristics (a) and (b) of the ideal training system described above.

[0075] The training image generation module also adds an auxiliary training image of equal area around the main training image (i.e., the real natural image after spatial frequency filtering). The auxiliary training image is a grating with orientational information. The spatial frequency of this grating falls within the last spatial frequency channel currently possessed by the low-vision training individual, and also within the range of spatial frequencies retained after spatial frequency filtering. The contrast of the grating is above a threshold. The orientation of the grating is basically consistent with the orientational information of the main content in the original training image (i.e., the real natural image). These settings ensure that the auxiliary training image is basically clearly visible, and also ensure that its information does not interfere with the main training image (because the main training image is located in the inner visual field center, while the auxiliary training image is located on the periphery). Simultaneously, the spatial frequency and orientational information provided by the auxiliary training image helps guide the integration of spatial frequency and orientational information in the main training image within the visual field center, assisting in the recognition of object contours in the main training image, thus conforming to the characteristics (c) of the aforementioned ideal training system.

[0076] The training process control module can continuously adjust the contrast (i.e., cue intensity) of the auxiliary training images during a single training session, thereby facilitating the second stage of image analysis in the visual system. This setup also aligns with the characteristics (c) of the ideal training system described above.

[0077] Finally, the data recording and analysis module can analyze and adjust the upper limit of the spatial frequency filtering segment used in the main training image in the next training session based on the completion status of each training session, making it gradually closer to real life (in low-pass filtering mode, the higher the upper limit of the filtering segment, the worse the filtering effect, and the closer the filtered image is to the original image), thus promoting the application of the improved visual system image analysis second stage in real-life scenarios. This setting conforms to the characteristics (d) of the ideal training system mentioned above.

[0078] The following provides further explanation for each working module:

[0079] The training image core parameter configuration module is configured to set the contrast segment and spatial frequency filtering segment of the main training image, as well as the spatial frequency of the auxiliary training image, based on the spatial contrast sensitivity curve of the training individual.

[0080] The specific method for setting the contrast and spatial frequency ranges of the main training image is as follows:

[0081] The lower limit of the contrast segment is set to the first value, and the upper limit is 1.0. The lower limit of the spatial frequency filtering segment is 0, and the upper limit is the x-coordinate of the point on the spatial contrast sensitivity curve of the training individual, where the vertical axis is the reciprocal of the lower limit of the contrast segment to be used.

[0082] When determining the contrast range to be used in the main training image, the lower limit of the contrast range should be set to a first value, and the upper limit should be 1.0. Generally, the first value should not be lower than 0.4. Preferably, the first value is 0.5.

[0083] This means that the main training images used have high contrast, corresponding to "contour recognition in a relatively bright environment".

[0084] This is because patients with low vision have very poor basic visual functions, and visual training can hardly improve these basic visual functions. Therefore, from the perspective of improving the quality of life, it is sufficient to achieve the goal of "improving the ability to recognize contours in relatively bright environments" through training, and there is no need to excessively pursue "improving the ability to recognize contours in dim environments".

[0085] In fact, since it is almost impossible to improve the basic visual function of patients with low vision, it is difficult to achieve even the goal of "improving the ability to recognize contours in dim environments".

[0086] It should be noted that visual acuity in low-vision patients ranges from 0.05 to 0.30, a wide range. Therefore, for low-vision patients with relatively good vision, a smaller value can be chosen for the first value, such as 0.4, while for low-vision patients with poor vision, a larger value should be chosen for the first value, such as 0.7.

[0087] When determining the spatial frequency filtering segment to be used for the main training image, the lower limit of the spatial frequency filtering segment should be set to 0 (i.e., low-pass filtering), and the upper limit should be set to the x-coordinate (i.e., spatial frequency) of the point on the spatial contrast sensitivity curve of the training individual where the y-axis is the reciprocal of the lower limit of the contrast segment to be used. For example, if the lower limit of the contrast segment is set to 0.5, its reciprocal is 2. Then, on the spatial contrast sensitivity curve of the training individual, find the point where the y-axis (i.e., contrast sensitivity) is 2. If the x-coordinate corresponding to this point is 6.0 cycles / degree, then it can be determined that the spatial frequency filtering segment to be used for the main training image should be 0 to 6.0 cycles / degree.

[0088] The method for determining the spatial frequency of auxiliary training images is as follows:

[0089] S1-1. Calculate the difference between the spatial contrast sensitivity curve of the training individual and the preset spatial contrast sensitivity curve reference value, and perform standardization processing to convert the difference values ​​at all spatial frequencies into relative difference values ​​with a numerical range of [0,1].

[0090] ;

[0091] ;

[0092] in, Spatial frequency, measured in periods per degree; The spatial contrast sensitivity is the contrast sensitivity at the corresponding spatial frequency on the preset spatial contrast sensitivity curve reference value. The preset spatial contrast sensitivity curve reference value should be the lower limit of the spatial contrast sensitivity curve in low vision populations, preferably the contrast sensitivity curve value of low vision patients with visual acuity of 0.05 and a cutoff spatial frequency of 2.0 cycles / degree. The contrast sensitivity at the corresponding spatial frequency on the spatial contrast sensitivity curve for patients with low vision before training; This represents the difference between the spatial contrast sensitivity curve and the reference value for low-vision patients at corresponding spatial frequencies. This is the set of differences between the spatial contrast sensitivity curves of low-vision patients and reference values ​​across all spatial frequencies. This represents the relative difference in contrast sensitivity at the corresponding spatial frequency after standardization. This is a function to find the maximum value.

[0093] S1-2. Use a pre-defined model to fit the relative differences to the data, and determine the optimal model based on statistical analysis; the formula for the pre-defined model is:

[0094] ;Formula 1;

[0095] ;Formula 2;

[0096] in, Spatial frequency; This represents the relative difference in contrast sensitivity at the corresponding spatial frequency after standardization, with a numerical range of [0,1]. , , All are fitted parameters;

[0097] S1-3. Determine the spatial frequencies that should be used for auxiliary training images. Specifically:

[0098] If the optimal model is Equation 1, then If the optimal model is Equation 2, then ; The parameter represents the bandwidth of the spatial frequency information processing channel in the visual system of patients with low vision under normal circumstances, and its value ranges from 0.8 to 1.2 octaves, preferably 1.0 octaves.

[0099] To determine the optimal model, we first need to obtain the goodness of fit when using Equation 1 and Equation 2 for data fitting. The calculation method is as follows:

[0100] ;

[0101] in, For data fit goodness; The data predicted by the model (i.e., the expected value of the difference in contrast sensitivity at different spatial frequencies). The actual difference in contrast sensitivity at different spatial frequencies; This is a function for taking the average value;

[0102] After obtaining the goodness-of-fit, a comparison between models is necessary. Since Equation 1 uses one fewer fitting parameter than Equation 2 (i.e., the baseline), if the goodness-of-fit of Equation 1 is not lower than that of Equation 2, then Equation 1 can be determined as the best model. However, if the goodness-of-fit of Equation 2 is greater than that of Equation 1, further analysis should be conducted to determine whether the improvement in goodness-of-fit of Equation 2 is statistically significant. If so, then Equation 2 can be determined as the best model; otherwise, Equation 1 remains the best model. To determine whether the improvement in goodness-of-fit of Equation 2 is statistically significant, an F-test can be used. The formula for calculating the F-value and its degrees of freedom is as follows:

[0103] ;

[0104] ;

[0105] ;

[0106] in, This represents the number of fitting parameters used in Formula 2 (i.e., 3 parameters). This represents the number of fitting parameters used in Formula 1 (i.e., 2 parameters). The number of data points used when fitting data using Formula 1 and Formula 2;

[0107] The steps described above for determining the spatial frequencies that the grating should use in the auxiliary training images are related to many important discoveries in the field of biomedical research, and are briefly described below:

[0108] (a) Numerous studies have shown that the human visual system perceives different spatial frequency information through different spatial frequency channels within the visual cortex. Spatial frequency channels typically exhibit band-pass characteristics, primarily processing spatial frequency information "within a certain range centered on a specific frequency," and their spectral characteristics resemble a Gaussian function. When a spatial frequency channel is affected, its function (e.g., sensitivity) will change significantly, but the functions of other spatial frequency channels will remain largely unaffected.

[0109] (b) Different spatial frequency channels process the visual image simultaneously, thus decomposing the image into multiple image information segments located in different spatial frequency ranges. For example... Figure 2 As shown, the common image of a dog's head can be decomposed into multiple sub-images containing different spatial frequency information by processing different spatial frequency channels, in order from low to high spatial frequency. These sub-images are then merged and summarized to form the complete original image.

[0110] (c) If it is confirmed that some spatial frequency channels of an individual organism are functionally deficient, the spatial contrast sensitivity curve can be compared with a standard spatial contrast sensitivity curve to identify potentially deficient spatial frequency channels. Of course, for patients with low vision, since their visual acuity and spatial contrast sensitivity curves are much lower than those of normal individuals, their spatial frequency channel impairment is generally more severe. In this case, the assessment of the functional status of their spatial frequency channels should aim to "explore which spatial frequency channels are normal besides the most basic spatial frequency channels," rather than "explore which spatial frequency channels are abnormal." Therefore, the spatial contrast sensitivity of patients with low vision can be compared with a preset spatial contrast sensitivity curve reference value (preferably the contrast sensitivity curve value for patients with low vision and a visual acuity of 0.05 and a cutoff spatial frequency of 2.0 cycles / degree). First, the difference between the spatial contrast sensitivity curves under the two conditions is calculated in S1-1. Then, through Gaussian model analysis in S1-2, the remaining functionally normal spatial frequency information processing channels of the patient with low vision are assessed compared to the reference condition. It should be noted that existing research shows that functional deficiencies are almost nonexistent at low spatial frequencies. Therefore, in order to improve the accuracy of model analysis during S1-2, data corresponding to spatial frequencies of 1.0 period / degree and below can be removed.

[0111] (d) If, besides the most basic spatial frequency channel, the low-vision patient has only one other functionally normal spatial frequency information processing channel, then the model analysis in Formula 1 of S1-2 can perform the evaluation well. In this case, the spatial frequency channel indicated by the model analysis must be the spatial frequency information processing channel within the visual system that processes the highest frequency information (i.e., the spatial frequency processed by this channel is higher than all other spatial frequency information processing channels). Its central position (i.e....) The frequency of the grating in the auxiliary training image must be within the passband of the spatial filter of the main training image and have the highest value at the center of all spatial frequency channels. This is most conducive to guiding the feature integration of the main training image. Therefore, the spatial frequency corresponding to the center position can be selected as the spatial frequency of the grating in the auxiliary training image.

[0112] (e) It should be noted that in rare cases, low-vision patients with good visual acuity may still have two or more remaining, functionally normal spatial frequency channels. To effectively identify this situation, a comprehensive analysis combining the baseline of the fitting parameters is necessary. If statistical analysis indicates that the existence of the baseline is statistically significant, then Formula 2 is the best-fit model. In this case, the sfpeak indicated by the model fitting results is affected by multiple functionally normal spatial frequency channels and cannot truly reflect the central location of the spatial frequency information processing channel that processes the highest frequency information within the visual system. Based on existing research experience and supporting data, the formula can be used... Determine the center location of the spatial frequency information processing channel within the vision system that processes the highest frequency information (i.e., the spatial frequency at which the grating should be used in the auxiliary training image). It should be noted that... The parameter represents the bandwidth of the spatial frequency information processing channel in the visual system of low-vision patients under normal circumstances. According to existing research data, the bandwidth of the spatial frequency information processing channel in the visual system of low-vision patients ranges from 0.8 to 1.2 octaves, with 1.0 octaves being preferred.

[0113] The various parameters set in other training parameter settings modules are all related to the generation of training images, training process control, and training effect.

[0114] The training volume set in other training parameter settings modules is directly related to the time spent in a single training session. Generally, the training volume ranges from 600 to 900 trials. One trial represents "the individual's recognition of a stimulus presented." Assuming a trial takes 4 seconds to complete, the training time corresponding to the above training volume would be 40 to 60 minutes. However, if necessary rest periods are considered during training, the time spent in a single training session corresponding to the above training volume would be approximately 50 to 75 minutes.

[0115] The training volume set in other training parameter settings modules is also related to the training effect. According to the general law of biological adaptation to the environment, when an organism receives a certain level of external stimulation, and the intensity and cumulative amount of the stimulation reach a certain level but do not cause serious damage to the body, the organism will strengthen the parts or tissues related to the stimulation reception or response after the stimulation ends, in order to enhance its responsiveness to the stimulation. At the behavioral level, this manifests as "practice makes improvement." Therefore, a certain training volume (i.e., training intensity) is required for each training session; otherwise, it will be impossible to promote the organism's self-reinforcement after stimulation. When the training volume set in other training parameter settings modules is insufficient, not only is the training intensity of a single training session insufficient, but it also affects the training difficulty. This is because if the initial difficulty of stimulus recognition is relatively low, and the training volume is small, it is very likely that the stimulus recognition difficulty will not have reached the recognition threshold by the end of the training. Generally, the threshold difficulty can be simply understood as the limit of ability that an organism can achieve. Failing to reach the threshold difficulty means that the average difficulty of the stimulus during the training process has not met the specified requirements. In this case, even if the training volume is relatively large, it is difficult to promote the organism's self-reinforcement. In real life, it is difficult for ordinary people to train their sprinting speed or long-distance running endurance through walking training, and the same principle applies here.

[0116] The other training parameter settings module also allows you to set the presentation time of the training images. Generally, when training low-vision patients, the presentation time of the training images in each trial is between 250 and 750 milliseconds. Preferably, the presentation time is 500 milliseconds.

[0117] Other training parameter settings modules also allow setting the minimum difficulty unit and initial difficulty for training image recognition. The recognition difficulty of the training image is determined by adjusting the overall contrast of the main training image (main recognition difficulty) and the contrast of the raster in the auxiliary training image (secondary recognition difficulty). In other words, both main and secondary recognition difficulty are reflected in the change in contrast. According to general principles in the field of visual training technology, the minimum difficulty unit (i.e., minimum contrast change unit) ranges from 0.5 to 2.0 dB. For patients with low vision, 0.5 dB is preferred. Since the functional level of patients with low vision is generally low, the initial difficulty can be within the selectable range, preferably the lowest difficulty. That is, for the main recognition difficulty, adjusting the overall contrast of the main training image (i.e., the minimum contrast change unit) is the minimum difficulty unit. The contrast of the raster in the auxiliary training image can be set to the maximum value of 1.0; for the difficulty of sub-recognition, the contrast of the raster in the auxiliary training image can also be set to the maximum value (i.e., The upper limit of the preferred value range is 0.8.

[0118] Other training parameter settings modules also allow for setting the stimulus size. Considering the characteristics of the human eye—that "spatial resolution gradually decreases from the center to the periphery in different visual fields, exhibiting a radial distribution"—and that "the macula, the area with the highest spatial resolution on the retina, generally does not exceed 6 degrees of visual angle," as well as the structural feature of the training image requiring "a double-layered region with an inner circle and an outer ring" as required by this system, the stimulus size is typically "5.0 to 6.0 degrees of visual angle in diameter." Preferably, the diameter of the inner circular region is 4.0 degrees of visual angle, the width of the outer ring region is 0.83 degrees of visual angle, and the overall diameter of the training image is 5.7 degrees of visual angle.

[0119] Other training parameter setting modules are configured to set training parameters, including training amount, presentation time of training images, range of recognition difficulty of training images, starting recognition difficulty and minimum change unit of recognition difficulty.

[0120] The training image generation module is configured to generate visual stimulus images for training based on the original training images in the database, the parameters determined by the core parameter configuration module of the training images, and the parameters passed by the flow control module.

[0121] Methods for generating training images include:

[0122] S2-1. The training image is set to a circle, consisting of an inner circular region and an outer annular region. The relationship between the diameter of the inner circular region and the width of the outer annular region is as follows:

[0123] ;

[0124] The width or diameter is expressed in terms of angle, with the unit being "degrees";

[0125] S2-2. Based on the original training images in the database, determine the main training image within the inner circular region according to the setting of the spatial frequency filtering segment to be used. Specifically:

[0126] ;

[0127] in, It is a two-dimensional Gaussian low-pass filter function; The original training images in the database; This is the upper limit of the spatial frequency filtering section to be used (i.e., the cutoff frequency of the low-pass filter). These are parameters that describe the smoothness of the Gaussian low-pass filter function. The parameter for adjusting the overall contrast of the image has a range of values ​​that is the contrast range of the main training image set in the core parameter configuration module of the training image.

[0128] S2-3. Based on the information recorded in the training image database, determine the orientation information of the main content in the original training images used in S2-2. ;

[0129] S2-4. Determine the auxiliary training images within the outer annular region. Specifically:

[0130] ;

[0131] in, This represents the brightness at each point (x, y) in the auxiliary training image; Average brightness; To assist in training the spatial frequency of the grating in the image; To enhance the contrast of the raster in the training image; The viewpoint occupied by each point (x, y) in the image; To assist in determining the orientation of the raster in the training image, the values ​​are set to the orientation information of the main content in the original training image. ; The phase of the grating in the training image is a random value; As an image mask, its function is to increase the contrast of the grating stimulation within the outer annular region. Along the radial direction from the inside out, it gradually decays to 0, that is, the image in the outer region gradually fades and blends into the background along the radial direction from the inside out.

[0132] It should be noted that, generally, the contrast value of an image ranges from [0,1]. However, considering that the auxiliary training image needs to guide the integration of multiple visual features in the main training image, therefore... The range of values ​​for is subject to certain limitations. Its lower limit cannot be too low, otherwise the auxiliary training image will appear too blurry and lose its guiding function; however, its upper limit cannot be too high, otherwise the auxiliary training image will appear too clear, thus distracting the trainee and affecting the integration of multiple visual features in the main training image. Practical data shows that... The preferred value range is [0.4, 0.8].

[0133] Figure 3 The image displays two training images. It can be seen that the inner main training image is a natural image ("lion" or "giraffe") filtered by spatial frequency, while the outer auxiliary training image is a grating with specific spatial frequency and orientation information. The orientation of the grating indicates the location of the main information content in the inner main training image. The spatial frequency of the grating corresponds to the spatial frequency range of the inner main training image. Guided by the outer grating, the content in the main training image is more easily recognized.

[0134] In addition, in the above formula It is related to the pixel size of the display device and the training distance, specifically:

[0135] ;

[0136] in, The viewpoint occupied by each point in the training image, in degrees per point; The actual size of each point in the training image (i.e., the pixel size of the display device), in centimeters; For distance measurement, the unit is centimeters;

[0137] It is important to emphasize that the circular structure of the training images, and the relationship between the diameter of the inner circular region and the width of the outer annular region, ensures that the training images conform to the physiological characteristics of the human visual system (i.e., visual resolution is highest in the central visual field and then decreases radially from the center to the periphery). Structurally, it divides the training images into two regions of equal area but different spatial locations to match the training design. This ensures that the auxiliary training image is basically clearly visible while also ensuring that its information does not interfere with the main training image (because the main training image is located in the center of the inner visual field, while the auxiliary training image is located on the periphery). At the same time, the spatial frequency and orientation information provided by the auxiliary training image helps guide the integration of spatial frequency and orientation information in the main training image within the center of the visual field, assisting in the recognition of object contours in the main training image.

[0138] The inner circular region contains the main training image, which has been spatially frequency filtered and is directly related to the training task requirements. During training, trainees are required to identify the contours of the main training image within the inner circular region.

[0139] For example, what animal is in the image? A cow or a horse? A cat or a mouse? The auxiliary training image presented in the outer ring region is only used to "guide the recognition of the main training image contour" and is not directly related to the training task requirements. In other words, to correctly complete the training task, the trainee will concentrate on identifying the main training image within the inner circular region. Simultaneously, because the outer ring region is adjacent to the inner ring region and its total area is not large (not exceeding 6.0 degrees of visual field), the trainee will "casually browse" the auxiliary training image appearing in the outer ring region while concentrating on their attention. Research shows that during visual perception learning, the functional level of the core task of focused observation is significantly improved, and is also influenced by the attributes of other stimuli presented around the core task stimulus, presented synchronously with the core task stimulus, repeatedly exposed, and casually browsed. When the characteristic attributes of the stimuli surrounding the core task are similar to the characteristic attributes of the core task, such influence is positive, facilitating, and beneficial. Combining the features (b) and (c) of the ideal training system for low vision patients mentioned above, it can be found that this training image design of "inner circle and outer ring double-layer structure, inner core filter image, and outer auxiliary feature prompts" perfectly meets the above requirements.

[0140] It is important to emphasize that although the training image used for recognition is the main training image located in the inner layer, the auxiliary training images in the outer layer also have a certain influence on the recognition of the inner layer images. Therefore, the recognition difficulty of the training images is jointly determined by the main recognition difficulty and the auxiliary recognition difficulty. The main recognition difficulty corresponds to the main training image and is determined by adjusting the parameters of the overall image contrast within the inner circular region. Decision. Under the same circumstances, The larger the value, the easier it is to identify. The initial value is the second value, preferably 1.0.

[0141] The difficulty of secondary recognition corresponds to the auxiliary training image, determined by the contrast of the grating in the auxiliary training image within the outer annular region. Decision. Under the same circumstances, The larger the value, the easier it is to identify. The initial value is a third value, preferably 0.8.

[0142] Furthermore, as the training effect of low-vision trainees gradually improves, to guide the application of this contour integration ability in real life, it is necessary to consider gradually increasing the realism and naturalness of the inner main training image and gradually reducing the guiding role of the outer auxiliary training image. The method to improve the realism and naturalness of the inner main training image mainly involves increasing the cutoff frequency of the low-pass filter; see the functional introduction of the data recording and analysis module later for details. The method to reduce the guiding role of the outer auxiliary training image can be "by adjusting the contrast of the auxiliary training image to reduce its sharpness"; details can be found in the relevant introduction of the training process control module later. Additionally, it is also possible to "adjust the orientation θ of the raster in the auxiliary training image and the orientation information of the main content in the original training image." The relationship between the two makes the former and the latter highly correlated, but not entirely consistent. Specifically:

[0143] ,and ;in, It is a normal distribution function.

[0144] After the training images are generated, they can be presented to the trainee through the display device in the human-computer interaction module. The human-computer interaction module can also obtain the trainee's recognition of the training images through the information input device. In addition, the human-computer interaction module includes various monitoring devices to confirm the standardization of the trainee's behavior in each training trial, or the stability of the training environment. These monitoring devices include ambient light monitoring devices, training distance monitoring devices, and eye-tracking monitoring devices.

[0145] Generally, training does not place high demands on the display device in the human-computer interaction module; most commonly used LCD monitors are sufficient. However, it's important to note that due to the diverse causes of low vision and the generally low visual acuity of patients, trainees may sometimes be insensitive to the training images, failing to achieve substantial functional improvement even after repeated training sessions. In such cases, it's necessary to use reinforcing visual stimuli, specifically flashing training images, without significantly altering other settings. It's important to understand that flashing visual images can automatically attract and maintain the trainee's attention. To avoid distraction, it should be ensured that "the outer auxiliary training image does not flash, only the inner main training image flashes." This setup helps trainees focus more on the training task, effectively improving training results.

[0146] The flicker frequency ranges from 1.0 to 40.0 Hz. It is important to emphasize that research shows that a 40 Hz flicker stimulus induces electrical activity of the same frequency in the visual cortex, allowing this rhythm to propagate to other brain regions (such as the hippocampus and prefrontal cortex), thereby synchronizing and enhancing gamma-wave activity throughout the brain and exerting a series of positive and beneficial physiological effects, including enhanced neuroplasticity and improved coordination between different brain regions. To ensure that the image flicker frequency reaches the maximum requirement of 40 Hz and to avoid image distortion that may occur when the display device continuously refreshes images, the presentation time of each image during flickering should be no less than 3 frames. Therefore, the number of refreshes per second of the display device should be no less than (3+3)×40=240 frames. Of these, 120 frames are training images, and 120 frames are gray backgrounds. That is, optionally, the refresh rate of the display device in the human-computer interaction module is no less than the fourth value. Preferably, the fourth value is 240 Hz.

[0147] Optionally, the eye-tracking monitoring device monitors the direction of the trainee's gaze when the training image appears. When it is detected that the angle by which the trainee's gaze deviates from the center of the training image exceeds a preset proportion corresponding to the viewing angle of the inner circular region of the training image, an abnormal situation prompt signal is sent to the training process control module in real time, and an audio prompt message "Correct gaze direction" is sent to the trainee via an audio prompt device. Preferably, the preset proportion is 20%.

[0148] It should be noted that due to the unique structure of the inner circle and outer ring of the training image, and the training task requiring the trainees to focus primarily on the inner circle region, this training system has higher and stricter requirements for monitoring eye movements than ordinary training systems. If the eye movements of the trainees cannot be effectively monitored, the design concept of simultaneously obtaining efficient training of primary and secondary frequencies cannot be implemented, and the training effect will inevitably be greatly reduced.

[0149] Optionally, the training distance monitoring device monitors the distance between the display device and the training individual in real time through a distance sensor; when the monitored distance exceeds a preset proportion of the predetermined training distance, it sends an abnormal situation prompt signal to the training process control module in real time, indicating that the training behavior is not standardized, and sends an audio prompt to the training individual to "correct training distance" through an audio prompt device; preferably, the preset proportion is 10%.

[0150] It should be noted that due to the parameters in the training image formula... Directly related to the training distance, and with DPD further affecting the spatial frequency of the training grating, this training system has higher and stricter requirements for monitoring the training distance than ordinary training systems. If the training distance cannot be effectively monitored, the design of the auxiliary training images will deviate significantly during implementation, thereby significantly reducing the training effect.

[0151] Optionally, the ambient illuminance monitoring device monitors the illuminance of the training environment in real time using an illuminance sensor. When the relative change in ambient illuminance compared to the initial illuminance at the start of training exceeds a preset proportion, an abnormal situation alert signal is sent to the training process control module in real time, indicating a significant change in the training environment. An audible alert, "Correct ambient illuminance," is also sent to the individual trainee via an audible alert device. Preferably, the preset proportion is 50%. It should be noted that ambient illuminance and object brightness are two different concepts. When an object does not emit its own light and only reflects ambient light, the lower the ambient illuminance, the lower the object's brightness. Therefore, in daily life, the quality of life for people with low vision is more significantly affected in dimly lit environments.

[0152] However, when objects are self-illuminating (for example, in this training system, training images are displayed on a self-illuminating display device), the level of ambient illumination has little impact on object recognition ability. On the contrary, higher ambient illumination reduces the contrast of the grating stimuli presented to the training individual, affecting the precise setting of training parameters, interfering with the control of the training process, and thus impacting the training effect.

[0153] To achieve better training results, the human-computer interaction module also includes a mid-frequency current stimulation device, which can generate a low-frequency modulated mid-frequency current to act on acupoints around the eyes during the presentation of training images. Optionally, both the mid-frequency carrier current and the low-frequency modulation signal use sinusoidal alternating current, with the frequency range of the mid-frequency carrier current being 1.0kHz to 5.0kHz and the frequency range of the low-frequency modulation signal being 10.0Hz to 120.0Hz.

[0154] When a low-frequency modulated mid-frequency current is applied to acupoints around the eyes, the output power is similar to that in the main training image. The relevant formula is:

[0155] ;

[0156] in, , These are the minimum and maximum values ​​of the output power, respectively; preferably, , They are 5.0W and 0.5W respectively.

[0157] It should be noted that perceptions from different sensory organs can be interconnected. The underlying mechanism is very complex, involving cognitive integration and neurotransmitter regulation, among other factors. In practical applications, using low-frequency modulated mid-frequency current applied to acupoints around the eyes during training image presentation can effectively create a linkage between "visually meaningful light stimulation" and "surface electrical stimulation" within the body, effectively promoting visual cognitive optimization and enhancing training results.

[0158] Furthermore, when the training images have low contrast and high recognition difficulty, the intensity of electrical stimulation can be increased to enhance visual cognitive optimization under high difficulty. Also, the above formula describes the relationship between two types of stimulus energies, namely the relationship between the visual signal energy carried by the training image and the electrical energy; therefore, the formula uses the square of the contrast ratio instead of the actual contrast ratio. For ordinary visual images, the specific visual signal energy they carry is:

[0159] ;

[0160] in, It is the contrast of the point with coordinates (x, y) on the training image.

[0161] Numerous scientific studies have shown that providing feedback to trainees during training, indicating correct or incorrect identification, helps them subconsciously adjust their identification strategies and promotes functional optimization, especially when the identification difficulty is high. Generally, this feedback is provided through sound. For example, a specific cue tone is given when the identification is correct; no cue tone is given when the identification is incorrect, or a different cue tone is given. However, considering the aforementioned sensory linkage effect, providing training feedback through electrical stimulation can actually achieve better results. Therefore, optionally, the mid-frequency current stimulation device can also provide training feedback through current stimulation, specifically: in each trial, after the trainee completes the identification of the target stimulus in the training pattern, if the identification is correct, a current stimulation with a specified power and duration is provided; otherwise, no current stimulation is provided. Preferably, the first specified power is 2.5W and the duration is 0.5s.

[0162] Optionally, to avoid confusion of training data between different training individuals and to improve the effectiveness and accuracy of training process evaluation (for details on training process evaluation, please refer to the introduction related to the data recording and analysis module later), basic information of the training individual can be entered through the information input device of the human-computer interaction module during each training session, which will be used as an identifier for recording, storing and retrieving training data.

[0163] Also includes:

[0164] The recognition difficulty of the training images is determined by the main recognition difficulty and the secondary recognition difficulty.

[0165] The main recognition difficulty corresponds to the main training image, and the parameters that adjust the overall contrast of the image are determined by the inner circular region. Decide, The initial value is the second value;

[0166] The sub-recognition difficulty corresponds to the auxiliary training image and is determined by the contrast of the grating in the auxiliary training image within the outer annular region. The initial contrast value of the grating is the third value.

[0167] The human-computer interaction module is configured to include a display device, an information input device, a sound prompt device, and a monitoring device; the refresh rate of the display device is not lower than the fourth value, enabling the training images to flicker at a maximum of 40Hz;

[0168] The training process control module is configured to control the sequence between different trials during a single training session, and, based on the individual trainee's recognition of the training images, sends control signals to the training image generation module to adjust the training parameters in the next trial.

[0169] The specific flow control method for determining the training parameters for the next trial in the current training iteration is as follows:

[0170] S3-1. If the training process control module receives an abnormal situation prompt signal during the presentation of the training image in the previous trial, it determines that the training in the previous trial is "invalid" and sends a control signal to the training stimulus generation module to determine that the training image will not be adjusted in the next trial and will continue to use the training image from the previous trial; otherwise, it determines that it is "valid" and proceeds to S3-2.

[0171] S3-2. If the training individual correctly identifies the training image in the previous trial, the sub-identification difficulty increases by 1 minimum change unit in the next trial; if the identification result is incorrect in the previous trial, the sub-identification difficulty decreases by a first specified number of minimum change units in the next trial; where the first specified number is a positive integer greater than 1.

[0172] S3-3. If, in the most recently completed second specified number of trials, the sub-identification difficulty reaches the upper limit of the sub-identification difficulty in at least the third specified number of trials, then the main identification difficulty in the next trial is increased by 1 minimum change unit, and the sub-identification difficulty is decreased to the lower limit of the sub-identification difficulty; if, in the most recently completed second specified number of trials, the sub-identification difficulty reaches the lower limit of the sub-identification difficulty in at least the third specified number of trials, then the main identification difficulty in the next trial is decreased by 1 specified number of minimum change units, and the sub-identification difficulty is increased to the upper limit of the sub-identification difficulty. Both the second specified number and the third specified number are positive integers greater than 1, and the second specified number is greater than the third specified number. Preferably, the second specified number is 12, and the third specified number is 8.

[0173] Generally, assuming one training session per day, the entire training process can last from ten days to several months. During this process, each completed training session lays the foundation for subsequent training.

[0174] Optionally, the data recording and analysis module can also determine the primary training difficulty to be used at the start of the next training session based on the training data obtained in this training session, specifically as follows:

[0175] ;

[0176] in, This is the initial value for the overall image contrast used in the main training image at the start of the next training iteration; This is the initial value for the overall image contrast used in the main training image at the start of this training. This is the threshold for the overall image contrast of the main training image obtained after the analysis at the end of this training period; This is a function that takes the minimum value.

[0177] The data recording and analysis module is configured to record and analyze the current training data, and adjust the training parameters for the next training session based on the analysis results.

[0178] The data recording and analysis module can also analyze and adjust the upper limit of the spatial frequency filtering segment used in the main training image in the next training session based on the completion status of each training session. Specifically:

[0179] If, during the current training, the difference between the primary recognition difficulty threshold and the upper limit of the primary recognition difficulty is less than the minimum change unit specified in the fourth order, then in the next training iteration, the upper limit of the spatial frequency filtering segment used for the primary training image will be increased, i.e.: .

[0180] If, during the current training, the difference between the primary recognition difficulty threshold and the primary recognition difficulty lower limit when the training individual identifies the training image is less than the minimum change unit specified in the fourth order, then in the next training iteration, the upper limit of the spatial frequency filtering segment used for the primary training image will be reduced, i.e.: ;

[0181] in, This is the upper limit of the spatial frequency filtering segment used in the main training image during this training; This is the upper limit of the spatial frequency filtering segment used in the main training image during the next training iteration; This is a parameter, and its specific meaning is the same as in S1-3. The fourth specified quantity is a positive integer greater than 1. Preferably, the fourth specified quantity is 5.

[0182] The orientation θ of the raster in the auxiliary training image and the orientation information of the main content in the original training image. The relationship can also be: ,and ;in, It is a normal distribution function.

[0183] It should be noted that the strategy of adjusting the "initial value of the overall image contrast used in the main training image" or even the "upper limit of the spatial frequency filtering segment used in the main training image" based on the results of this training is to achieve better training results. Generally, after each training session, the contour recognition ability of the trained individual will improve to some extent, especially in the early stages of training. At this time, appropriately adjusting the initial value of the overall image contrast used in the main training image helps to reach the threshold level corresponding to the main training frequency more quickly in the next training session, thereby obtaining more training time near the threshold level and achieving better training results.

[0184] Furthermore, if the primary recognition difficulty threshold when training individuals to identify training images approaches the current upper limit of primary recognition difficulty, it indicates that the current training difficulty can no longer meet the required training difficulty. In this case, adjusting the upper limit of the spatial frequency filtering segment used in the primary training images can both increase the recognition difficulty of the primary training images and gradually make the training images more realistic, thus promoting the application of training results in real-life scenarios.

[0185] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0186] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0187] It should be noted that, in this document, the use of relational terms such as "first" and "second" is merely for distinguishing one entity or operation from another, and does not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0188] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0189] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0190] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0191] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0192] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0193] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0194] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A training system for improving contour recognition function in visually impaired individuals, characterized in that, include: The training image core parameter configuration module is configured to set the contrast segment and spatial frequency filtering segment of the main training image, as well as the spatial frequency of the auxiliary training image, based on the spatial contrast sensitivity curve of the training individual. The training image generation module is configured to generate visual stimulus images for training based on the original training images in the database, the parameters determined by the core parameter configuration module of the training images, and the parameters passed by the flow control module. Methods for generating training images include: S2-1. The training image is set to a circle, consisting of an inner circular region and an outer annular region. The relationship between the diameter of the inner circular region and the width of the outer annular region is as follows: ； S2-2. Based on the original training images in the database, determine the main training image within the inner circular region according to the setting of the spatial frequency filtering segment to be used. Specifically: ； in, It is a two-dimensional Gaussian low-pass filter function; The original training images in the database; This is the upper limit of the spatial frequency filtering section that should be used; These are parameters that describe the smoothness of the Gaussian low-pass filter function. Parameters for adjusting the overall contrast of the image; S2-3. Based on the information recorded in the training image database, determine the orientation information of the main content in the original training images used in S2-2. ; S2-4. Determine the auxiliary training images within the outer annular region. Specifically: ； in, This represents the brightness at each point (x, y) in the auxiliary training image; Average brightness; To assist in training the spatial frequency of the grating in the image; To enhance the contrast of the raster in the training images; The viewpoint occupied by each point (x, y) in the image; To assist in determining the orientation of the raster in the training image, the values ​​are set to the orientation information of the main content in the original training image. ; The phase of the grating in the training image is a random value; For image masking; The training process control module is configured to control the sequence between different trials during a single training session, and, based on the individual trainee's recognition of the training images, sends control signals to the training image generation module to adjust the training parameters in the next trial. The data recording and analysis module is configured to record and analyze the current training data, and adjust the training parameters for the next training session based on the analysis results.

2. The training system for improving contour recognition function in visually impaired individuals according to claim 1, characterized in that, The specific method for setting the contrast and spatial frequency ranges of the main training image is as follows: The lower limit of the contrast segment is set to the first value, and the upper limit is 1.

0. The lower limit of the spatial frequency filtering segment is 0, and the upper limit is the x-coordinate of the point on the spatial contrast sensitivity curve of the training individual, where the vertical axis is the reciprocal of the lower limit of the contrast segment to be used.

3. A training system for improving contour recognition function in visually impaired individuals according to claim 1, characterized in that, The method for determining the spatial frequency of auxiliary training images is as follows: S1-1. Calculate the difference between the spatial contrast sensitivity curve of the training individual and the preset spatial contrast sensitivity curve reference value, and perform standardization processing to convert the difference values ​​at all spatial frequencies into relative difference values ​​with a numerical range of [0,1]. S1-2. Use a preset model to fit the relative difference values ​​to the data, and determine the best model based on statistical analysis; S1-3. Determine the spatial frequencies to be used for auxiliary training images. .

4. A training system for improving contour recognition function in visually impaired individuals according to claim 1, characterized in that, Also includes: The recognition difficulty of the training images is determined by the main recognition difficulty and the secondary recognition difficulty. The main recognition difficulty corresponds to the main training image, and the parameters that adjust the overall contrast of the image are determined by the inner circular region. Decide, The initial value is the second value; The sub-recognition difficulty corresponds to the auxiliary training image and is determined by the contrast of the grating in the auxiliary training image within the outer annular region. The initial contrast value of the grating is the third value.

5. A training system for improving contour recognition function in visually impaired individuals according to claim 4, characterized in that, The specific process control method for determining the training parameters for the next trial in the current training iteration within the training process control module is as follows: S3-1. If the training process control module receives an abnormal situation prompt signal during the presentation of the training image in the previous trial, it determines that the training in the previous trial is "invalid" and sends a control signal to the training stimulus generation module to determine that the training image will not be adjusted in the next trial and will continue to use the training image from the previous trial; otherwise, it determines that it is "valid" and proceeds to S3-2. S3-2. If the training individual correctly identifies the training image in the previous trial, the sub-identification difficulty increases by 1 minimum change unit in the next trial; if the identification result is incorrect in the previous trial, the sub-identification difficulty decreases by a first specified number of minimum change units in the next trial; where the first specified number is a positive integer greater than 1. S3-3. If, in the most recently completed second specified number of trials, the sub-identification difficulty reaches the upper limit of the sub-identification difficulty in at least the third specified number of trials, then the main identification difficulty in the next trial is increased by 1 minimum change unit, and the sub-identification difficulty is reduced to the lower limit of the sub-identification difficulty; if, in the most recently completed second specified number of trials, the sub-identification difficulty reaches the lower limit of the sub-identification difficulty in at least the third specified number of trials, then the main identification difficulty in the next trial is reduced by 1 specified number of minimum change units, and the sub-identification difficulty is increased to the upper limit of the sub-identification difficulty.

6. A training system for improving contour recognition function in visually impaired individuals according to claim 5, characterized in that, The data recording and analysis module can also analyze and adjust the upper limit of the spatial frequency filtering segment used in the main training image in the next training session based on the completion status of each training session. Specifically: If, during the current training, the difference between the primary recognition difficulty threshold and the upper limit of the primary recognition difficulty when the training individual identifies the training image is less than the minimum change unit specified by the fourth number, then in the next training, the upper limit of the spatial frequency filtering segment used in the primary training image will be increased.

7. A training system for improving contour recognition function in visually impaired individuals according to claim 6, characterized in that, If, during the current training, the difference between the primary recognition difficulty threshold and the primary recognition difficulty lower limit when the training individual identifies the training image is less than the minimum change unit specified by the fourth number, then in the next training, the upper limit of the spatial frequency filtering segment used in the primary training image will be reduced.

8. A training system for improving contour recognition function in visually impaired individuals according to claim 1, characterized in that, Also includes: The human-computer interaction module is configured to include a display device, an information input device, a sound prompt device, and a monitoring device. The refresh rate of the display device is no less than the fourth value, enabling the training images to flicker at a maximum of 40Hz.

Images