Vision recovery device

Abstract

The invention relates to the field of vision recovery, in particular to a vision recovery device which comprises an eyeshade trainer, a biosensor, a training sensor, a right connecting cable, a left connecting cable, an adjusting buckle, a fixing buckle, an intelligent module, a fixing belt and an eye tracking sensor. A display screen is arranged in the eyeshade trainer, and a contact part is made of medical silica gel The biosensor is integrated with an intraocular pressure sensor, an ocular axis length sensor and an eye muscle electric signal sensor, the training sensor is integrated with a photosensitive element and a register, the connecting cable is made of high-purity oxygen-free copper and a shielding insulating material, a microprocessor and an image generation chip are arranged in the intelligent module, and the eye tracking sensor is integrated with an infrared light source and a camera. According to the device, customization and optimization adjustment of a real-time training mode and a periodic training mode are achieved through the intelligent module, different training modes are adopted according to the vision conditions of different users, real-time monitoring and adjustment are achieved, and the safety, effectiveness, pertinence and efficiency of vision recovery are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of vision restoration, and in particular to a vision restoration device. Background Art

[0002] With the widespread use of electronic devices in modern life and the change of eye - using habits, vision problems have become increasingly common. Vision impairments such as myopia, hyperopia, and astigmatism trouble a large number of people. Traditional vision restoration training methods lack accuracy and personalization. They mostly adopt a single fixed mode and cannot flexibly adjust the training plan according to the user's real - time eye condition and vision restoration progress. For example, common vision training instruments only provide simple image stimulation. They can neither comprehensively collect the user's eye physiological data nor dynamically optimize the training parameters during the training process, resulting in poor training effects and even possibly causing additional burden on the eyes due to improper training. Therefore, it is of great practical significance to develop a vision restoration device that can accurately analyze the user's eye condition and adjust the training mode in real - time and in a personalized manner.

[0003] Chinese Patent Publication No.: CN108379043A discloses a vision restoration treatment device, which includes a treatment darkroom component, a lens degree selection and positioning mechanism, 2 sets of lens mounting components, and an operating rod component; a visual window is provided on the front cover of the treatment darkroom component, a visual display screen is installed on the inner wall of the rear cover, an operation screen is fixed on the inclined surface at the side end of the rear cover, the visual display screen is communicatively connected to the operation screen, and the operation screen is communicatively connected to the operating rod component. The visual display screen plays a dynamic visual picture for treatment according to the operation instruction; the lens degree selection and positioning mechanism is installed inside the treatment darkroom component, and the 2 sets of lens mounting components are respectively slidably connected to the lens degree selection and positioning mechanism, and the distance between the two sets of lens mounting components is adjusted by the lens degree selection and positioning mechanism. Different - degree lenses can be automatically replaced, and the user watches the dynamic visual picture. Through the cooperation of lens adjustment and the display of the treatment dynamic visual picture for treatment, the muscles around the eyes are relaxed, thereby helping the user restore visual function. However, this solution still has a relatively single monitoring of the user's eye condition. It only adapts to different vision conditions by changing the lens degree, and cannot accurately obtain multi - dimensional eye physiological data, resulting in a lack of sufficient pertinence in the treatment plan. Moreover, during the training process, it lacks a real - time dynamic adjustment mechanism, which makes it difficult for the training to fit the changes of the user during the training process, easily resulting in poor training effects and difficult to fully tap the user's vision restoration potential. The overall treatment efficiency and effect are limited. Therefore, the existing technology has obvious deficiencies in the accuracy of vision restoration treatment, real - time dynamic adjustment ability, and personalized cycle planning, and there is an urgent need for a more advanced and intelligent vision restoration device to solve these problems. Summary of the Invention

[0004] To this end, the present invention provides a vision restoration device to overcome the problem of low efficiency in vision restoration treatment in the prior art due to the inability to perform real-time adjustment during vision restoration treatment and the inability to provide real-time feedback adjustment for cycle planning.

[0005] To achieve the above object, the present invention provides a vision restoration device, comprising:

[0006] An eye mask trainer, one side of which close to the fixing strap is connected to a biological sensor, a training sensor and an eye tracking sensor, and a display screen is arranged inside it for generating vision training images;

[0007] A biological sensor, which is installed on one side of the eye mask trainer close to the fixing strap, for collecting the eye physiological data of the user;

[0008] A training sensor, which is installed on one side of the eye mask trainer close to the fixing strap and away from the biological sensor, for collecting the image training data of the user;

[0009] A right connection cable and a left connection cable, one end of each of which is connected to the eye mask trainer and the other end is connected to the intelligent module, for transmitting the image training data of the user, the eye physiological data of the user and the vision training process data of the user;

[0010] An adjustment buckle, which is installed on the fixing strap for adjusting the length of the fixing strap;

[0011] A fixing buckle, which is installed at the junction of the intelligent module and the right connection cable and the left connection cable for fixing the right connection cable and the left connection cable;

[0012] An intelligent module, which is connected to the fixing strap, the right connection cable and the left connection cable, for performing conversion processing on data and generating a real-time training mode and a cycle training mode;

[0013] A fixing strap, one end of which is connected to the eye mask trainer and the other end is connected to the intelligent module for wearing and fixing;

[0014] An eye tracking sensor, which is installed on one side of the eye mask trainer away from the biological sensor and the training sensor, for collecting the vision training process data of the user;

[0015] Further, the intelligent module includes:

[0016] An acquisition unit for acquiring the image training data of the user, the eye physiological data of the user and the vision training process data transmitted by the right connection cable and the left connection cable, taking the image training data of the user as the first target real-time data, taking the eye physiological data of the user as the first target analysis data, and taking the vision training process data of the user as the first target comparison data;

[0017] A processing unit for performing conversion processing on first target real-time data, first target analysis data, and first target comparison data to obtain second target analysis data, second target real-time data, and target comparison data;

[0018] A real-time training mode generation unit for generating a real-time training mode according to the second target analysis data;

[0019] A real-time adjustment unit for comparing and analyzing the second target real-time data and the target comparison data, and adjusting the real-time training mode according to the analysis result to obtain an optimal real-time training mode;

[0020] A periodic training mode generation unit for generating a periodic training mode according to the second target analysis data;

[0021] A periodic adjustment unit for performing periodic analysis on the second target analysis data, and optimizing the periodic training mode according to the analysis result to obtain an optimal periodic training mode.

[0022] Further, when the obtaining unit is used to obtain the image training data, the eye physiological data of the user, and the vision training process data transmitted by the right connection cable and the left connection cable, it obtains the first target real-time data by obtaining the data of the user's eye movement trajectory and the image quality index through the training sensor, obtains the first target analysis data by obtaining the pressure value inside the eyeball, the anteroposterior diameter length of the eyeball, and the bioelectric signals generated during the contraction and relaxation of the eye muscles through the biosensor, and obtains the first target comparison data by obtaining the coincidence degree between the user's eye movement trajectory and the preset eye movement trajectory through the eye tracking sensor.

[0023] Further, when the processing unit is used to perform conversion processing on the first target real-time data and the first target analysis data, it constructs a composite filtering conversion model according to the historical conversion data, divides the historical conversion data into a 70% simulated training set, a 20% simulated validation set, and a 10% simulated test set, sets the hidden_size in the recurrent neural network model parameters to 128 and the num_layers to 4 layers, inputs the simulated training set into the recurrent neural network model with the parameters set for training, inputs the simulated validation set into the trained recurrent neural network model to optimize the parameters of the recurrent neural network model, then inputs the test set into the recurrent neural network model with the optimized parameters for testing, outputs the test accuracy rate R, compares the test accuracy rate R with the preset test accuracy rate R0, judges the compliance of the recurrent neural network model with the optimized parameters according to the comparison result, and outputs the recurrent neural network model with the optimized parameters according to the judgment result, where:

[0024] When R≥R0, it is determined that the recurrent neural network model after parameter optimization meets the standard, and the recurrent neural network model is output as a composite filtering conversion model;

[0025] When R<R0, it is determined that the recurrent neural network model after parameter optimization does not meet the standard. Obtain the second historical conversion data, and train the recurrent neural network model according to the second historical conversion data until the comparison result between the test accuracy rate R of the recurrent neural network model and the preset test accuracy rate R0 is R≥R0, and output the recurrent neural network model as a composite filtering conversion model;

[0026] Input the first target real-time data, the first target analysis data, and the first target comparison data into the composite filtering conversion model for filtering conversion to obtain the second target real-time data, the second target analysis data, and the target comparison data.

[0027] Further, when the real-time training mode generation unit generates the real-time training mode according to the second target analysis, compare the second target analysis data with the preset analysis data in the real-time special training mode database, where:

[0028] When there is preset analysis data in the real-time special training mode database that is consistent with the second target analysis data, output the image moving distance, the image contrast value, and the image brightness value corresponding to the preset analysis data as the initial training parameter values;

[0029] When there is no preset analysis data in the real-time special training mode database that is consistent with the second target analysis data, push the second target analysis data to the special user terminal to obtain the initial training parameter values given by the special user terminal;

[0030] Input the initial training parameter values into the intelligent module to obtain the real-time training mode.

[0031] Further, when the real-time adjustment unit is used to compare and analyze the second target real-time data and the target comparison data, and adjust the real-time training mode according to the analysis result, obtain the second target real-time data training outlier P, compare the second target real-time data training outlier P with the preset training outlier P0, judge the abnormal degree of the second target real-time data training according to the comparison result, and adjust the real-time training mode according to the judgment result, where:

[0032] When P≤P0, it is determined that the abnormal degree of the second target real-time data training is low, and the real-time training mode is not adjusted;

[0033] When P>P0, it is determined that the abnormal degree of the second target real-time data training is high, and the real-time training mode is adjusted;

[0034] Obtain the coincidence degree M between the eye movement trajectory in the second target real-time data and the preset eye movement trajectory according to the target comparison data. Compare the coincidence degree M between the eye movement trajectory of the second target real-time data and the visual stimulus image trajectory with the preset coincidence degree M0. Judge the compliance of the coincidence degree M between the eye movement trajectory of the second target real-time data and the visual stimulus image trajectory according to the comparison result, and update the training outlier P of the second target real-time data according to the judgment result, where:

[0035] When M≥M0, it is determined that the coincidence degree between the eye movement trajectory of the second target real-time data and the preset eye movement trajectory is high, and the training outlier P of the second target real-time data is not updated;

[0036] When M<M0, it is determined that the coincidence degree between the eye movement trajectory of the second target real-time data and the preset eye movement trajectory is low, and the training outlier P of the second target real-time data is updated. Set the update coefficient as A, A = 1.6 + 0.6×e -0.7×(M0-M) , and the updated training outlier of the second target real-time data is P1, P1 = P×A;

[0037] Obtain the image quality index Z in the second target real-time data. Compare the image quality Z in the second target real-time data with the preset quality index Z0. Judge the clarity of the image in the second target real-time data according to the comparison result, and optimize the coincidence degree M between the eye movement trajectory in the second target real-time data and the preset eye movement trajectory according to the judgment result, where:

[0038] When Z≥Z0, it is determined that the clarity of the image in the second target real-time data is normal, and the coincidence degree M between the eye movement trajectory in the second target real-time data and the preset eye movement trajectory is not optimized;

[0039] When Z<Z0, it is determined that the clarity of the image in the second target real-time data is abnormal, and the coincidence degree M between the eye movement trajectory in the second target real-time data and the preset eye movement trajectory is optimized. Set the optimization coefficient as B, B = 0.4 + 0.6×e -0.3×(Z0-Z) , and the optimized coincidence degree between the eye movement trajectory in the second target real-time data and the preset eye movement trajectory is M1, M1 = M×(1 - B);

[0040] Obtain the image moving distance S within a short time period according to the target comparison data. Compare the image moving distance S within a short time period with the preset moving distance S0. Judge the image moving speed according to the comparison result, and adjust the image moving speed according to the judgment result, where:

[0041] When S = S0, it is determined that the image moving speed is normal, and the image moving speed is not adjusted;

[0042] When S > S0, it is determined that the image moving speed is too fast, and the image moving speed is slowed down until S = S0;

[0043] When S < S0, it is determined that the image moving speed is too slow, and the image moving speed is accelerated until S = S0;

[0044] When the real-time adjustment unit adjusts the real-time training mode, it obtains the image contrast value D, compares the image contrast value D with the preset image contrast standard value D0, judges the image contrast value according to the comparison result, and adjusts the image contrast value according to the judgment result, where:

[0045] When D = D0, it is determined that the image contrast value is normal, and the image contrast value is not adjusted;

[0046] When D > D0, it is determined that the image contrast value is too high, and the image contrast value is decreased until D = D0;

[0047] When D < D0, it is determined that the image contrast value is too low, and the image contrast value is increased until D = D0;

[0048] When the real-time adjustment unit adjusts the real-time training mode, it obtains the image brightness value C, compares the image brightness value C with the preset image brightness standard value C0, judges the magnitude of the image brightness according to the comparison result, and adjusts the image brightness value according to the judgment result, where:

[0049] When C = C0, it is determined that the image brightness value is normal, and the image brightness value is not adjusted;

[0050] When C > C0, it is determined that the image brightness value is too high, and the image brightness value is reduced until C = C0;

[0051] When C < C0, it is determined that the image brightness value is too low, and the image brightness value is increased until C = C0.

[0052] Furthermore, when the periodic training mode generation unit generates the periodic training mode according to the second target analysis data, it compares the second target analysis data with the preset analysis data in the periodic special training mode database, where:

[0053] When there is a preset recovery degree in the periodic special training mode database that is consistent with the second target analysis data, the preset periodic training mode corresponding to the preset analysis data is output as the periodic training mode;

[0054] When there is no preset recovery degree in the periodic special training mode database that is consistent with the second target analysis data, push the second target analysis data to the special user terminal and obtain the periodic training mode given by the special user terminal.

[0055] Further, when the periodic adjustment unit is used to perform a periodic analysis on the second target analysis data and optimize the periodic training mode according to the analysis results, an update period is set. At the end of each update period, obtain the vision abnormality value G of the second target analysis data, compare the vision abnormality value G of the second target analysis data with the preset vision abnormality value G0, judge the vision abnormality degree of the second target analysis data according to the comparison result, and adjust the periodic training mode according to the judgment result, where:

[0056] When G ≤ G0, it is determined that the vision abnormality degree of the second target analysis data is low, and the periodic training mode is not adjusted;

[0057] When G > G0, it is determined that the vision abnormality degree of the second target analysis data is high, and the periodic training mode is adjusted;

[0058] Obtain the intraocular pressure value H of the second target analysis data, compare the intraocular pressure value H of the second target analysis data with the preset intraocular pressure value H0, judge the intraocular pressure abnormality degree of the second target analysis data according to the comparison result, and update the vision abnormality value G of the second target analysis data according to the judgment result, where:

[0059] When H ≤ H0, it is determined that the intraocular pressure abnormality value degree of the second target analysis data is low, and the vision abnormality value G of the second target analysis data is not updated;

[0060] When H > H0, it is determined that the intraocular pressure abnormality value degree of the second target analysis data is high, and the vision abnormality value G of the second target analysis data is updated. Set the update coefficient as A1, A1 = 1.4 - 0.4×e -0.8×(H-H0) , and the updated vision abnormality value of the second target analysis data is G1, G1 = G × A1;

[0061] Obtain the eye axis length abnormality value K of the second target analysis data, compare the eye axis length abnormality value K of the second target analysis data with the preset eye axis length abnormality value K0, judge the eye axis length abnormality degree of the second target analysis data according to the comparison result, and update the intraocular pressure abnormality value G of the second target analysis data according to the judgment result, where:

[0062] When K ≤ K0, it is determined that the eye axis length abnormality degree of the second target analysis data is low, and the vision abnormality value G of the second target analysis data is not updated;

[0063] When K > K0, it is determined that the abnormal degree of the axial length of the second target analysis data is high, and the abnormal value G of the vision of the second target analysis data is updated. The update coefficient is set as A2, and A2 = 1.59 - 0.51×e -0.7×(K-K0) , and the updated abnormal value G2 of the vision of the second target analysis data is G2 = G×A2;

[0064] Obtain the abnormal value L of the biological signal of the eye muscles of the second target analysis data, compare the abnormal value L of the biological signal of the eye muscles of the second target analysis data with the preset abnormal value L0 of the biological signal of the eye muscles, judge the abnormal degree of the biological signal of the eye muscles of the data according to the comparison result, and update the abnormal value G of the intraocular pressure of the second target analysis data according to the judgment result, where:

[0065] When L ≤ L0, it is determined that the abnormal degree of the biological signal of the eye muscles of the second target analysis data is low, and the abnormal value G of the vision of the second target analysis data is not updated;

[0066] When L > L0, it is determined that the abnormal degree of the biological signal of the eye muscles of the second target analysis data is high, and the abnormal value G of the vision of the second target analysis data is updated. The update coefficient is set as A3, and A3 = 2.58 - 1.52×e -0.6×(L-L0) , and the updated abnormal value G3 of the vision of the second target analysis data is G3 = G×A3;

[0067] Obtain the abnormal degree G1 of vision at the end of the previous cycle of training, the abnormal degree G2 of vision at the end of the current cycle of training, the preset adjustment coefficient k, and the preset vision recovery coefficient as X. According to the formula Obtain the vision recovery coefficient X, compare the recovery degree X of the target cycle data with the preset recovery degree X0, judge the vision recovery degree according to the comparison result, and adjust the training intensity of the training mode according to the judgment result, where:

[0068] When X < X0, it is determined that the vision recovery degree is low, and the training intensity of the training mode is strengthened until X = X0;

[0069] When X = X0, it is determined that the vision recovery degree is normal, and the training intensity of the training mode is not adjusted;

[0070] When X > X0, it is determined that the vision recovery degree is high, and the training intensity of the training mode is reduced until X = X0.

[0071] Compared with the prior art, the beneficial effects of the present invention are that, based on multi-dimensional eye physiological data, a real-time and periodic training mode suitable for individuals is generated, training data is collected in real time, training parameters are accurately adjusted, and according to the vision recovery process, multi-factor dynamic adjustment of training intensity and plan is performed periodically to ensure the safety and effectiveness of training and improve the recovery efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 is a schematic structural diagram of the vision restoration device of this embodiment;

[0073] Figure 2 is a schematic structural diagram of the intelligent module of this embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0074] In order to make the objectives and advantages of the present invention more clear and understandable, the present invention will be further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0075] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present invention and do not limit the protection scope of the present invention.

[0076] It should be noted that in the description of the present invention, the terms indicating directions or positional relationships such as "upper", "lower", "left", "right", "inner", "outer", etc. are based on the directions or positional relationships shown in the drawings. This is only for convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention.

[0077] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and defined, the terms "installation", "connection", and "connection" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two elements. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0078] Please refer to Figure 1 as shown, which is a schematic structural diagram of the vision restoration device of this embodiment. The device includes:

[0079] An eye mask trainer 1, one side of which close to the fixing band 7 is connected to a biological sensor 201, a training sensor 202, and an eye tracking sensor 8, and a display screen is arranged inside it for generating vision training images;

[0080] A biological sensor 201, which is installed on one side of the eye mask trainer 1 close to the fixing band 7, for collecting the physiological data of the user's eyes;

[0081] The training sensor 202 is installed on one side of the eye mask trainer 1 close to the fixing strap 7 and far from the biosensor 201 for collecting the image training data of the user.

[0082] The right connection cable 301 and the left connection cable 302, one end of which is connected to the eye mask trainer 1 and the other end is connected to the intelligent module 6, are used for transmitting the image training data of the user, the eye physiological data of the user and the vision training process data of the user.

[0083] The adjusting buckle 4 is installed on the fixing strap 7 for adjusting the length of the fixing strap 7.

[0084] The fixing buckle 5 is installed at the connection of the intelligent module 6 with the right connection cable 301 and the left connection cable 302 for fixing the right connection cable 301 and the left connection cable 302.

[0085] The intelligent module 6 is connected to the fixing strap 7, the right connection cable 301 and the left connection cable 302 for converting and processing data and generating a real-time training mode and a periodic training mode.

[0086] The fixing strap 7, one end of which is connected to the eye mask trainer 1 and the other end is connected to the intelligent module 6, is used for wearing and fixing.

[0087] The eye tracking sensor 8 is installed on one side of the eye mask trainer 1 far from the sensors 201 and 202 for collecting the vision training process data of the user.

[0088] Specifically, the device is applied to vision rehabilitation training, adopts different training modes according to the vision conditions of different users, monitors the training process data of the user in real time during use, feeds the data back to the intelligent module for analysis, adjusts the accuracy of vision training in real time, improves the safety and effectiveness of vision recovery, and sets a periodic vision training mode to adjust vision training according to the vision recovery situation of the user, thus improving the vision recovery efficiency.

[0089] Specifically, the eye contact part of the eye mask trainer 1 is made of medical silicone material, and the internal display screen displays vision training images.

[0090] Specifically, this embodiment does not limit the display screen, and those skilled in the relevant art can freely set it according to actual needs as long as the image display requirements are met, such as liquid crystal materials and quantum dot materials. This embodiment does not limit the vision training images, and those skilled in the relevant art can freely set them according to actual needs as long as the images that can prompt the eye muscles of the user to react are met, such as black and white stripes and color gradient patterns with different frequencies.

[0091] Specifically, the eye mask trainer 1 displays the vision training images generated within the intelligent module on the internal display screen through the connection cable 3, utilizes the softness of medical silicone to ensure comfort and a tight fit, blocks external light interference, and effectively conducts vision training.

[0092] Specifically, the biosensor 201 integrates an intraocular pressure sensor, an axial length sensor of the eye, and an electrophysiological signal sensor of the eye muscles.

[0093] Specifically, the intraocular pressure sensor refers to a device specifically used to measure the internal pressure of the eyeball, the axial length sensor of the eye refers to a device used to measure the length of the anteroposterior diameter of the eyeball, and the electrophysiological signal sensor of the eye muscles refers to a device used to detect the bioelectrical signals generated during the contraction and relaxation of the eye muscles.

[0094] Specifically, according to the eye physiological data of the user collected by the biosensor 201, the best vision training mode is selected, and the vision training is adjusted in a timely manner according to the subtle changes in the user's eye condition, ensuring that the vision restoration treatment always fits the actual vision condition of the user and improving the effectiveness of the user's vision restoration.

[0095] Specifically, the training sensor 202 integrates a photosensitive element and a register.

[0096] Specifically, the photosensitive element refers to a component that captures light and distinguishes the color, brightness, and texture information of an object, and the register refers to a component used to temporarily store instruction and data information.

[0097] Specifically, according to the image training data of the user collected by the training sensor 202, the training condition during the user's real-time training process is adjusted to improve the accuracy and safety of vision restoration.

[0098] Specifically, the right connection cable 301 and the left connection cable 302 are made of high-purity oxygen-free copper and are externally covered with a shielding insulation material.

[0099] Specifically, the high-purity oxygen-free copper refers to a copper material with a copper content of more than 99.95% and a very low oxygen content. In this embodiment, the shielding insulation material is not limited, and those skilled in the relevant art can freely set it according to actual needs, as long as the requirements for good insulation performance and mechanical performance are met, such as polyvinyl chloride and polyethylene.

[0100] Specifically, the right connection cable 301 and the left connection cable 302 connect the eye mask trainer and the host, and efficiently transmit the image training data of the user, the eye physiological data of the user, and the vision training process data of the user. The high-purity oxygen-free copper and the shielding insulation material are used to effectively resist external electromagnetic interference and ensure the reliability of communication between modules.

[0101] Specifically, the adjustment buckle 4 is a sliding buckle made of metal material.

[0102] Specifically, in this embodiment, the sliding buckle is not limited, and those skilled in the relevant art can freely set it according to actual needs, as long as it meets the requirement of being adjustable, such as a D-ring buckle and an 8-shaped buckle.

[0103] Specifically, the adjustment buckle 4 can quickly and conveniently adjust the length of the fixing belt according to the user's own needs, improve the flexibility of the device, and ensure the visual training effect.

[0104] Specifically, the fixing buckle 5 is a fixed buckle.

[0105] Specifically, in this embodiment, the fixed buckle is not limited, and those skilled in the relevant art can freely set it according to actual needs, as long as it meets the requirement of fixation, such as a magnetic buckle and a snap buckle.

[0106] Specifically, the fixing buckle 5 ensures the fixing effect of the device, reduces the risk of injury during wearing, and improves the stability and safety of the device.

[0107] Specifically, the intelligent module 6 is built-in with a microprocessor and an image generation chip.

[0108] Specifically, the microprocessor refers to an integrated circuit chip that is the core component of a computer system, and the image generation chip refers to an integrated circuit chip specifically designed to process image and graphic data.

[0109] Specifically, the microprocessor of the intelligent module 6 has computing power and task processing ability, can quickly analyze the preset program instructions and parameters, improve the training efficiency of the device, and the image generation chip can generate rich and diverse, high-precision specific visual stimulation images, improving the applicability and effectiveness of the device.

[0110] Specifically, the fixing belt 7 is made of elastic material.

[0111] Specifically, the elastic material refers to a material that can deform when subjected to an external force and can return to its original shape and size when the external force is removed.

[0112] Specifically, the fixing belt 7 can firmly fix the device on the user's head according to the different sizes and shapes of the user's head, ensure the stable position of the device during use, and improve the stability of the device.

[0113] Specifically, the eye tracking sensor 8 integrates an infrared light source, a camera, an image processor, and a data transmission interface.

[0114] Specifically, the infrared light source refers to a light source that can emit infrared rays, the camera refers to a device that uses the principle of optical imaging to convert a scene into an electrical signal or a digital signal, the image processor refers to a device specifically used to process image data, and the data transmission interface refers to the physical interface and communication protocol for transmitting data between different devices.

[0115] Specifically, the eye tracking sensor 8 collects the data of the user's vision training process, and transmits the collected data of the user's vision training process to the intelligent module 6 through the right connection cable 301 and the left connection cable 302 in real time for intelligent analysis, adjusts the real-time training mode in a timely manner, and improves the vision recovery efficiency.

[0116] Please refer to Figure 2 as shown, which is a schematic structural diagram of the intelligent module of this embodiment. The intelligent module includes:

[0117] An acquisition unit for acquiring the user's image training data, the user's eye physiological data, and the user's vision training process data transmitted by the right connection cable and the left connection cable, taking the user's image training data as the first target real-time data, taking the user's eye physiological data as the first target analysis data, and taking the user's vision training process data as the first target comparison data;

[0118] A processing unit for performing conversion processing on the first target real-time data, the first target analysis data, and the first target comparison data to obtain the second target analysis data, the second target real-time data, and the target comparison data. The processing unit is connected to the real-time training mode generation unit;

[0119] A real-time training mode generation unit for generating a real-time training mode according to the second target analysis data. The real-time training mode generation unit is connected to the processing unit;

[0120] A real-time adjustment unit for comparing and analyzing the second target real-time data and the target comparison data, and adjusting the real-time training mode according to the analysis result to obtain the real-time optimal training mode. The real-time adjustment unit is connected to the real-time training mode generation unit;

[0121] A periodic training mode generation unit for generating a periodic training mode according to the second target analysis data. The periodic training mode generation unit is connected to the processing unit;

[0122] A periodic adjustment unit for performing periodic analysis on the second target analysis data, and optimizing the periodic training mode according to the analysis result to obtain the periodic optimal training mode. The periodic adjustment unit is connected to the periodic training mode generation unit.

[0123] Specifically, the intelligent module is applied to the vision restoration device described in this embodiment. By intelligently controlling the vision restoration device, it realizes the customization of the real-time training mode and the cycle training mode during the treatment process of the vision restoration device, as well as the optimized adjustment of the customization of the real-time training mode and the cycle training mode, so as to achieve the intelligent adjustment and control of the vision restoration device, ensure the accuracy and safety of the device treatment operation. The acquisition unit integrates various types of sensors to improve the richness and accuracy of data. The processing unit performs conversion processing on the acquired data to remove interference signals and improve the accuracy and stability of the treatment effect. The real-time adjustment unit monitors the vision training situation in real time, enabling the user to obtain the best training plan during use, improving the vision restoration efficiency, and enhancing the treatment safety. The cycle adjustment unit adjusts the periodic training mode for the user in line with the physiological rhythm, adapts to individual differences, and improves the treatment precision.

[0124] Specifically, when the acquisition unit is used to acquire the user's image training data, the user's eye physiological data, and the user's vision training process data transmitted by the right connection cable and the left connection cable, it acquires the data of the user's eye movement trajectory and the image quality index through the training sensor to obtain the first target real-time data, acquires the pressure value inside the eyeball, the anteroposterior diameter length of the eyeball, and the bioelectric signals generated during the contraction and relaxation of the eye muscles through the biosensor to obtain the first target analysis data, and acquires the coincidence degree between the user's eye movement trajectory and the preset eye movement trajectory through the eye tracking sensor to obtain the first target comparison data.

[0125] Specifically, the image quality index refers to an index used to measure the image quality, and the bioelectric signal refers to an electric signal generated in a living body due to the metabolic activities or physiological function changes of cells.

[0126] Specifically, the acquisition unit acquires the user's image training data, the user's eye physiological data, and the user's vision training process data through multiple sensors, improves the data accuracy, facilitates subsequent adjustment of the training mode, and increases the training safety.

[0127] Specifically, when the processing unit is used to perform conversion processing on the first target real-time data and the first target analysis data, a composite filtering conversion model is constructed according to the historical conversion data. The historical conversion data is divided into a 70% simulation training set, a 20% simulation validation set, and a 10% simulation test set. The hidden_size in the parameters of the recurrent neural network model is set to 128, and the num_layers is set to 4 layers. The simulation training set is input into the recurrent neural network model with the parameters set for training, and the simulation validation set is input into the trained recurrent neural network model to optimize the parameters of the recurrent neural network model. Then, the test set is input into the recurrent neural network model with the optimized parameters for testing, and the test accuracy rate R is output. The test accuracy rate R is compared with the preset test accuracy rate R0, and the compliance of the recurrent neural network model with the optimized parameters is judged according to the comparison result, and the recurrent neural network model with the optimized parameters is output according to the judgment result, where:

[0128] When R≥R0, it is determined that the recurrent neural network model with the optimized parameters is compliant, and the recurrent neural network model is output as the composite filtering conversion model;

[0129] When R<R0, it is determined that the recurrent neural network model with the optimized parameters is not compliant. The second historical conversion data is obtained, and the recurrent neural network model is trained according to the second historical conversion data until the comparison result between the test accuracy rate R of the recurrent neural network model and the preset test accuracy rate R0 is R≥R0, and the recurrent neural network model is output as the composite filtering conversion model;

[0130] The first target real-time data, the first target analysis data, and the first target comparison data are input into the composite filtering conversion model for filtering conversion to obtain the second target real-time data, the second target analysis data, and the target comparison data.

[0131] Specifically, the historical conversion data refers to the recorded information of data conversions that have occurred in the past and the conversion result data, such as interference signals and data conversion information. In this embodiment, the method for obtaining the historical conversion data is not limited, and those skilled in the relevant art can freely set it according to actual needs, as long as it meets the requirements for constructing the recurrent neural network model. For example, the method for obtaining the historical conversion data can be set to obtain it through big data. The simulation training set refers to the data set in the historical conversion data used to train the recurrent neural network model. The validation set refers to the data set used to adjust the parameters of the recurrent neural network model during the training process. The test set refers to the data set used to evaluate the final performance of the recurrent neural network model after the training of the recurrent neural network model is completed. The recurrent neural network model refers to a neural network model used to process sequential data. The hidden_size refers to the number of neurons in the hidden layer, and the num_layers refers to the number of hidden layers in the recurrent neural network model.

[0132] Specifically, the processing unit constructs a composite filtering conversion model to remove interference signals and perform signal conversion on the first target real-time data, the first target analysis data, and the first target comparison data, improving the accuracy of the system, providing a precise basis for subsequent generation and adjustment of the training mode, and enhancing the effectiveness of the vision restoration treatment effect.

[0133] Specifically, when the real-time training mode generation unit generates the real-time training mode based on the second target analysis, it compares the second target analysis data with the preset analysis data in the real-time special training mode database, where:

[0134] When there is preset analysis data in the real-time special training mode database that is consistent with the second target analysis data, the image movement distance, the image contrast value, and the image brightness value corresponding to the preset analysis data are output as the initial training parameter values;

[0135] When there is no preset analysis data in the real-time special training mode database that is consistent with the second target analysis data, the second target analysis data is pushed to the special user terminal to obtain the initial training parameter values given by the special user terminal;

[0136] The initial training parameter values are input into the intelligent module to obtain the real-time training mode.

[0137] Specifically, the real-time special training mode database refers to a database system that stores and integrates vision training knowledge, vision training experience, and vision training research results, including preset analysis data and initial parameter values. In this embodiment, the setting method of the real-time special training mode database is not limited, and those skilled in the art can freely set it according to the actual situation, as long as the output requirements for the initial parameter values are met. For example, it can be set to set the real-time special training mode database through big data and update the real-time special training mode database in real time. The preset analysis data refers to the pre-set and stored cases of simulated real user's eye physiological data, and the initial parameter values refer to the standardized training parameters pre-designed and stored for the preset analysis data. The special user terminal refers to the expert user's terminal with the ability to set analysis data. In this embodiment, the acquisition method of the initial training parameter values given by the special user terminal is not limited, and relevant personnel in the art can freely set it according to the actual situation, as long as the acquisition requirements for the initial training parameter values given by the special user terminal are met. For example, it can be set to push an acquisition window to the special user terminal, and the special user inputs in the acquisition window, and the content input by the special user in the acquisition window is taken as the initial training parameter values given by the special user terminal. The real-time training mode refers to a mode of real-time training setting based on the user's vision condition.

[0138] Specifically, the real-time training mode generation unit compares the first target analysis data with the preset analysis data in the real-time special training mode database to generate a real-time training mode, and realizes the precise customization of the real-time training mode according to the user's eye condition, improving the pertinence and effectiveness of the system's vision training.

[0139] Specifically, when the real-time adjustment unit is used to compare and analyze the second target real-time data and the target comparison data and adjust the real-time training mode according to the analysis result, the training outlier P of the second target real-time data is obtained, and the training outlier P of the second target real-time data is compared with the preset training outlier P0. According to the comparison result, the outlier degree of the second target real-time data training is judged, and the real-time training mode is adjusted according to the judgment result, where:

[0140] When P ≤ P0, it is determined that the outlier degree of the second target real-time data training is low, and the real-time training mode is not adjusted;

[0141] When P > P0, it is determined that the outlier degree of the second target real-time data training is high, and the real-time training mode is adjusted;

[0142] Obtain the coincidence degree M between the eye movement trajectory in the second target real-time data and the preset eye movement trajectory according to the target comparison data, compare the coincidence degree M between the eye movement trajectory of the second target real-time data and the visual stimulus image trajectory with the preset coincidence degree M0, judge the compliance of the coincidence degree M between the eye movement trajectory of the second target real-time data and the visual stimulus image trajectory according to the comparison result, and update the training outlier P of the second target real-time data according to the judgment result, where:

[0143] When M≥M0, it is determined that the coincidence degree between the eye movement trajectory of the second target real-time data and the preset eye movement trajectory is high, and the training outlier P of the second target real-time data is not updated;

[0144] When M<M0, it is determined that the coincidence degree between the eye movement trajectory of the second target real-time data and the preset eye movement trajectory is low, and the training outlier P of the second target real-time data is updated. Set the update coefficient as A, A = 1.6 + 0.6×e -0.7×(M0-M) , and the updated training outlier of the second target real-time data is P1, P1 = P×A;

[0145] Obtain the image quality index Z in the second target real-time data, compare the image quality Z in the second target real-time data with the preset quality index Z0, judge the clarity of the image in the second target real-time data according to the comparison result, and optimize the coincidence degree M between the eye movement trajectory in the second target real-time data and the preset eye movement trajectory according to the judgment result, where:

[0146] When Z≥Z0, it is determined that the clarity of the image in the second target real-time data is normal, and the coincidence degree M between the eye movement trajectory in the second target real-time data and the preset eye movement trajectory is not optimized;

[0147] When Z<Z0, it is determined that the clarity of the image in the second target real-time data is abnormal, and the coincidence degree M between the eye movement trajectory in the second target real-time data and the preset eye movement trajectory is optimized. Set the optimization coefficient as B, B = 0.4 + 0.6×e -0.3×(Z0-Z) , and the optimized coincidence degree between the eye movement trajectory in the second target real-time data and the preset eye movement trajectory is M1, M1 = M×(1 - B);

[0148] Obtain the image movement distance S within a short time period according to the target comparison data, compare the image movement distance S within the short time period with the preset movement distance S0, judge the image movement speed according to the comparison result, and adjust the image movement speed according to the judgment result, where:

[0149] When S = S0, it is determined that the image movement speed is normal, and the image movement speed is not adjusted;

[0150] When S > S0, it is determined that the image moving speed is too fast, and the image moving speed is slowed down until S = S0;

[0151] When S < S0, it is determined that the image moving speed is too slow, and the image moving speed is accelerated until S = S0;

[0152] When the real-time adjustment unit adjusts the real-time training mode, it obtains the image contrast value D, compares the image contrast value D with the preset image contrast standard value D0, judges the image contrast value according to the comparison result, and adjusts the image contrast value according to the judgment result, where:

[0153] When D = D0, it is determined that the image contrast value is normal, and the image contrast value is not adjusted;

[0154] When D > D0, it is determined that the image contrast value is too high, and the image contrast value is reduced until D = D0;

[0155] When D < D0, it is determined that the image contrast value is too low, and the image contrast value is increased until D = D0;

[0156] When the real-time adjustment unit adjusts the real-time training mode, it obtains the image brightness value C, compares the image brightness value C with the preset image brightness standard value C0, judges the size of the image brightness according to the comparison result, and adjusts the image brightness value according to the judgment result, where:

[0157] When C = C0, it is determined that the image brightness value is normal, and the image brightness value is not adjusted;

[0158] When C > C0, it is determined that the image brightness value is too high, and the image brightness value is reduced until C = C0;

[0159] When C < C0, it is determined that the image brightness value is too low, and the image brightness value is increased until C = C0.

[0160] Specifically, this embodiment does not limit the short time period, and relevant personnel in the field can freely set it according to the actual situation, as long as it meets the requirement of measuring the image moving speed, such as setting the short time period to one cycle every 2 seconds.

[0161] Specifically, the real-time adjustment unit improves the accuracy of judging the training situation by comparing and analyzing the second target real-time data and the target comparison data, and timely adjusts the real-time training mode to ensure that the real-time training is always in the best state, improving the visual recovery efficiency and stability.

[0162] Specifically, when the periodic training mode generation unit generates a periodic training mode based on the second target analysis data, it compares the second target analysis data with the preset analysis data in the periodic special training mode database, where:

[0163] When there is a preset recovery degree in the periodic special training mode database that is consistent with the second target analysis data, the preset periodic training mode corresponding to the preset analysis data is output as the periodic training mode;

[0164] When there is no preset recovery degree in the periodic special training mode database that is consistent with the second target analysis data, the second target analysis data is pushed to the special user terminal to obtain the periodic training mode given by the special user terminal.

[0165] Specifically, the periodic special training mode database refers to a database system that stores and integrates periodic vision training knowledge, periodic vision training experience, and periodic vision training research results, including preset analysis data and periodic training modes. In this embodiment, the setting method of the periodic special training mode database is not limited. Those skilled in the art can freely set it according to the actual situation, as long as the output requirements of the periodic training mode are met. For example, it can be set to set the periodic special training mode database through big data and update the periodic special training mode database in real time. The preset analysis data refers to the pre-set and stored cases of simulated real user eye physiological data. The periodic training mode refers to the training mode that is periodically planned and set based on the user's vision condition. The special user terminal refers to the expert user terminal with the ability to set analysis data. In this embodiment, the method for obtaining the periodic training mode given by the special user terminal is not limited. Relevant personnel in the field can freely set it according to the actual situation, as long as the acquisition requirements of the periodic training mode given by the special user terminal are met. For example, a acquisition window can be set to be pushed to the special user terminal, and the special user inputs in the acquisition window, and the content input by the special user in the acquisition window is obtained as the periodic training mode given by the special user terminal.

[0166] Specifically, by comparing the eye physiological data of the user with the periodic special training mode database to customize the training cycle plan, the training pertinence is improved. According to the eye physiological data collected at different stages during the vision recovery process, the training mode is dynamically adjusted periodically to adapt to the changing actual situation of the user during the vision recovery process, ensuring that the training always fits the user's needs and further improving the training effect.

[0167] Specifically, when the cycle adjustment unit is used to perform periodic analysis on the second target analysis data and optimize the cycle training mode according to the analysis results, an update cycle is set. At the end of each update cycle, the vision abnormality value G of the second target analysis data is obtained, and the vision abnormality value G of the second target analysis data is compared with the preset vision abnormality value G0. According to the comparison result, the vision abnormality degree of the second target analysis data is judged, and the cycle training mode is adjusted according to the judgment result, where:

[0168] When G ≤ G0, it is determined that the vision abnormality degree of the obtained second target analysis data is low, and the cycle training mode is not adjusted;

[0169] When G > G0, it is determined that the vision abnormality degree of the obtained second target analysis data is high, and the cycle training mode is adjusted;

[0170] The intraocular pressure value H of the second target analysis data is obtained, and the intraocular pressure value H of the second target analysis data is compared with the preset intraocular pressure value H0. According to the comparison result, the intraocular pressure abnormality degree of the second target analysis data is judged, and the vision abnormality value G of the second target analysis data is updated according to the judgment result, where:

[0171] When H ≤ H0, it is determined that the intraocular pressure abnormality value degree of the second target analysis data is low, and the vision abnormality value G of the second target analysis data is not updated;

[0172] When H > H0, it is determined that the intraocular pressure abnormality value degree of the second target analysis data is high, and the vision abnormality value G of the second target analysis data is updated. The update coefficient is set as A1, A1 = 1.4 - 0.4×e -0.8×(H-H0) , and the updated vision abnormality value of the second target analysis data is G1, G1 = G×A1;

[0173] The axial length abnormality value K of the second target analysis data is obtained, and the axial length abnormality value K of the second target analysis data is compared with the preset axial length abnormality value K0. According to the comparison result, the axial length abnormality degree of the second target analysis data is judged, and the vision abnormality value G of the second target analysis data is updated according to the judgment result, where:

[0174] When K ≤ K0, it is determined that the axial length abnormality degree of the second target analysis data is low, and the vision abnormality value G of the second target analysis data is not updated;

[0175] When K > K0, it is determined that the axial length abnormality degree of the second target analysis data is high, and the vision abnormality value G of the second target analysis data is updated. The update coefficient is set as A2, A2 = 1.59 - 0.51×e -0.7×(K-K0) , and the updated vision abnormality value of the second target analysis data is G2, G2 = G×A2;

[0176] Obtain the abnormal value L of the bio-signal of the eye muscles in the second target analysis data. Compare the abnormal value L of the bio-signal of the eye muscles in the second target analysis data with the preset abnormal value L0 of the bio-signal of the eye muscles. Judge the degree of abnormality of the bio-signal of the eye muscles based on the comparison result, and update the abnormal value G of the intraocular pressure in the second target analysis data according to the judgment result, where:

[0177] When L ≤ L0, it is determined that the degree of abnormality of the bio-signal of the eye muscles in the second target analysis data is low, and the abnormal value G of the visual acuity in the second target analysis data is not updated;

[0178] When L > L0, it is determined that the degree of abnormality of the bio-signal of the eye muscles in the second target analysis data is high, and the abnormal value G of the visual acuity in the second target analysis data is updated. Set the update coefficient as A3, A3 = 2.58 - 1.52×e -0.6×(L-L0) , and the updated abnormal value G3 of the visual acuity in the second target analysis data is G3 = G×A3;

[0179] Obtain the degree of visual acuity abnormality G1 at the end of the previous cycle of training, the degree of visual acuity abnormality G2 at the end of the current cycle of training, the preset adjustment coefficient k, and the preset visual acuity recovery coefficient as X. According to the formula Obtain the visual acuity recovery coefficient X. Compare the recovery degree X of the target cycle data with the preset recovery degree X0. Judge the visual acuity recovery degree based on the comparison result, and adjust the training intensity of the training mode according to the judgment result, where:

[0180] When X < X0, it is determined that the visual acuity recovery degree is low, and the training intensity of the training mode is strengthened until X = X0;

[0181] When X = X0, it is determined that the visual acuity recovery degree is normal, and the training intensity of the training mode is not adjusted;

[0182] When X > X0, it is determined that the visual acuity recovery degree is high, and the training intensity of the training mode is reduced until X = X0.

[0183] Specifically, the intraocular pressure value refers to the pressure value inside the eyeball. In this embodiment, the preset adjustment coefficient is not limited, and those skilled in the art can freely set it as long as it meets the calculation requirements for visual acuity recovery, such as conducting a special user-end evaluation according to the user's visual acuity condition.

[0184] Specifically, the cycle adjustment unit updates and adjusts the second target analysis data and the cycle training mode periodically and dynamically, so that the cycle training mode can closely fit the actual eye condition of the user, improving the pertinence and effectiveness of the training.

[0185] So far, the technical solution of the present invention has been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easily understood by those skilled in the art that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims

1. A schematic diagram of the structure of a vision restoration device, characterized in that: include: An eye mask trainer, the side of which close to the fixing belt is connected to the biosensor, the training sensor and the eye tracking sensor, and a display screen is provided inside the eye mask trainer for generating vision training images; A biosensor is installed on one side of the eye mask trainer close to the fixing belt to collect the eye physiological data of the user; A training sensor is installed on a side of the eye mask trainer close to the fixing belt and away from the biosensor, and is used to collect image training data of the user; A right connecting cable and a left connecting cable, one end of which is connected to the eye mask trainer, and the other end is connected to the intelligent module, and is used to transmit the user's image training data, the user's eye physiological data and the user's vision training process data; An adjusting buckle, which is installed on the fixing belt and is used to adjust the length of the fixing belt; A fixing buckle, which is installed at the connection between the smart module and the right connecting cable and the left connecting cable, and is used to fix the right connecting cable and the left connecting cable; An intelligent module, which is connected to the fixing belt, the right connecting cable and the left connecting cable, and is used for converting and processing the data and generating a real-time training mode and a periodic training mode; A fixing belt, one end of which is connected to the eye mask trainer, and the other end is connected to the smart module for wearing and fixing; The eye tracking sensor is installed on the side of the eye mask trainer away from the biological sensor and the training sensor, and is used to collect data on the user's vision training process.

2. The vision restoration device according to claim 1, characterized in that: The smart module comprises: an acquisition unit, used to acquire the user's image training data, the user's eye physiological data and the user's vision training process data transmitted by the right connecting cable and the left connecting cable, and use the user's image training data as the first target real-time data, the user's eye physiological data as the first target analysis data, and the user's vision training process data as the first target comparison data; A processing unit, used for converting the first target real-time data, the first target analysis data and the first target comparison data to obtain the second target analysis data and the second target real-time data and the target comparison data; A real-time training pattern generating unit, generating a real-time training pattern according to the second target analysis data; A real-time adjustment unit is used to compare and analyze the second target real-time data and the target comparison data, and adjust the real-time training mode according to the analysis result to obtain the real-time optimal training mode; A periodic training pattern generating unit, used for generating a periodic training pattern according to the second target analysis data; The periodic adjustment unit is used to perform periodic analysis on the second target analysis data, and optimize the periodic training mode according to the analysis result to obtain the periodic optimal training mode.

3. The vision restoration device according to claim 1, characterized in that: When the acquisition unit is used to acquire the user's image training data, the user's eye physiological data and the user's vision training process data transmitted by the right connecting cable and the left connecting cable, the data of the user's eye movement trajectory and the image quality index are acquired through the training sensor to obtain the first target real-time data, the pressure value inside the eyeball, the anterior-posterior diameter length of the eyeball and the bioelectric signals generated by the eye muscles during contraction and relaxation are acquired through the biosensor to obtain the first target analysis data, and the overlap between the user's eye movement trajectory and the preset eye movement trajectory is acquired through the eye tracking sensor to obtain the first target comparison data.

4. The vision restoration device according to claim 1, characterized in that: When the processing unit is used to convert the first target real-time data and the first target analysis data, a composite filtering conversion model is constructed according to the historical conversion data, the historical conversion data is divided into a 70% simulation training set, a 20% simulation verification set and a 10% simulation test set, the hidden_size in the recurrent neural network model parameters is set to 128, num_layers is set to 4 layers, the simulation training set is input into the recurrent neural network model after the parameter setting for training, and the simulation verification set is input into the trained recurrent neural network model, the parameters of the recurrent neural network model are optimized, and then the test set is input into the recurrent neural network model after the parameter optimization for testing, and the test accuracy R is output, and the test accuracy R is compared with the preset test accuracy R0, and the compliance of the recurrent neural network model after the parameter optimization is judged according to the comparison result, and the recurrent neural network model after the parameter optimization is output according to the judgment result, wherein: When R≥R0, it is determined that the recurrent neural network model after parameter optimization meets the standard, and the recurrent neural network model is output as a composite filtering conversion model; When R<R0, it is determined that the recurrent neural network model after parameter optimization does not meet the standard, the second historical conversion data is obtained, and the recurrent neural network model is trained according to the second historical conversion data until the comparison result of the test accuracy R of the recurrent neural network model and the preset test accuracy R0 is R≥R0, and the recurrent neural network model is output as a composite filtering conversion model; The first target real-time data, the first target analysis data and the first target comparison data are input into the composite filtering conversion model for filtering conversion to obtain the second target real-time data, the second target analysis data and the target comparison data.

5. The vision restoration device according to claim 1, characterized in that: When the real-time training mode generation unit generates the real-time training mode according to the second target analysis, the second target analysis data is compared with the preset analysis data in the real-time special training mode database, wherein: When there is preset analysis data consistent with the second target analysis data in the real-time special training mode database, the image movement distance, image contrast value and image brightness value corresponding to the preset analysis data are output as initial training parameter values; When there is no preset analysis data consistent with the second target analysis data in the real-time special training mode database, the second target analysis data is pushed to the special user terminal, and the initial training parameter value given by the special user terminal is obtained; The initial training parameter values ​​are input into the intelligent module to obtain a real-time training mode.

6. The vision restoration device according to claim 1, characterized in that: The real-time adjustment unit is used to compare and analyze the second target real-time data and the target comparison data, and when adjusting the real-time training mode according to the analysis result, obtain the second target real-time data training abnormal value P, compare the second target real-time data training abnormal value P with the preset training abnormal value P0, judge the abnormal degree of the second target real-time data training according to the comparison result, and adjust the real-time training mode according to the judgment result, wherein: When P≤P0, it is determined that the abnormality of the second target real-time data training is low, and the real-time training mode is not adjusted; When P>P0, it is determined that the abnormality of the second target real-time data training is high, and the real-time training mode is adjusted; According to the target comparison data, the overlap degree M between the eye movement trajectory in the second target real-time data and the preset eye movement trajectory is obtained, and the overlap degree M between the eye movement trajectory of the second target real-time data and the visual stimulation image trajectory is compared with the preset overlap degree M0. According to the comparison result, the compliance of the overlap degree M between the eye movement trajectory of the second target real-time data and the visual stimulation image trajectory is judged, and the training abnormal value P of the second target real-time data is updated according to the judgment result, wherein: When M≥M0, it is determined that the overlap degree between the eye movement trajectory of the second target real-time data and the preset eye movement trajectory is high, and the training abnormal value P of the second target real-time data is not updated; When M<M0, it is determined that the overlap between the eye movement trajectory of the second target real-time data and the preset eye movement trajectory is low, and the training abnormal value P of the second target real-time data is updated, and the update coefficient is set to A, A=1.6+0.6×e -0.7×(M0-M) , the updated second target real-time data training anomaly value is P1, P1 = P × A; Obtain an image quality index Z in the second target real-time data, compare the image quality Z in the second target real-time data with a preset quality index Z0, judge the image clarity in the second target real-time data according to the comparison result, and optimize the overlap degree M between the eye movement trajectory in the second target real-time data and the preset eye movement trajectory according to the judgment result, wherein: When Z≥Z0, the image clarity in the second target real-time data is determined to be normal, and the overlap degree M between the eye movement trajectory in the second target real-time data and the preset eye movement trajectory is not optimized; When Z<Z0, the image clarity in the second target real-time data is determined to be abnormal, and the overlap degree M between the eye movement trajectory in the second target real-time data and the preset eye movement trajectory is optimized, and the optimization coefficient is set to B, B=0.4+0.6×e -0.3×(Z0-Z) , the overlap between the eye movement trajectory in the optimized second target real-time data and the preset eye movement trajectory is M1, M1 = M×(1-B); The image moving distance S in a short period of time is obtained according to the target comparison data, the image moving distance S in the short period of time is compared with the preset moving distance S0, the image moving speed is judged according to the comparison result, and the image moving speed is adjusted according to the judgment result, wherein: When S=S0, the image moving speed is determined to be normal and the image moving speed is not adjusted; When S>S0, it is determined that the image moving speed is too fast, and the image moving speed is slowed down until S=S0; When S<S0, it is determined that the image moving speed is too slow, and the image moving speed is accelerated until S=S0; When the real-time adjustment unit adjusts the real-time training mode, it obtains the image contrast value D, compares the image contrast value D with the preset image contrast standard value D0, judges the image contrast value according to the comparison result, and adjusts the image contrast value according to the judgment result, wherein: When D=D0, the image contrast value is determined to be normal and the image contrast value is not adjusted; When D>D0, it is determined that the image contrast value is too high, and the image contrast value is reduced until D=D0; When D<D0, it is determined that the image contrast value is too low, and the image contrast value is increased until D=D0; When the real-time adjustment unit adjusts the real-time training mode, it obtains the image brightness value C, compares the image brightness value C with the preset image brightness standard value C0, judges the image brightness according to the comparison result, and adjusts the image brightness value according to the judgment result, wherein: When C=C0, the image brightness value is determined to be normal and the image brightness value is not adjusted; When C>C0, it is determined that the image brightness value is too high, and the image brightness value is reduced until C=C0; When C<C0, it is determined that the image brightness value is too low, and the image brightness value is increased until C=C0.

7. The vision restoration device according to claim 1, characterized in that: When the periodic training mode generation unit generates the periodic training mode according to the second target analysis data, the second target analysis data is compared with the preset analysis data in the periodic special training mode database, wherein: When there is a preset recovery degree consistent with the second target analysis data in the periodic special training mode database, the preset periodic training mode corresponding to the preset analysis data is output as the periodic training mode; When there is no preset recovery degree consistent with the second target analysis data in the periodic special training mode database, the second target analysis data is pushed to the special user terminal to obtain the periodic training mode given by the special user terminal.

8. The vision restoration device according to claim 1, characterized in that: The periodic adjustment unit is used to perform periodic analysis on the second target analysis data, and when optimizing the periodic training mode according to the analysis result, an update period is set, and at the end of each update period, the visual acuity abnormality value G of the second target analysis data is obtained, and the visual acuity abnormality value G of the second target analysis data is compared with the preset visual acuity abnormality value G0, and the visual acuity abnormality degree of the second target analysis data is judged according to the comparison result, and the periodic training mode is adjusted according to the judgment result, wherein: When G≤G0, it is determined that the degree of visual acuity abnormality in acquiring the second target analysis data is low, and the periodic training mode is not adjusted; When G>G0, it is determined that the degree of visual acuity abnormality in obtaining the second target analysis data is high, and the periodic training mode is adjusted; Obtain the intraocular pressure value H of the second target analysis data, compare the intraocular pressure value H of the second target analysis data with the preset intraocular pressure value H0, judge the degree of abnormality of the intraocular pressure of the second target analysis data according to the comparison result, and update the abnormal intraocular pressure value G of the second target analysis data according to the judgment result, wherein: When H≤H0, it is determined that the abnormal value of the intraocular pressure of the second target analysis data is low, and the abnormal value of the visual acuity of the second target analysis data G is not updated; When H>H0, it is determined that the abnormal value of the intraocular pressure of the second target analysis data is high, and the abnormal value of the visual acuity of the second target analysis data G is updated, and the update coefficient is set to A1, A1=1.4-0.4×e -0.8×(H-H0) , the updated abnormal visual acuity value of the second target analysis data is G1, G1 = G × A1; Obtain the axial length abnormal value K of the second target analysis data, compare the axial length abnormal value K of the second target analysis data with the preset axial length abnormal value K0, judge the abnormal degree of the axial length of the second target analysis data according to the comparison result, and update the intraocular pressure abnormal value G of the second target analysis data according to the judgment result, wherein: When K≤K0, it is determined that the abnormality of the axial length of the second target analysis data is low, and the visual acuity abnormality value G of the second target analysis data is not updated; When K>K0, it is determined that the abnormality of the axial length of the second target analysis data is high, and the abnormal visual acuity value G of the second target analysis data is updated, and the update coefficient is set to A2, A2=1.59-0.51×e -0.7×(K-K0) , the updated abnormal visual acuity value of the second target analysis data is G2, G2 = G × A2; Obtain the eye muscle biosignal abnormal value L of the second target analysis data, compare the eye muscle biosignal abnormal value L of the second target analysis data with the preset eye muscle biosignal abnormal value L0, judge the degree of abnormality of the eye muscle biosignal data according to the comparison result, and update the intraocular pressure abnormal value G of the second target analysis data according to the judgment result, wherein: When L≤L0, it is determined that the abnormality of the eye muscle biological signal of the second target analysis data is low, and the visual acuity abnormality value G of the second target analysis data is not updated; When L>L0, it is determined that the abnormality of the eye muscle biological signal of the second target analysis data is high, and the abnormal value of the vision of the second target analysis data G is updated, and the update coefficient is set to A3, A3=2.58-1.52×e -0.6×(L-L0) , the updated abnormal visual acuity value of the second target analysis data is G3, G3 = G × A3; Get the degree of abnormal vision G1 at the end of the previous training cycle, the degree of abnormal vision G2 at the end of the current training cycle, the preset adjustment coefficient k, the preset vision recovery coefficient X, according to the formula Obtain the vision recovery coefficient X, compare the recovery degree X of the target cycle data with the preset recovery degree X0, judge the vision recovery degree according to the comparison result, and adjust the training intensity of the training mode according to the judgment result, where: When X<X0, it is determined that the degree of visual recovery is low, and the training intensity of the training mode is increased until X=X0; When X=X0, it is determined that the degree of visual recovery is normal, and the training intensity of the training mode is not adjusted; When X>X0, it is determined that the degree of visual recovery is high, and the training intensity of the training mode is reduced until X=X0.

Images