A design method for progressive lenses and the resulting progressive lenses

Abstract

The present invention relates to the technical field of progressive lenses, and a design method for progressive lenses and the obtained progressive lenses, including: defining a visual demand coefficient to dynamically adjust the change rates of the main curvature and the average curvature; dynamically adjusting the width of the transition region according to the hyperopia and myopia demands of the user; dividing the transition region in the width direction into regions, and establishing a new weight distribution function for each region; establishing a satisfaction model for prediction; providing initial τ 平均曲率0 , τ 主曲率0 , W base , W base(x) and reference values of each demand coefficient, and adjusting the curvature control module, the region division module, and the multi-interval power distribution according to the calculation. The present invention establishes a feedback and production guidance mechanism, which can effectively solve the deficiencies of the existing progressive lens design methods, overcome the problem that the fixed design cannot meet the personalized needs, ensure that the diopter continuously changes in the far, middle, and near fields of vision, and provide a more comfortable visual experience for users.

Description

Technical Field

[0001] The present invention relates to the technical field of progressive lenses, and particularly to a design method of progressive lenses and the obtained progressive lenses. Background Art

[0002] A progressive lens is a lens that can make the diopter continuously change in the far, middle, and near visual fields. It is usually divided into four regions, including a far vision region, a near vision region, a transition region, and a peripheral region. Among them, the transition region connects the far vision region and the near vision region, achieving a smooth refractive power transition position. Through the smooth change of refractive power, users will not feel a sudden visual jump when looking far and near.

[0003] In order to design high-performance progressive lenses, in the existing design methods, mathematical models are usually used to minimize the optical errors of the lenses. The mathematical models consider the principal curvature difference and the average curvature of the lens surface, and affect the calculation results of the models by using different weight distribution functions. In the current publicly available technologies, the spline interpolation method is often used to design multiple groups of weight distribution functions. These weight functions correspond to different change speeds of the transition region and are used to connect different regions on the lens: the high-weight region and the low-weight region. This design method aims to study the influence of changing the width of the weight transition on the refractive power and astigmatism distribution of the lens.

[0004] In the above design method, the weight distribution functions are preset and remain unchanged in different regions of the lens. The weight distribution functions are used to control the optical performance of each region of the lens and adjust the optical design of these regions. For example, by adjusting the changes in the principal curvature difference and the average curvature, to meet the optical requirements of different regions; in the far vision region and the near vision region, the weights are often larger to ensure clear visual effects, while in the peripheral region, the weights are smaller to reduce peripheral astigmatism. These weight distribution functions are usually set in advance and will not be adjusted according to the needs of individual users or other external factors.

[0005] In order to achieve a smooth transition, although the existing progressive lens design methods have used the spline interpolation method to generate weight distribution functions, thus ensuring a smooth and controllable change in optical performance between different regions, due to the use of fixed weight distribution functions, especially in the design of the transition region, the width and the change speed of the weight are fixed and unchanged. Therefore, the personalized needs of each wearer cannot be fully met. Summary of the Invention

[0006] The present invention provides a design method of progressive lenses and the obtained progressive lenses, which can effectively solve the problems in the background art.

[0007] In order to achieve the above object, the technical solution adopted by the present invention is:

[0008] A design method for progressive lenses, comprising:

[0009] S1: Obtain visual demand data of different users and define a visual demand coefficient D u，

[0010] D u = f(D far , D near , D astig , D activity );

[0011] Among them, D far is the hyperopia demand coefficient, D near is the myopia demand coefficient, D astig is the astigmatism demand coefficient, P activity is the living habit demand coefficient;

[0012] S2: According to the visual demand coefficient D u , dynamically adjust the change speeds of the main curvature and the average curvature. The formula is as follows:

[0013] τ 平均曲率 = τ 平均曲率0 ·(1 + m1·(D u - D ref ));

[0014] τ 主曲率 = τ 主曲率0 ·(1 + m2·(D u - D ref ));

[0015] Among them, τ 平均曲率0 and τ 主曲率0 are the standard curvature change speeds of the average curvature and the main curvature respectively, m1 and m2 are the first adjustment coefficient and the second adjustment coefficient respectively, D ref is the visual demand coefficient reference value;

[0016] S3: According to the hyperopia and myopia demands of the user, dynamically adjust the width W (x,y) of the transition region. The formula is:

[0017] W (x,y) = W base ·(1 + k1·(D far - D ref,far ) + k2·(D near - D ref,near ));

[0018] Among them, W base is the standard width of the transition region, k1 and k2 are the first control coefficient and the second control coefficient respectively, D ref,far and Dref,near They are the reference values of the hyperopia and myopia demand coefficients respectively;

[0019] S4: Divide the transition region in the width direction, and establish a new weight distribution function W for each of the regions adjusted(x) , and the formula is:

[0020] W adjusted(X) = W base(X) ·(1 + a·exp(b·(D u - D ref )) + c·ln(1 + d·|D u - D ref |));

[0021] Among them, W base(x) is the basic weight distribution function of the corresponding region, representing the initial state in the progressive lens design; a, b, c, and d are the first control parameter, the second control parameter, the third control parameter, and the fourth control parameter respectively, which are used to adjust the influence degrees of the exponential and logarithmic parts and are specifically valued according to the corresponding region;

[0022] S5: Establish a satisfaction model, predict m1, m2, k1, k2 and a, b, c, d of each corresponding region according to the user feedback data, and calculate τ 平均曲率 , τ 主曲率 , W (x,y) and W of each region adjusted(x) ;

[0023] S6: Provide the initial τ 平均曲率0 , τ 主曲率0 , W base , W base(x) and the reference values of each demand coefficient for the lens processing equipment. Adjust the curvature control module according to the calculated m1, m2, τ 平均曲率 and τ 主曲率 , adjust the region division module according to the calculated k1, k2 and W (x,y) , and adjust the multi-interval power distribution by using a, b, c, d and W adjusted(x) of each corresponding region.

[0024] Furthermore, the D u is a weighting function, and the calculation formula is:

[0025] D u = Q far ·D far + W near ·D near + W astig ·D astig + W activity ·Dactivity

[0026] Among them, W far 、W near 、W astig 、W activity are the weight coefficients of each requirement respectively, and are dynamically adjusted according to the specific situation and living habits of the user.

[0027] Furthermore, the training data of the satisfaction model includes user feedback data, design parameter data, and requirement data;

[0028] The user feedback data is a feature combination obtained by calculating subjective feedback data and objective data through feature interaction technology; the design parameter data is a feature combination obtained by combining and transforming existing design parameters; the requirement data is a feature combination obtained by calculating different requirement features through feature interaction technology.

[0029] Furthermore, establishing a satisfaction model includes:

[0030] A1: Determine the model objective to meet the visual satisfaction of the user;

[0031] A2: Obtain training data and divide the training data into a training set, a validation set, and a test set;

[0032] A3: Construct a model structure for outputting predicted design parameter data by inputting the user feedback data and requirement data;

[0033] A4: Model training and optimization, using machine learning algorithms to train the model, using validation set data to validate the model, and using test set data for final testing.

[0034] Furthermore, the machine learning algorithm is one of linear regression, decision tree, random forest, and neural network.

[0035] Furthermore, establish a non-linear fusion function F combined , with the input being three feature combinations of user feedback data, design parameter data, and requirement data, and the output being the fused comprehensive feature vector, and the formula is as follows;

[0036] F combined = ReLU(W2·ReLU(W1[F feedbace ,F design ,F demand +b1)+b2)

[0037] Among them, F feedback is the feature combination of user feedback data, F design is the feature combination of design parameter data, F demandIt is a feature combination of demand data, W1 and W2 are weight matrices of a neural network, b1 and b2 are bias terms, and ReLU is a non-linear activation function.

[0038] Further, the satisfaction model includes:

[0039] An input layer that receives the comprehensive feature vector as input;

[0040] A feature extraction layer that extracts high-level features from the comprehensive feature vector;

[0041] An output layer that generates the final model prediction output.

[0042] Further, the same feature interaction technology is adopted for the user feedback data and the demand data.

[0043] Further, the feature interaction technology is a feature interaction technology based on a random forest.

[0044] A progressive lens is obtained by using the design method of the progressive lens as described above.

[0045] Through the technical solution of the present invention, the following technical effects can be achieved:

[0046] By obtaining personalized visual demand data, dynamically adjusting the curvature and weight distribution, and establishing a feedback and production guidance mechanism, the present invention can effectively solve the deficiencies of the existing progressive lens design method, overcome the problem that the fixed design cannot meet personalized needs, allow the dynamic adjustment of the optical performance of the lens according to the needs of each user, ensure that the diopter continuously changes in the far, middle, and near fields of view, and provide a more comfortable visual experience for users. Description of the Drawings

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments recorded in the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0048] Figure 1 It is a flowchart of the design method of the progressive lens;

[0049] Figure 2 It is a schematic diagram of the composition of the training data of the satisfaction model;

[0050] Figure 3 It is a flowchart of establishing the satisfaction model;

[0051] Figure 4 It is a framework diagram of the satisfaction model. Detailed implementation manners

[0052] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments.

[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this invention belongs. The terms used in the description of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0054] Embodiment 1

[0055] As shown in Figure 1 , a design method for progressive lenses includes:

[0056] S1: Obtain the visual demand data of different users, and define the visual demand coefficient D u ,

[0057] D u = f(D far , D near , D astig , D activity );

[0058] Among them, D far is the hyperopia demand coefficient, D near is the myopia demand coefficient, D astig is the astigmatism demand coefficient, P activity is the living habit demand coefficient; this step ensures that the starting point of the design scheme is based on targeted differential data, reflecting the specific visual needs and lifestyles of each user, enabling subsequent designs to be precisely adjusted for each wearer;

[0059] S2: According to the visual demand coefficient D u , dynamically adjust the change speeds of the main curvature and the average curvature. The formula is as follows:

[0060] τ 平均曲率 = τ 平均曲率0 ·(1 + m1·(D u - D ref ));

[0061] τ 主曲率 = τ 主曲率0 ·(1 + m2·(D u - D ref ));

[0062] Among them, τ 平均曲率0 and τ 主曲率0 are the standard curvature change rates of the mean curvature and the principal curvature respectively, m1 and m2 are the first adjustment coefficient and the second adjustment coefficient respectively, and D ref is the reference value of the visual demand coefficient; during implementation, m1 and m2 are used to adjust the amplitude and direction of the curvature change. This dynamic adjustment overcomes the defects of the fixed curvature design, allows for flexible adjustment of the curvature change on the lens surface according to individual needs, ensures a smooth transition of vision, and solves the different needs of different wearers for far, middle, and near fields during diopter changes, optimizing the optical performance of each region;

[0063] S3: Dynamically adjust the width W of the transition region according to the hyperopia and myopia needs of the user (x,y) , and the formula is:

[0064] W (x,y) =W base ·(1 + k1·(D far -D ref,far ) + k2·(D near -D ref,near ));

[0065] Among them, W base is the standard width of the transition region, k1 and k2 are the first control coefficient and the second control coefficient respectively, and D ref,far and D ref,near are the reference values of the hyperopia and myopia demand coefficients respectively; the dynamic adjustment of the transition region solves the problem of fixed width in the existing design. By adjusting the width of the transition region according to the differences in different needs, it ensures a smoother change in diopter within the transition region, reduces the visual jump, and improves visual comfort;

[0066] S4: Divide the transition region in the width direction into regions, and establish a new weight distribution function W adjusted(x) for each region, and the formula is:

[0067] W adjusted(X) =W base(X) ·(1 + a·exp(b·(D u -D ref )) + c·ln(1 + d·|D u -D ref |));

[0068] Among them, W base(x)is the basic weight distribution function of the corresponding area, representing the initial state in progressive lens design; a, b, c, and d are the first control parameter, the second control parameter, the third control parameter, and the fourth control parameter respectively, which are used to adjust the influence degrees of the exponential and logarithmic parts and are specifically valued according to the corresponding area; the new weight distribution function can dynamically adjust the distribution of diopter and astigmatism according to the different requirements of each area, making the optical performance of the lens more personalized and optimized;

[0069] S5: Establish a satisfaction model, predict m1, m2, k1, k2, and a, b, c, d of each corresponding area according to the user feedback data, and calculate τ according to the prediction results 平均曲率 、τ 主曲率 、W (x,y) and W of each area adjusted(x) ; This mechanism allows the design process to have the ability of self - adaptation and continuous improvement, ensuring that the design can be continuously optimized to meet the personalized needs and actual wearing effects of users, and solving the problem in the prior art that dynamic adjustment cannot be made according to external factors or individual feedback;

[0070] S6: For the lens processing equipment, provide the initial τ 平均曲率0 、τ 主曲率0 、W base 、W base(x) and the reference values of each demand coefficient. Adjust the curvature control module according to the calculated m1, m2, τ 平均曲率 and τ 主曲率 ; Adjust the area division module according to the calculated k1, k2, and W (x,y) , and adjust the multi - interval power distribution by using a, b, c, d and W of each corresponding area adjusted(x) . Directly apply the optimized parameters to the actual production process to ensure that the manufactured lens meets the personalized optical design requirements.

[0071] The visual demand coefficient D u is a parameter that comprehensively considers the different visual needs of users, including hyperopia, myopia, astigmatism, and living habits. These needs will directly affect the optical performance requirements of the lens. D far and D near determine the diopter adjustment of the lens in the cases of hyperopia and myopia, which need to be adapted by the changes in the principal curvature and the mean curvature; D astig determines the requirement for the diopter consistency of the lens in different directions, which needs to minimize astigmatism by adjusting the principal curvature and the mean curvature; D activityIt reflects the usage patterns of users in daily life, such as working at close range or watching from a long distance for a long time, which need to be considered in the curvature design. Combining all requirements helps to comprehensively optimize the optical performance of the lens, ensuring visual clarity and comfort for users in various usage scenarios. Step S1 is the starting point of the entire design process, providing a basis for the dynamic adjustment and optimization of parameters in subsequent steps; Step S2 provides the basic curvature data for adjusting the width of the transition area to ensure the rationality of area division and the smoothness of refractive power transition. The subsequent area division provides a clear area boundary for the establishment of the weight distribution function, ensuring that the design of the weight distribution function meets the user's needs.

[0072] By obtaining personalized visual demand data, dynamically adjusting the curvature and weight distribution, and establishing a feedback and production guidance mechanism, the present invention can effectively solve the deficiencies of existing progressive lens design methods, overcome the problem that fixed designs cannot meet personalized needs, allow for the dynamic adjustment of the optical performance of the lens according to the needs of each user, ensure continuous change of refractive power in the far, middle, and near fields of view, and provide a more comfortable visual experience for users.

[0073] As a preference of the above embodiment, D u is a weighting function, and the calculation formula is:

[0074] D u =W far ·D far +W near ·D near +W astig ·D astig +W activity ·D activity ;

[0075] Among them, W far , W near , W astig , W activity are the weight coefficients for each demand, which are dynamically adjusted according to the specific situation and living habits of the user. Specifically, W far In the case where distant visual clarity is more important to the user, such as driving, watching sports games or meetings, this coefficient will be given a higher weight; W near In the case where near visual clarity is more important to the user, such as reading, using electronic devices, working, this coefficient will be given a higher weight; W astig For users who require special astigmatism correction, this coefficient will be given a higher weight; W activity is dynamically adjusted according to the monitoring data of the user's living habits, including the wearing effect of the user and activity pattern data. For example, if the user reports more glare or visual fatigue when driving at night, the weight coefficient can be adjusted to enhance the performance of the lens in this specific activity.

[0076] In the above preferred method, after assigning weights, the calculated D u is more in line with the actual needs and lifestyle of users. By comprehensively considering the importance and priority of various visual needs in users' lives, it improves the visual experience and comfort, enhances the efficiency and interpretability of design optimization, and ensures the consistency and stability of the design.

[0077] As a preference of the above embodiment, as Figure 2 shown, the training data of the satisfaction model includes user feedback data, design parameter data, and requirement data; the user feedback data is a feature combination obtained by calculating subjective feedback data and objective data through feature interaction technology; the design parameter data is a feature combination obtained by performing feature combination and feature transformation on existing design parameters; the requirement data is a feature combination obtained by calculating different requirement features through feature interaction technology.

[0078] Among them, the subjective feedback in the user feedback data refers to the user's feelings and satisfaction with the visual effect after wearing the lens, such as the comfort and fatigue when looking far, near, and using it for a long time; the objective data refers to the accurate values such as visual clarity, astigmatism change, and diopter accuracy measured by equipment; the feature interaction technology for the two analyzes the mutual relationship between multiple features to generate new combined features to more accurately reflect the user's overall visual experience. For example, the subjective feedback data may have a strong correlation with specific objective data (such as diopter error), and new feature combinations can be constructed through feature interaction to optimize the prediction ability of the model.

[0079] The design parameter data refers to various parameters involved in the current lens design, such as the curvature of the lens surface, the weight distribution function, the width of the transition region, etc. After these parameters undergo feature combination and feature transformation, new features are generated to enable the model to better understand the relationship between design parameters and user feedback; by performing feature combination on different design parameters, the model can identify which combinations of parameters have the greatest impact on the visual effect, and the feature transformation technology can convert the design parameters into different representations, such as enhancing the expressive ability of the model through non-linear transformation to capture more complex relationships.

[0080] Demand data refers to the personalized demands of different users (such as hyperopia demand, myopia demand, astigmatism correction, living habits, etc.). These data generate new feature combinations through feature interaction technology to optimize the adaptability of the design. Demand data can come from users' lifestyle surveys, activity pattern monitoring, etc. Demand data interacts with other features (such as design parameters and feedback data) through feature interaction technology to generate new combined features. For example, a user may have a high demand for long-distance clarity and a strong myopia correction demand at the same time. Through interaction technology, the mutual influence between these demands can be discovered to form a more accurate demand feature combination.

[0081] As a preference of the above embodiment, as Figure 3 shown, establishing a satisfaction model includes:

[0082] A1: Determine the model objective to meet the visual satisfaction of users;

[0083] A2: Obtain training data and divide the training data into a training set, a validation set, and a test set;

[0084] A3: Construct a model structure for outputting predicted design parameter data by inputting users' feedback data and demand data;

[0085] A4: Model training and optimization, using machine learning algorithms to train the model, using the validation set data to validate the model, and using the test set data for final testing.

[0086] Determining the model objective is the basis of the entire model design, which clarifies the direction of modeling. By setting visual satisfaction as the optimization objective, the model will automatically adjust design parameters (such as the width of the transition area, the speed of curvature change, etc.) to improve the user's wearing experience. In this preferred solution, the core structure of the satisfaction model is a machine learning model. The input is users' feedback data and demand data, and the output is predicted design parameter data. This model needs to learn the relationship between a large number of features (such as hyperopia, myopia, astigmatism demands, etc.) and the users' satisfaction feedback.

[0087] Among them, the machine learning algorithm is one of linear regression, decision tree, random forest, and neural network; specifically, the complexity of the model depends on the complexity of the data and the feature interaction relationships to be captured. For example, simpler relationships can use linear regression, while complex and non-linear relationships can be modeled through deep neural networks.

[0088] As a preference of the above embodiment, establish a non-linear fusion function F combined , with the input being three feature combinations of users' feedback data, design parameter data, and demand data, and the output being the fused comprehensive feature vector. The formula is as follows;

[0089] F combined = ReLU(W2·ReLU(W1[F feedbace ,F design ,F demand +b1)+b2);

[0090] Among them, F feedback is the feature combination of the user feedback data, F design is the feature combination of the design parameter data, F demand is the feature combination of the requirement data, W1, W2 are the weight matrices of the neural network, b1, b2 are the bias terms, and ReLU is the non-linear activation function.

[0091] In the implementation process, using the non-linear fusion function of the multi-layer perceptron structure can capture the complex non-linear relationships between the input features. Specifically, through the combination of the weight matrices W1, W2 and the non-linear activation function ReLU in multiple fully connected layers, the model can learn the deep interaction relationships between the user feedback data, the design parameter data and the requirement data. In the actual scenario, the relationships between the user's visual feedback, the lens design parameters and the requirement features are usually not linear. Through this formula, the model can mine the complex patterns hidden in different feature combinations, thereby improving the prediction ability and accuracy; each fully connected layer composed of the weight matrices W1, W2 and the bias terms b1, b2 in the formula can linearly transform the input features and introduce non-linearity through the ReLU activation function, making the fused output features have stronger expression ability.

[0092] In the implementation process, by concatenating all the input features together and performing fusion through two layers of fully connected layers and non-linear activation functions, all the features can be efficiently processed and fused within a unified framework. This processing method reduces the feature engineering work that needs to be manually designed and hands over the data fusion process to the model for automatic learning, and can achieve effective feature fusion in less training time; the weight matrices W1, W2 and the bias terms b1, b2 are trainable parameters, and they will be continuously updated during the model training process, so that the model can automatically learn the relative importance of the input features.

[0093] By performing feature fusion before the input satisfaction model, the model structure can be simplified, the calculation efficiency and feature learning ability can be improved, the generalization ability and self-adaptability of the model can be enhanced, the dependence on feature engineering can be reduced, and the training stability can be improved. This fusion strategy can better integrate the feature information of different data sources and improve the prediction effect and optimization performance of the model. Specifically, as a specific structural form of the satisfaction model that can accept the fused feature vectors, such as Figure 4As shown in the figure, the satisfaction model includes: an input layer that receives a comprehensive feature vector as input; a feature extraction layer that extracts high-level features from the comprehensive feature vector; and an output layer that generates the final model prediction output.

[0094] During the implementation process, the input layer receives a single feature vector, reducing the complexity of multi-channel input. For the implementation of the feature extraction layer, 1 to 2 fully connected layers can be used to extract non-linear features. Finally, a fully connected layer can be used to output the optimized design parameters and the user satisfaction score.

[0095] As a preference of the above embodiment, the same feature interaction technology is adopted for the user feedback data and the requirement data. By using the same feature interaction technology to process the user feedback data and the requirement data, the consistency of feature interaction can be improved, the model structure can be simplified, the generalization ability and interpretability can be enhanced, and at the same time, the risk of overfitting is reduced. The model can automatically capture the complex interaction relationship between feedback and requirements, effectively process multi-dimensional and multi-type data, and improve the personalization and overall performance of progressive lens design.

[0096] Specifically, the feature interaction technology is the feature interaction technology based on random forest. Random forest is an ensemble model composed of multiple decision trees, which naturally has the ability to capture feature interactions. The splitting process of each tree is to divide the space layer by layer according to the data features, and this splitting will naturally form the interaction of different features. For example, the decision tree can first split according to the hyperopia requirement, and then further split according to the astigmatism requirement, thus forming the interaction feature of the two. Since the random forest is an ensemble model composed of multiple trees, it can capture more complex and subtle feature interactions through the diversity of the trees.

[0097] During the implementation process, for the user feedback data (subjective feedback and objective data) and the requirement data (such as hyperopia requirement, myopia requirement, living habits, etc.), there may be complex non-linear relationships between these features. Through the splitting mechanism of the random forest, the model can automatically process these non-linear relationships without manually setting interaction terms, reducing the complexity of manual intervention. At the same time, the random forest can better cope with data noise or the contingency of some feedback data, thereby improving the robustness and generalization ability of the model. In practical applications, the feedback data of users may have the characteristics of volatility or strong subjectivity. The random forest can smooth these noises through the way of ensemble learning, so that the model still performs well in different user scenarios.

[0098] Embodiment 2

[0099] A progressive lens is obtained by using the design method of the progressive lens as described in Embodiment 1. The technical effects achieved in this embodiment are as described in Embodiment 1 above, and will not be elaborated here.

[0100] The basic principles, main features and advantages of the present invention have been shown and described above. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and what is described in the above embodiments and the specification only illustrates the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A design method for progressive lenses, characterized in that, Including: S1: Obtain the visual requirement data of different users and define the visual requirement coefficient D u , D u = f(D far , D rear , D astig , D activity ); Among them, D far is the hyperopia demand coefficient, D rear is the myopia demand coefficient, D astig is the astigmatism demand coefficient, P activity is the living habit demand coefficient; S2: According to the visual demand coefficient D u , dynamically adjust the change speeds of the principal curvature and the mean curvature, and the formula is as follows: τ 平均曲率 = τ 平均曲率0 ·(1 + m1·(D u - D ref )); τ 主曲率 = τ 主曲率0 ·(1 + m2·(D u - D ref )); Among them, τ 平均曲率0 and T 主曲率0 are respectively the standard curvature change rates of the mean curvature and the principal curvature, m1 and m2 are respectively the first adjustment coefficient and the second adjustment coefficient, and D ref is the reference value of the visual demand coefficient; S3: Dynamically adjust the width W of the transition region according to the user's hyperopia and myopia needs (x，y) , and the formula is: W (x，y) = W base ·(1 + k1·(D far - D ref,far ) + k2·(D near - D ref,near )); Among them, W base is the standard width of the transition region, k1 and k2 are the first control coefficient and the second control coefficient respectively, D ref，far and D ref，near are the reference values of the hyperopia and myopia demand coefficients respectively; S4: Divide the transition region in the width direction and establish a new weight distribution function W for each of the regions adjusted(x) , and the formula is: W adjusted(X) = W base(X) ·(1 + a·exp(b·(D u - D ref )) + c·ln(1 + d·|D u - D ref |)); Among them, W base(x) is the basic weight distribution function of the corresponding area, representing the initial state in the progressive lens design; a, b, c, and d are the first control parameter, the second control parameter, the third control parameter, and the fourth control parameter respectively, which are used to adjust the influence degrees of the exponential and logarithmic parts, and are specifically valued according to the corresponding area; S5: Establish a satisfaction model, predict m1, m2, k1, k2 and a, b, c, d in their respective corresponding regions based on the feedback data from users, and calculate τ according to the prediction results 平均曲率 , τ 主曲率 , W (x，y) and the W of each region adjusted(x) ; S6: Provide an initial τ for the lens processing equipment 平均曲率0 , τ 主曲率0 , W base , W base(x) and the reference values of each demand coefficient. Adjust the curvature control module according to the calculated m1, m2, τ 平均曲率 and τ 主曲率 Adjust the region division module according to the calculated k1, k2 and W (x，y) and adjust the multi - interval power distribution by using a, b, c, d and W in each corresponding region. adjusted(x) Adjust the multi - interval power distribution.

2. The design method of the progressive lens according to claim 1, characterized in that, The said D u is a weighting function, and its calculation formula is: D u = W far ·D far + W nean ·D near + W astig ·D astig + W activity ·D activity ; Among them, W far , W near , W astig , W activity are the weight coefficients for each type of requirement, which are dynamically adjusted according to the specific situation and living habits of the user.

3. The design method of the progressive lens according to claim 1, characterized in that, The training data of the satisfaction model includes user feedback data, design parameter data, and requirement data; The user feedback data is a feature combination obtained by calculating subjective feedback data and objective data through feature interaction technology; the design parameter data is a feature combination obtained by combining and transforming existing design parameters; the requirement data is a feature combination obtained by calculating different requirement features through feature interaction technology.

4. The design method of the progressive lens according to claim 3, characterized in that Establishing a satisfaction model includes: A1: Determining the model objective to meet the visual satisfaction of users; A2: Obtaining training data and dividing the training data into a training set, a validation set, and a test set; A3: Constructing a model structure for outputting predicted design parameter data by inputting the user feedback data and requirement data; A4: Model training and optimization, training the model using a machine learning algorithm, validating the model using validation set data, and performing a final test using test set data.

5. The design method of the progressive lens according to claim 4, characterized in that, The machine learning algorithm is one of linear regression, decision tree, random forest, and neural network.

6. The design method of the progressive lens according to claim 3, characterized in that, Establish a non-linear fusion function F combined , with the input being three feature combinations of user feedback data, design parameter data, and requirement data, and the output being the fused comprehensive feature vector. The formula is as follows; F combined = ReLU(W2 · ReLU(W1[F feedbace ,F design ,F demand + b1)+ b2); Among them, F feedback is the feature combination of the user feedback data, F design is the feature combination of the design parameter data, F demand is the feature combination of the requirement data, W1 and W2 are the weight matrices of the neural network, b1 and b2 are the bias terms, and ReLU is the non-linear activation function.

7. The design method of the progressive lens according to claim 6, characterized in that, The satisfaction model includes: An input layer that receives the comprehensive feature vector as input; A feature extraction layer that extracts high-level features from the comprehensive feature vector; An output layer that generates the final model prediction output.

8. The design method of the progressive lens according to claim 3, characterized in that, The same feature interaction technology is adopted for the user feedback data and the requirement data.

9. The design method of the progressive lens according to claim 8, characterized in that The feature interaction technology is a feature interaction technology based on random forest.

10. A progressive lens, characterized in that, Obtained by using the progressive lens design method according to any one of claims 1 to 9.

Images