An intelligent matching method for private customized rimless glasses

Abstract

The application relates to the field of intelligent glasses, and discloses an intelligent matching method for rimless glasses customized privately, which comprises the following steps: feature extraction is performed on face information data of a glasses wearer and three-dimensional pictures of rimless glasses in a database; the positions of the key points of the face of the glasses wearer after feature extraction are preliminarily screened with the key points of a three-dimensional model of the rimless glasses; further matching analysis is performed on the glasses matching degree in combination with face feature parameters of the glasses wearer; the upper limit value of the budget of the glasses wearer and the glasses matching degree are collected to perform optimal recommendation; three-dimensional rendering technology is used to render the virtual effect of the glasses wearer wearing each type of glasses in real time; the adaptation and the fit of the recommended rimless glasses and the face features of the wearer are evaluated and adjusted; finally, the type parameters of the rimless glasses selected by the user are transmitted to a full-automatic edging machine connected with a virtual try-on device to perform self-starting edging processing operation, so that the satisfaction and loyalty of the user are improved.

Description

Technical Field

[0001] This invention relates to the field of smart glasses technology, and more specifically to a smart matching method for custom-made rimless glasses. Background Technology

[0002] Traditional rimless eyeglass matching relies primarily on experience and manual fitting. Due to the design characteristics of rimless eyeglasses, their fit is highly dependent on frame size, shape, and the wearer's facial features. This method is inefficient and struggles to achieve accurate matching. However, advancements in computer vision, artificial intelligence algorithms, and 3D modeling technologies have brought new opportunities to the eyewear industry. These technologies enable precise identification and measurement of consumers' facial features, providing more personalized eyeglass matching services. Furthermore, consumers can virtually try on eyeglasses online, selecting the most suitable rimless glasses based on system recommendations. This significantly enhances the shopping experience, reduces the number of fittings, and shortens the purchase time.

[0003] However, the above process still has the following drawbacks:

[0004] Firstly, although existing virtual glasses try-on systems provide users with a variety of 3D images of glasses styles to choose from, the actual size of the glasses represented by these images is fixed. Even if they can match and recommend based on the user's facial features, they cannot fully meet the user's personalized needs for the size and shape of the glasses.

[0005] Secondly, existing technologies offer more of a virtual try-on experience than true personal customization. Users can only choose from a limited range of options and cannot create a fully personalized design based on their preferences and facial features. Furthermore, there is a lack of technology that connects the virtual try-on device to a fully automatic edging machine to automatically start edging processing based on the wearer's facial features and the designed rimless glasses.

[0006] Thirdly, because the size, shape and material of the actual glasses differ from the simulation images, users may find that the actual effect does not match the simulation when wearing them. This may require users to try on and adjust them multiple times, increasing the complexity and time cost of purchasing. Summary of the Invention

[0007] In order to overcome the above-mentioned defects of the prior art, the present invention provides an intelligent matching method for custom-made rimless glasses to solve the problems existing in the background art.

[0008] This invention provides the following technical solution: a smart matching method for custom-made rimless glasses, comprising:

[0009] S1: By inputting hundreds of 3D images of different shapes of rimless glasses lenses, bridge styles, and temple styles into the database, the 3D images of rimless glasses entered into the database are then categorized and stored according to the shape of a person's face.

[0010] S2: The scanner of the simulated trial fitting device is used to accurately scan the face of the person applying glasses, thereby collecting facial information data of the person applying glasses;

[0011] S3: By extracting features from the facial information data of the eyeglass wearer and the 3D images of rimless glasses in the database, the key points of the eyeglass wearer's face and the key points of the 3D model of the rimless glasses are extracted. Then, the key points of the eyeglass wearer's face and the key points of the 3D model of the rimless glasses are preliminarily screened to obtain feature similarity. The parameters of the rimless glasses that match the facial features of the eyeglass wearer are selected by feature similarity and transmitted to S4.

[0012] S4: Perform further matching analysis based on the facial feature parameters of the person applying for glasses and the parameters of the frameless glasses that match the facial features of the person applying for glasses, calculate the glasses matching degree, and transmit the calculation results of all glasses matching degrees to S5;

[0013] S5: Calculate the optimal recommendation index by collecting the upper limit of the user's budget and the matching degree of the glasses, and rank and recommend the selected rimless glasses according to the optimal recommendation index, and transmit the recommendation results of the rimless glasses to S6;

[0014] S6: Using 3D rendering technology, the recommended frameless glasses parameters are combined with the wearer's facial parameters to render the virtual effect of the wearer wearing each pair of glasses in real time. An interactive interface is provided, through which the wearer can freely choose the lens shape, bridge, and temples of the frameless glasses according to their own preferences from the sorted frameless glasses.

[0015] S7: Calculate the fit evaluation coefficient based on the scores of the fit and fit items in the virtual try-on, and use the fit evaluation coefficient to evaluate the fit and fit of the recommended frameless glasses to the wearer's facial features. Then adjust the try-on parameters based on the fit evaluation results.

[0016] S8: By transmitting the frameless glasses model parameters confirmed by the user to the fully automatic edging machine connected to the virtual try-on device, when the fully automatic edging machine receives the frameless glasses model parameters, it performs a verification of the frameless glasses model parameters and calculates the verification error coefficient. It then checks whether the verification result is within the allowable error range by using the verification error coefficient. For the verification result that passes the verification, it automatically generates an execution command for lens edging and transmits the execution command to S9.

[0017] S9: Upon receiving the execution command, immediately start the lens edging process, while monitoring the processing status in real time until the lens processing is completed. Then, automatically stop the processing and provide feedback information indicating that the processing is complete.

[0018] Preferably, step S1 uses a high-precision 3D scanner to scan the lens shape, bridge style, and temple style of hundreds of rimless glasses with different shapes to generate high-precision 3D data. The scanned 3D data is then denoised, and the denoised 3D data is registered and repaired to form complete 3D images of different styles of rimless glasses. The 3D images of the rimless glasses lenses are then classified and stored according to the shape of the face, including round, square, rectangular, and oval.

[0019] Preferably, step S2 involves having the person wearing glasses stand in front of a selected high-precision 3D scanner while maintaining a natural expression and posture, and then starting to capture the person's facial information data, including facial contours, interpupillary distance, and the position of facial features.

[0020] Preferably, step S3 extracts facial key points from the collected facial information data features, including glasses, nose tip, corners of mouth, and cheekbones, uses the SIFT algorithm to detect facial key points, and constructs a facial feature vector of the eyeglass wearer based on the location of the facial key points as F = [(x1, y1, z1), (x2, y2, z2), (x3, y3, z3), ..., (x n y n , z n This study extracts key points of a 3D model of rimless glasses from 3D images in a database, including the frame edges, nose pads, and temple connection points. The SIFT algorithm is used to detect these key points, and a 3D feature vector of the rimless glasses is constructed based on their locations: G = [(X1, Y1, Z1), (X2, Y2, Z2), (X3, Y3, Z3), ..., (X...]. m Y m Z m Then, a similarity metric is used to compare the facial feature vector of the person wearing glasses with the three-dimensional feature vector of the rimless glasses. The formula for calculating the feature similarity is as follows:

[0021]

[0022] Among them, (x i y i , z i (X) represents the coordinates of the i-th key point on the face of the person wearing glasses. j Y j Z j) represents the coordinates of the j-th keypoint in the 3D model of the rimless glasses, N represents the number of keypoints matching the wearer's facial keypoints with keypoints in the 3D model of the rimless glasses, and d represents the number of keypoints that match the wearer's facial keypoints. max This represents the maximum distance between all keypoint pairs, with the specific value determined by the maximum dimensions of the wearer's face and the rimless glasses model;

[0023] The feature similarity D(F, G) is compared with the preset feature similarity threshold τ. If the feature similarity D(F, G) ≤ the preset feature similarity threshold τ, it is determined that the rimless glasses have a high similarity to the facial features of the wearer, and the rimless glasses that meet the requirements are selected. The style and parameters of the selected rimless glasses are then transmitted to S4. If the feature similarity D(F, G) > the preset feature similarity threshold τ, it is determined that the rimless glasses have a low similarity to the facial features of the wearer, and the next rimless glasses will be selected.

[0024] Preferably, the frameless glasses parameter design values ​​g selected in S4 that match the wearer's facial features include frame width, frame height, nose pad width, and temple length; the actual facial feature parameters f of the wearer include face width, face height, nose bridge width, and cheekbone height; by further matching the selected frameless glasses parameters that match the wearer's facial features with the wearer's facial feature parameters, the formula for calculating the glasses matching degree is as follows:

[0025]

[0026] Among them, f p w represents the actual value of the p-th facial feature parameter of the eyeglass wearer. p g represents the weighting coefficient of the p-th facial feature parameter of the eyeglass wearer. p This represents the p-th parameter of the frameless glasses selected that match the facial features of the person wearing the glasses.

[0027] Preferably, the formula for calculating the optimal recommendation index in S5 is as follows:

[0028]

[0029] Where B represents the upper limit of the wearer's budget, P represents the price of rimless glasses, M represents the glasses fit, and α represents the weighting coefficient of the wearer's upper limit of budget.

[0030] A recommended list of rimless glasses is automatically generated based on the optimal recommendation index value, and the selected rimless glasses are sorted in the list from highest to lowest according to the optimal recommendation index value.

[0031] Preferably, step S6 allows the wearer to view and compare the scores of the virtual fitting and fit rating items for different rimless glasses, allows the wearer to adjust the position and angle of the rimless glasses by dragging and rotating, and allows the wearer to freely enlarge or reduce the size of the lens shape according to the size of their face until satisfied. At the same time, the virtual fitting process of the wearer is monitored in real time, and then the virtual fitting results are transmitted to step S7.

[0032] Preferably, the formula for calculating the adaptability evaluation coefficient in step S7 is as follows:

[0033]

[0034] Among them, s k h represents the score of the k-th fitness rating item. k t represents the weight of the k-th fitness score item. q v represents the score for the q-th fit rating item. q β1 represents the weight of the q-th fit rating item, β2 represents the total weight of fit, and β1 represents the total weight of fit.

[0035] The fit evaluation coefficient A is compared with the preset fit evaluation threshold θ. When the fit evaluation coefficient A ≥ the preset fit evaluation threshold θ, the fit and fit of the virtual trial are most suitable for the eyeglass wearer, and no adjustment is needed. When the fit evaluation coefficient A < the preset fit evaluation threshold θ, the fit and fit of the virtual trial need to be adjusted.

[0036] The size, shape, position of the bridge and temples, and width of the nose pads of the rimless glasses are adjusted using a fit assessment coefficient. The fit of the virtual try-on results is evaluated in real time until the fit and fit of the virtual try-on are most suitable for the wearer.

[0037] Preferably, step S8 connects the virtual try-on device to the fully automatic edging machine via an interface, and wirelessly transmits the frameless glasses style parameters selected and confirmed by the user on the virtual try-on device to the connected fully automatic edging machine. After receiving the parameters, the fully automatic edging machine calculates the verification error coefficient based on the selected frameless glasses style parameters using the following formula:

[0038] ΔK=|K-μ

[0039] Where K represents the actual check value and μ represents the standard check value;

[0040] The verification error coefficient ΔK is compared with the standard deviation coefficient v. If the verification error coefficient ΔK ≤ the standard deviation coefficient v, it indicates that the verification result is within the allowable error range, and the edge grinding verification is deemed to have passed. An execution command for lens edge grinding is automatically generated. If the verification error coefficient ΔK > the standard deviation coefficient v, it indicates that the verification result exceeds the allowable error range, and the edge grinding verification is deemed to have failed. A warning message is immediately issued and the edge grinding operation is refused.

[0041] Preferably, step S9 monitors the processing status during the lens edging process from the start of the fully automatic edging machine. When the fully automatic edging machine detects a processing completion signal, it automatically stops the edging operation and sends a completion message to the interactive interface.

[0042] The technical effects and advantages of this invention are as follows:

[0043] This invention involves inputting 3D images of hundreds of rimless eyeglasses with varying lens shapes, bridge styles, and temple designs into a database, categorized and stored according to facial features. A scanner in a simulated try-on device is used to precisely scan the wearer's face, collecting facial information data. Feature extraction is performed on the wearer's facial data and the 3D images of rimless eyeglasses in the database. The extracted key facial points are then initially matched with key points in the 3D models of the rimless eyeglasses. Further matching analysis is conducted based on the wearer's facial feature parameters and the parameters of the selected rimless eyeglasses that match their facial features, calculating the eyeglasses' fit degree. An optimal recommendation index is calculated by collecting the wearer's budget limit and the eyeglasses' fit degree. The selected rimless eyeglasses are then ranked and recommended using this optimal recommendation index. Finally, 3D rendering technology is used to combine the recommended rimless eyeglasses parameters with the wearer's facial parameters, rendering a virtual effect of the wearer wearing each pair of glasses in real time. The fit is then assessed based on a virtual try-on fitting score. The scores for the fit and fit rating items are used to calculate the fit evaluation coefficient. This coefficient is then used to assess and adjust the fit and fit of the recommended rimless glasses to the wearer's facial features. Finally, the parameters of the rimless glasses style selected by the user are transmitted to a fully automatic edging machine connected to the virtual try-on device for self-starting edging processing. This leverages the customizable shape of rimless glasses lenses, allowing users to create completely personalized designs based on their preferences and facial features. By connecting the virtual try-on device to the fully automatic edging machine, the edging process is automatically started based on the wearer's facial features and the designed rimless glasses. This allows for more fully automated and personalized designs based on the user's facial features and preferences. Because users can participate in the design process, they are more likely to achieve a satisfactory wearing effect, increasing user satisfaction and loyalty, and bringing better reputation and performance to the optical shop. Through the virtual try-on device, users can quickly see the effect of their design, greatly reducing the number of try-ons and time costs, and improving purchasing efficiency. Attached Figure Description

[0044] Figure 1 This is a flowchart of an intelligent matching method for custom-made rimless glasses according to the present invention. Detailed Implementation

[0045] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. In addition, the forms of the various structures described in the following embodiments are merely illustrative. The intelligent matching method for custom-made frameless glasses involved in the present invention is not limited to the structures described in the following embodiments. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] like Figure 1 The embodiment shown provides a smart matching method for custom-made rimless glasses, including:

[0047] S1: By inputting hundreds of 3D images of different rimless eyeglass lens shapes, bridge styles, and temple styles into the database, the 3D images of rimless eyeglasses in the database are then categorized and stored according to the shape of a person's face.

[0048] In this embodiment, step S1 uses a high-precision 3D scanner to scan the lens shape, bridge style, and temple style of hundreds of rimless glasses with different shapes, generating high-precision 3D data. The scanned 3D data is then denoised, and the denoised 3D data is registered and repaired to form complete 3D images of different styles of rimless glasses. The 3D images of the rimless glasses lenses are then classified and stored according to the shape of the face, including round, square, rectangular, and oval.

[0049] S2: A scanner using a simulated try-on device is used to precisely scan the face of the person applying glasses, thereby collecting facial information data.

[0050] In this embodiment, step S2 begins to capture facial information data of the eyeglasses wearer, including facial contours, interpupillary distance, and the position of facial features, by having the wearer stand in front of the selected high-precision 3D scanner and maintain a natural expression and posture.

[0051] S3: By extracting features from the facial information data of the eyeglass wearer and the 3D images of rimless glasses in the database, the key points of the eyeglass wearer's face and the key points of the 3D model of the rimless glasses are extracted. Then, the key points of the eyeglass wearer's face and the key points of the 3D model of the rimless glasses are preliminarily screened to obtain feature similarity. The parameters of the rimless glasses that match the facial features of the eyeglass wearer are selected by feature similarity and transmitted to S4.

[0052] In this embodiment, step S3 extracts facial key points from the collected facial information data features, including glasses, nose tip, corners of mouth, and cheekbones. The SIFT algorithm is used to detect these facial key points, and a facial feature vector for the eyeglass wearer is constructed based on the positions of these key points as F = [(x1, y1, z1), (x2, y2, z2), (x3, y3, z3), ..., (x...]. n y n , z nThis study extracts key points of a 3D model of rimless glasses from 3D images in a database, including the frame edges, nose pads, and temple connection points. The SIFT algorithm is used to detect these key points, and a 3D feature vector of the rimless glasses is constructed based on their locations: G = [(X1, Y1, Z1), (X2, Y2, Z2), (X3, Y3, Z3), ..., (X...]. m Y m Z m Then, a similarity metric is used to compare the facial feature vector of the person wearing glasses with the three-dimensional feature vector of the rimless glasses. The formula for calculating the feature similarity is as follows:

[0053]

[0054] Among them, (x i y i , z i (X) represents the coordinates of the i-th key point on the face of the person wearing glasses. j Y j Z j ) represents the coordinates of the j-th keypoint in the 3D model of the rimless glasses, N represents the number of keypoints matching the wearer's facial keypoints with keypoints in the 3D model of the rimless glasses, and d represents the number of keypoints that match the wearer's facial keypoints. max This represents the maximum distance between all keypoint pairs, with the specific value determined by the maximum dimensions of the wearer's face and the rimless glasses model;

[0055] The feature similarity D(F, G) is compared with the preset feature similarity threshold τ. If the feature similarity D(F, G) ≤ the preset feature similarity threshold τ, it is determined that the rimless glasses have a high similarity to the facial features of the wearer, and the rimless glasses that meet the requirements are selected. The style and parameters of the selected rimless glasses are then transmitted to S4. If the feature similarity D(F, G) > the preset feature similarity threshold τ, it is determined that the rimless glasses have a low similarity to the facial features of the wearer, and the next rimless glasses will be selected.

[0056] It should be noted that if the number of key points in the facial feature vector F of the person wearing glasses and the three-dimensional feature vector G of the glasses are different, then key point interpolation processing needs to be performed on the facial feature vector and the three-dimensional feature vector first to ensure that the same number of key points can be compared before feature similarity calculation.

[0057] The steps for keypoint lookup include:

[0058] Step 1: If there are more key points in F than in G, then determine the number of new key points that need to be inserted.

[0059] Step 2: For each new point to be inserted, find the two closest key points in G, and use linear interpolation to calculate the coordinates of the new point:

[0060] X new = (1-b)×X j1 +b×X j2

[0061] Y new = (1-b)×Y j1 +b×Y j2

[0062] Z new = (1-b)×Z j1 +b×Z j2

[0063] Among them, X new Y new Z new The coordinates of the newly inserted point, X j1 Y j1 Z j1 X represents the coordinates of the first keypoint in G that is closest to the new keypoint. j2 Y j2 Z j2 Let represent the coordinates of the second closest keypoint in G to the new keypoint, and b represent the interpolation coefficient, i.e., the position of the new keypoint relative to the two known keypoints, where 0 ≤ b ≤ 1. When b = 0, the coordinates of the new keypoint are interpolated with (X... j1 Y j1 Z j1 The coordinates of the new keypoint are exactly the same as those of (X). When b = 1, the coordinates of the new keypoint are the same as those of (X). j2 Y j2 Z j2 The coordinates of the new keypoint are exactly the same, when 0 < b < 1, the coordinates of the new keypoint are located at (X). j1 Y j1 Z j1 ) and (X j2 Y j2 Z j2 Any position between ).

[0064] S4: Perform further matching analysis based on the facial feature parameters of the person applying for glasses and the parameters of the frameless glasses that match the facial features of the person applying for glasses, calculate the glasses matching degree, and transmit the calculation results of all glasses matching degrees to S5.

[0065] In this embodiment, the frameless glasses parameter design values ​​g selected by S4 that match the facial features of the wearer include frame width, frame height, nose pad width, and temple length; the actual facial feature parameters f of the wearer include face width, face height, nose bridge width, and cheekbone height; by further matching the selected frameless glasses parameters that match the wearer's facial features with the wearer's facial feature parameters, the formula for calculating the glasses matching degree is as follows:

[0066]

[0067] Among them, f p w represents the actual value of the p-th facial feature parameter of the eyeglass wearer. p g represents the weighting coefficient of the p-th facial feature parameter of the eyeglass wearer. p This represents the p-th parameter of the frameless glasses selected that match the facial features of the person wearing the glasses.

[0068] S5: Calculate the optimal recommendation index by collecting the upper limit of the user's budget and the matching degree of the glasses, and rank and recommend the selected rimless glasses according to the optimal recommendation index, and transmit the recommendation results of the rimless glasses to S6.

[0069] In this embodiment, the formula for calculating the optimal recommendation index in step S5 is as follows:

[0070]

[0071] Where B represents the upper limit of the wearer's budget, P represents the price of rimless glasses, M represents the glasses fit, and α represents the weighting coefficient of the wearer's upper limit of budget.

[0072] A recommended list of rimless glasses is automatically generated based on the optimal recommendation index value, and the selected rimless glasses are sorted in the list from highest to lowest according to the optimal recommendation index value.

[0073] S6: Using 3D rendering technology, the recommended frameless glasses parameters are combined with the wearer's facial parameters to render a virtual effect of the wearer wearing each pair of glasses in real time. An interactive interface is provided, through which the wearer can freely choose the lens shape, bridge, and temples of the frameless glasses according to their own preferences from the sorted rimless glasses.

[0074] In this embodiment, step S6 allows the eyeglass wearer to view and compare the scores of the virtual fitting and fit rating items for different rimless glasses. The eyeglass wearer is allowed to adjust the position and angle of the rimless glasses by dragging and rotating them, and to freely enlarge or reduce the size of the lens shape according to the size of their face until they are satisfied. At the same time, the virtual fitting process of the eyeglass wearer is monitored in real time, and then the virtual fitting results are transmitted to step S7.

[0075] S7: Calculate the fit evaluation coefficient based on the scores of the fit and fit items in the virtual try-on. Then, use the fit evaluation coefficient to assess the fit and fit of the recommended rimless glasses to the wearer's facial features. Finally, adjust the try-on parameters based on the fit evaluation results.

[0076] In this embodiment, the formula for calculating the fit evaluation coefficient in step S7 is as follows:

[0077]

[0078] Among them, s k h represents the score of the k-th fitness rating item. k t represents the weight of the k-th fitness score item. q v represents the score for the q-th fit rating item. q β1 represents the weight of the q-th fit rating item, β2 represents the total weight of fit, and β1 represents the total weight of fit.

[0079] The fit evaluation coefficient A is compared with the preset fit evaluation threshold θ. When the fit evaluation coefficient A ≥ the preset fit evaluation threshold θ, the fit and fit of the virtual trial are most suitable for the eyeglass wearer, and no adjustment is needed. When the fit evaluation coefficient A < the preset fit evaluation threshold θ, the fit and fit of the virtual trial need to be adjusted.

[0080] The size, shape, position of the bridge and temples, and width of the nose pads of the rimless glasses are adjusted using a fit assessment coefficient. The fit of the virtual try-on results is evaluated in real time until the fit and fit of the virtual try-on are most suitable for the wearer.

[0081] It should be noted that the scoring criteria and score ranges for the fit and conformity rating items are as follows:

[0082] Compatibility rating criteria:

[0083] a. Face shape compatibility (0-10 points)

[0084] b. Nasal bridge compatibility (0-10 points)

[0085] c. Ear position fit (0-10 points)

[0086] d. Eye spacing fit (0-10 points);

[0087] Fit rating criteria:

[0088] a. Fit of the frames to the face (0-10 points)

[0089] b. How well the temples fit the ears (0-10 points)

[0090] c. Fit of the nose pads to the bridge of the nose (0-10 points).

[0091] S8: By transmitting the frameless glasses model parameters confirmed by the user to the fully automatic edging machine connected to the virtual try-on device, the fully automatic edging machine performs a verification of the frameless glasses model parameters after receiving them, calculates the verification error coefficient, checks whether the verification result is within the allowable error range by using the verification error coefficient, automatically generates an execution command for lens edging for the verification result that passes the verification, and transmits the execution command to S9.

[0092] In this embodiment, step S8 connects the virtual try-on device to the fully automatic edging machine via an interface, and wirelessly transmits the frameless glasses model parameters selected and confirmed by the user on the virtual try-on device to the connected fully automatic edging machine. After receiving the parameters, the fully automatic edging machine calculates the verification error coefficient based on the selected frameless glasses model parameters using the following formula:

[0093] ΔK=|K-μ

[0094] Where K represents the actual check value and μ represents the standard check value;

[0095] The verification error coefficient ΔK is compared with the standard deviation coefficient v. If the verification error coefficient ΔK ≤ the standard deviation coefficient v, it indicates that the verification result is within the allowable error range, and the edge grinding verification is deemed to have passed. An execution command for lens edge grinding is automatically generated. If the verification error coefficient ΔK > the standard deviation coefficient v, it indicates that the verification result exceeds the allowable error range, and the edge grinding verification is deemed to have failed. A warning message is immediately issued and the edge grinding operation is refused.

[0096] It should be noted that the actual formula for calculating the check value K is as follows:

[0097] K=h1×L1+h2×L2+h3×(L3+L4)+h4×C

[0098] Where L1 represents the lens diameter, L2 represents the lens center thickness, L3 represents the temple length, L4 represents the nose pad width, C represents the lens hardness, and h1, h2, h3, and h4 are weighting coefficients.

[0099] S9: Upon receiving the execution command, immediately start the lens edging process, while monitoring the processing status in real time until the lens processing is completed. Then, automatically stop the processing and provide feedback information indicating that the processing is complete.

[0100] In this embodiment, S9 starts monitoring the processing status during the lens edging process when the fully automatic edging machine starts the edging process. When the fully automatic edging machine detects the processing completion signal, it automatically stops the edging operation and sends a prompt message to the interactive interface indicating that the processing is complete.

[0101] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0102] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A smart matching method for private customization of rimless eyeglasses, characterized in that: The method comprises the following steps: S1: by entering hundreds of three-dimensional pictures of frameless glasses lens shapes, frameless glasses bridge styles and frameless glasses leg styles into a database, and then storing the frameless glasses three-dimensional pictures in the database according to human face shapes; S2: using a scanner of a simulation fitting device to accurately scan the face of the lens wearer, so as to collect the face information data of the lens wearer; S3: by extracting the features of the face information data of the lens wearer and the frameless glasses three-dimensional pictures in the database, the face key point positions of the lens wearer and the three-dimensional model key points of the frameless glasses are extracted, and then the face key point positions of the lens wearer and the three-dimensional model key points of the frameless glasses are preliminarily screened, the feature similarity is obtained, and the frameless glasses parameters conforming to the face features of the lens wearer are screened out through the feature similarity and transmitted to S4; S4: according to the face feature parameters of the lens wearer and the frameless glasses parameters conforming to the face features of the lens wearer, further matching analysis is carried out, the glasses matching degree is calculated, and the calculation results of all glasses matching degrees are transmitted to S5; S5: by collecting the upper limit value of the budget of the lens wearer and the glasses matching degree, the optimal recommendation index is calculated, the frameless glasses screened out are sorted and recommended through the optimal recommendation index, and the recommendation results of the frameless glasses are transmitted to S6; S6: using three-dimensional rendering technology, the recommended frameless glasses parameters are combined with the face parameters of the lens wearer, the virtual effect of the lens wearer wearing each pair of glasses is rendered in real time, and an interactive interface is provided, through which the lens wearer can randomly select the lens shape, bridge and leg of the frameless glasses according to his own preferences; S7: according to the scores of the virtual fitting adaptation degree score item and the fit degree score item, the adaptation evaluation coefficient is calculated, and the adaptation degree and the fit degree of the recommended frameless glasses and the face features of the wearer are evaluated by the adaptation evaluation coefficient, and then the fitting parameters are adjusted according to the evaluation results of the adaptation; S8: by transmitting the frameless glasses model parameters selected by the user to the full-automatic edging machine connected with the virtual fitting device, when the full-automatic edging machine receives the frameless glasses model parameters, the verification of the frameless glasses model parameters is performed, the verification error coefficient is calculated, the verification result is detected through the verification error coefficient whether it is within the allowable error range, the execution instruction of lens edging is automatically generated for the detection result passing the verification, and the execution instruction is transmitted to S9; S9: according to the received execution instruction, the lens edging process is immediately started, and the processing state in the edging process is monitored in real time until the lens processing is completed, and then the processing is automatically stopped and the feedback information of processing completion is prompted.

2. A smart matching method for private customization of rimless eyeglasses as claimed in claim 1, wherein: The S1 scans the lens shapes of hundreds of frameless glasses with different shapes, the bridge styles of the frameless glasses and the temple styles of the frameless glasses by using a high-precision three-dimensional scanner, generates high-precision three-dimensional data, performs denoising processing on the three-dimensional data obtained by scanning, performs data registration and repair on the three-dimensional data after denoising processing, forms complete three-dimensional pictures of different types of frameless glasses, and classifies and stores the three-dimensional pictures of the lens of the frameless glasses according to the face shapes including a circle, a square, a rectangle and an ellipse.

3. The intelligent matching method for private customization of rimless eyeglasses as claimed in claim 1, wherein: The S2 starts to capture the facial information data of the lens wearer, including the facial contour, the interpupillary distance and the position of the five facial features, when the lens wearer stands in front of the selected high-precision three-dimensional scanner and keeps a natural expression and posture.

4. The intelligent matching method for private customization of rimless eyeglasses as claimed in claim 1, wherein: The S3 extracts facial key points including glasses, nose tip, mouth corner and cheekbone from the collected facial information data features, detects the facial key points using the SIFT algorithm, and constructs a frame wearer facial feature vector F = [(x1, y1, z1), (x2, y2, z2), (x3, y3, z3), …, (x n , y n , z n )] according to the facial key point positions. The S3 extracts rimless glasses three-dimensional model key points including frame edge, nose pad and temple connecting point from the rimless glasses three-dimensional picture features in the database, detects the rimless glasses three-dimensional model key points using the SIFT algorithm, and constructs a rimless glasses three-dimensional feature vector G = [(X1, Y1, Z1), (X2, Y2, Z2), (X3, Y3, Z3), …, (X m , Y m , Z m )] according to the rimless glasses three-dimensional model key point positions. The S3 then uses a similarity measure to compare the frame wearer facial feature vector and the rimless glasses three-dimensional feature vector, and calculates a feature similarity formula as follows: wherein (x i , y i , z i ) represents the coordinates of the i-th key point of the face of the wearer, (X j , Y j , Z j ) represents the coordinates of the corresponding j-th key point in the three-dimensional model of rimless eyeglasses, N represents the number of matching between the key points of the face of the wearer and the key points of the three-dimensional model of rimless eyeglasses, d max represents the maximum distance between all key point pairs, the specific value being determined by the maximum size of the face of the wearer and the model of rimless eyeglasses; The feature similarity D(F, G) is compared with a preset feature similarity threshold τ, if the feature similarity D(F, G) ≤ the preset feature similarity threshold τ, it is determined that the frameless glasses have a high similarity with the facial features of the lens wearer, the frameless glasses that meet the requirement at this time are screened out, and the frameless glasses style and parameters screened out are transmitted to the S4, if the feature similarity D(F, G) > the preset feature similarity threshold τ, it is determined that the frameless glasses have a low similarity with the facial features of the lens wearer, and the next frameless glasses will be screened out.

5. The intelligent matching method for private customization of rimless eyewear according to claim 1, characterized in that: The S4 screens out the frameless glasses parameters design value g that is consistent with the facial features of the lens wearer, including the frame width, the frame height, the nose pad width and the temple length; the lens wearer's facial feature parameter actual value f, including the face width, the face height, the nose bridge width and the cheekbone height; the frameless glasses parameters that are consistent with the facial features of the lens wearer are further matched with the lens wearer's facial feature parameters, and the formula for calculating the glasses matching degree is as follows: wherein f p represents the actual value of the pth facial feature parameter of the lens wearer, w p represents the weight coefficient of the pth facial feature parameter of the lens wearer, g p represents the pth eyewear parameter of the rimless eyewear that is filtered out to be consistent with the facial features of the lens wearer.

6. The intelligent matching method for private customization of rimless eyewear according to claim 1, wherein: The formula for calculating the optimal recommendation index is as follows: Wherein, B represents the upper limit of the budget of the lens wearer, P represents the price of the frameless glasses, M represents the glasses matching degree, and a represents the weight coefficient of the upper limit of the budget of the lens wearer. An optimal recommendation index value is automatically generated to generate a frameless glasses recommendation list, and the frameless glasses screened out are sorted in the list from high to low according to the size of the optimal recommendation index value.

7. The intelligent matching method for private customization of rimless eyewear according to claim 1, wherein: The S6 allows the lens wearer to view and compare the virtual try-on fitting degree score and the fit degree score of different frameless glasses, allows the lens wearer to adjust the position and angle of the frameless glasses by dragging and rotating, and allows the lens wearer to arbitrarily enlarge or reduce the size of the lens shape according to the size of the face, until the lens wearer is satisfied, and at the same time, the virtual try-on process of the lens wearer is monitored in real time, and the virtual try-on result is transmitted to the S7.

8. The intelligent matching method for private customization of rimless eyewear according to claim 1, wherein: The formula for calculating the adaptability evaluation coefficient is as follows: wherein s k represents the score of the kth degree of adaptation score item, h k represents the weight of the kth degree of adaptation score item, t q represents the score of the qth degree of fit score item, v q represents the weight of the qth degree of fit score item, β1 represents the total weight of the degree of adaptation, and β2 represents the total weight of the degree of fit. The adaptability evaluation coefficient A is compared with a preset adaptability evaluation threshold θ, when the adaptability evaluation coefficient A ≥ the preset adaptability evaluation threshold θ, the virtual try-on fitting degree and the fit degree are most suitable for the lens wearer, and there is no need to adjust the fitting degree and the fit degree, when the adaptability evaluation coefficient A < the preset adaptability evaluation threshold θ, the virtual try-on fitting degree and the fit degree need to be adjusted. The size, shape, size, position of the bridge and the leg, and the width of the nose pad of the rimless glasses are adjusted through the adaptability evaluation coefficient, and the virtual try-on result is evaluated in real time until the adaptability and fit of the virtual try-on are most suitable for the wearer.

9. The intelligent matching method for private customization of rimless eyewear according to claim 1, wherein: The S8 connects the virtual try-on device to the full-automatic edger through an interface, and transmits the rimless glasses model parameters selected and confirmed by the user on the virtual try-on device to the full-automatic edger connected thereto in a wireless manner. After receiving the parameters, the full-automatic edger calculates the verification error coefficient according to the selected rimless glasses model parameters, and the formula is as follows: ΔK = |K - μ| Wherein, K represents the actual verification value, and μ represents the standard verification value. The verification error coefficient ΔK is compared with the standard deviation coefficient v. If the verification error coefficient ΔK ≤ the standard deviation coefficient v, it is indicated that the verification result is within the allowable error range, the edging verification is determined to pass, and the execution instruction of lens edging is automatically generated. If the verification error coefficient ΔK > the standard deviation coefficient v, it is indicated that the verification result exceeds the allowable error range, the edging verification is determined to fail, and a warning information is immediately sent and the edging processing operation is refused.

10. The intelligent matching method for private customization of rimless eyewear according to claim 1, wherein: The S9 starts to monitor the processing state in the edging process when the full-automatic edger starts the edging process of the lens, and automatically stops the edging operation when the full-automatic edger monitors the processing completion signal, and sends a processing completion prompt information to the interactive interface.

Images