Intelligent AR display method and system

Abstract

The invention discloses an intelligent AR display method and system, and belongs to the technical field of intelligent furniture, and the technical scheme is characterized in that the method comprises the steps: receiving a fitting image of a user; extracting a first human body contour of the user according to the fitting image; obtaining user posture information according to the first human body contour; obtaining information of recommended clothes according to the fashion clothes and the posture information of the user; the information of the recommended clothes is sent to the intelligent AR fitting mirror, so that the intelligent AR fitting mirror displays the recommended clothes, the size data of all parts of the user can be accurately calculated through the human body contour of the user in the dressing state, the posture information of the user is obtained, then the clothes are recommended for the user based on the posture information of the user and the fashion wearing trend, and the user experience is improved. Meanwhile, the three-dimensional human body model of the user can be generated according to the posture information of the user, the fitting effect of the model is displayed in the intelligent AR fitting mirror, and the experience feeling of the user is enhanced.

Description

Technical Field

[0001] This invention relates to the field of smart furniture technology, and more specifically to a smart AR display method and system. Background Technology

[0002] As the economy develops, people are constantly improving their quality of life, and current shopping methods can no longer meet users' needs and experience. With the development of virtual reality and augmented reality technologies, many solutions have been developed to enable virtual try-on, allowing users to intuitively feel the effect of trying on clothes without actually wearing them, thus enhancing their shopping experience.

[0003] For example, Chinese patent CN109685911B provides an AR glasses system for virtual try-on and a method thereof. The AR glasses acquire images of clothing and use a modeling system to create 3D models of them. Then, the system acquires images of the human body and uses the modeling system to create 3D models of the human body. Finally, the 3D models are intelligently combined and presented to the user.

[0004] However, this existing technology requires users to wear AR glasses to see the fitting effect and requires inputting the user's human body image. It cannot perform virtual fitting directly online based on the user's current fitting image, so the existing technology has shortcomings. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the present invention aims to provide an intelligent AR display method and system that recommends clothing to users based on their body shape information and popular fashion trends. Simultaneously, it can generate a three-dimensional human body model of the user based on their body shape information and display the model's fitting effect in an intelligent AR fitting mirror.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] This invention provides an intelligent AR display method, executed by a server, the method comprising:

[0008] Receive the user's fitting room image;

[0009] Extract the user's first human body contour from the fitting image;

[0010] The user's body posture information is obtained based on the first human body contour;

[0011] Based on popular fashion trends and the user's body shape information, clothing recommendations are obtained.

[0012] The information of the recommended clothing items is sent to the smart AR fitting mirror so that the smart AR fitting mirror can display the recommended clothing items.

[0013] As a further improvement of the present invention, extracting the user's first human body contour based on the fitting image includes:

[0014] The fitting image is converted into an HSV color space image, and the saturation S channel in the HSV color space image is extracted.

[0015] Based on the multi-directional Sobel operator, the gradient images of the saturation S channel in four directions are obtained;

[0016] Based on the gradient images in the four directions and linear weighted fusion, an absolute gradient image is obtained;

[0017] The absolute gradient image is binarized to obtain the first human body contour.

[0018] As a further improvement of the present invention, obtaining user posture information based on the first human body contour includes:

[0019] The user's first data is obtained based on the first human body contour, and the first data is the size data of the user's exposed parts.

[0020] The size information of the clothing currently being tried on by the user is obtained from the fitting image;

[0021] Based on the first data and the size information, the size ratio between the user and the currently tried-on clothing is obtained;

[0022] The second data of the user is obtained based on the size ratio relationship, and the second data is the size data of the user's obscured part;

[0023] The user's body posture information is obtained based on the second data.

[0024] As a further improvement of the present invention, obtaining the user's body posture information based on the second data includes:

[0025] Based on the second data and the preset neural network model, multiple second human body contours of the user are obtained;

[0026] The user's third human body contour is calculated based on the plurality of second human body contours;

[0027] The user's body shape information is calculated based on the third human body contour, and the user's body shape information includes the user's shoulder width, chest circumference, waist circumference, and hip circumference.

[0028] As a further improvement of the present invention, the shoulder width ,in Indicates the user's actual height. This represents the height of the third human body silhouette. , representing the shoulder width of the third human body contour. and They are respectively and The derivative of, where , , and This represents the coordinates of the two endpoints of the shoulder of the third human body contour. The coordinates represent the intersection of the tangents at the two endpoints. .

[0029] As a further improvement of the present invention, the waist circumference ,in This indicates the length of the user's waist. and These represent the widths of the user's abdomen and lower abdomen, respectively. , representing the width of the user's waist.

[0030] As a further improvement of the present invention, the step of obtaining recommended clothing information based on popular fashion trends and the user's body shape information includes:

[0031] The style and design of the clothing being tried on are obtained based on the fitting images;

[0032] Based on the style and design of the currently tried-on clothing, the user's body shape information, and a preset database, information on recommended clothing is obtained.

[0033] As a further improvement of the present invention, the user's joint position is predicted based on the third human body contour and the convolutional neural network;

[0034] The joint positions are fitted to three-dimensional joint points to obtain the user's three-dimensional model.

[0035] This invention provides a smart AR display method, executed by a smart AR fitting mirror, the method comprising:

[0036] The smart AR fitting mirror captures images of the user trying on clothes using a camera, which is installed in the mirror.

[0037] The fitting image is sent to the server, so that the server can extract the user's first human body contour based on the fitting image, obtain the user's body shape information based on the first human body contour, and then obtain recommended clothing information based on popular clothing and the user's body shape information, and finally send the recommended clothing information to the smart AR fitting mirror.

[0038] Receive the information about the recommended clothing items and display the recommended clothing items.

[0039] This invention provides an intelligent AR display system, including a server and an intelligent AR fitting mirror;

[0040] The server includes:

[0041] The first receiving module is used to receive the user's fitting images;

[0042] Recommendation module: used to extract the user's first human body contour based on the fitting image, then obtain the user's body shape information based on the first human body contour, and finally obtain recommended clothing information based on popular outfits and the user's body shape information;

[0043] The sending module is used to send the information of the recommended clothing to the smart AR fitting mirror;

[0044] The intelligent AR fitting mirror includes:

[0045] A camera is used to capture images of users trying on clothes;

[0046] The communication module is used to send the fitting image to the server, so that the server can extract the user's first human body contour based on the fitting image, obtain the user's body shape information based on the first human body contour, and finally obtain recommended clothing information based on popular clothing and the user's body shape information.

[0047] The second receiving module is used to receive the recommended clothing information sent by the server;

[0048] A display screen is used to show the try-on effect of the recommended clothing.

[0049] This invention calculates the size data of various parts of the user's body by observing the user's body contours while dressed, thereby obtaining the user's body posture information. Based on the user's body posture information and popular clothing trends, it recommends clothing to the user. At the same time, it can generate a three-dimensional human body model of the user based on the user's body posture information and display the model's fitting effect in a smart AR fitting mirror, thereby enhancing the user's experience. Attached Figure Description

[0050] Figure 1 This is a schematic diagram of the method steps of the present invention;

[0051] Figure 2 This is a schematic diagram of the first and third human body contours in this invention;

[0052] Figure 3 for Figure 2 A diagram of a hand;

[0053] Figure 4 for Figure 2 A diagram of the waist;

[0054] Figure 5 for Figure 2 A diagram of the legs;

[0055] Figure 6 This is a schematic diagram of the first human body outline, the second human body outline, and the third human body outline in this invention;

[0056] Figure 7 A schematic diagram of the shoulder of a third human figure;

[0057] Figure 8 A schematic diagram of the waist of a third human figure;

[0058] Figure 9 This is a schematic diagram of the system structure of the present invention.

[0059] Explanation of reference numerals in the attached drawings: 1. First human body outline; 2. Second human body outline; 3. Third human body outline; 4. Display screen; 5. Camera; 6. Communication module; 7. Second receiving module. Detailed Implementation

[0060] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof.

[0061] The term "and / or" in the following text is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0062] like Figure 1 As shown in the embodiment of this application, a smart AR display method is provided, executed by a server, and the method includes:

[0063] Receive the user's fitting room image;

[0064] Extract the user's first human silhouette 1 from the fitting image;

[0065] The user's body posture information is obtained based on the first human body contour 1;

[0066] The system recommends clothing based on popular fashion trends and user body shape information.

[0067] Send the recommended clothing information to the smart AR fitting mirror so that the smart AR fitting mirror can display the recommended clothing.

[0068] The AR display method provided in this embodiment obtains the user's body shape information by recognizing the user's human body contour, and recommends clothing to the user based on the user's body shape information and popular clothing styles. Finally, the user's fitting effect is displayed in the smart AR fitting mirror. The method provided in this embodiment allows the user to observe the final fitting effect without trying on clothes, enhancing the user's experience and the convenience of shopping.

[0069] Furthermore, this embodiment provides a step for extracting the user's first human body contour 1 based on a fitting image, including:

[0070] Convert the fitting image to an HSV color space image and extract the saturation S channel from the HSV color space image;

[0071] Based on the multi-directional Sobel operator, the gradient images of the saturation S channel in four directions are obtained;

[0072] The absolute gradient image is obtained by fusing gradient images from four directions with linear weighting.

[0073] Binarize the absolute gradient image to obtain the first human body contour 1.

[0074] Specifically, saturation Gradient images of the four directions of the channel , , and They are respectively:

[0075]

[0076]

[0077]

[0078]

[0079] in, Represents the HSV image The gradient images of the saturation S channel in four directions are obtained using the multi-directional Sobel operator. for:

[0080]

[0081] in, , , and They are respectively , , and The weight, To enhance parameters, Next Binarization is performed, i.e., traversal. For each pixel in the middle, when When, record the pixel as 1; when... When that happens, the pixel is recorded as 0, where Representing an image The Middle Line 1 The pixel values ​​of the columns of pixels. This is a threshold value; this embodiment does not impose any limitations on it, but it is exemplary. Finally, only all pixels marked as 1 are retained to obtain the first human body contour image.

[0082] This embodiment extracts contours based on the HSV image space. Compared to the RGB color space, the HSV color space has the H component specifically responsible for representing the hue of a pixel, while the S and V components represent the saturation and brightness of the pixel. Since common shooting backgrounds are mostly red and blue with high saturation, the image is converted from the RGB color space to the HSV color space. Then, the complete human contour is extracted in the saturation S channel image using the multi-directional Sobel operator, avoiding the influence of ambient light.

[0083] Furthermore, this embodiment provides a step for obtaining user posture information based on a first human body contour 1, including:

[0084] The user's first data is obtained based on the first human body contour 1. The first data is the size data of the user's exposed parts.

[0085] The size information of the clothing currently being tried on by the user is obtained from the fitting images;

[0086] Based on the initial data and size information, the size ratio between the user and the currently tried-on clothing is obtained;

[0087] The second data of the user is obtained based on the size ratio relationship. The second data is the size data of the user's obscured part.

[0088] The user's body posture information is obtained based on the second data.

[0089] For example, such as Figure 2 As shown, the solid line represents the first human body contour 1, which is the user's human body contour when dressed, while the dashed line represents the user's actual human body contour, which is the third human body contour 3 mentioned below. Figures 3-5 for Figure 2 A partial schematic diagram. The first data point is the length of the user's exposed wrist. The second data point is the user's waist length as it is covered by clothing. and waist width The specific steps are as follows.

[0090] Firstly, based on image recognition technology, the length of the user's exposed wrist can be accurately obtained from the first human body contour 1. Length of cuffs and the length of the waist of the garment Next, the circumference of the cuffs of the garment currently being tried on by the user is obtained from the fitting image. and waist circumference ;

[0091] according to and This allows us to obtain the length-to-size ratio between the user and the currently tried-on clothing. This ratio can be used to obtain... ;

[0092] like Figure 3 and Figure 4 As shown, the curves of the garment's cuff, the user's wrist, and the garment's waist can all be approximated as elliptical curves. Therefore, according to the formula for the circumference of an ellipse: , and This allows you to determine the waist width of the garment. The width of the cuffs of the garment and the width of the user's wrist ,in The circumference of the user's wrist;

[0093] according to and This allows us to obtain the width-to-width ratio between the user and the currently tried-on garment. This ratio can be used to obtain... .

[0094] For example, when the first data is the length of the user's exposed ankle... The second data point is the length of the user's legs that are covered by clothing. and leg width Similarly, the calculation can be performed using the steps described above, that is, firstly, the length of the user's exposed ankle is obtained based on image recognition technology. Length of trouser legs and the length of the clothes' legs The circumference of the trouser leg of the garment currently being tried on by the user is obtained from the fitting image. and leg circumference And obtain the length-to-size ratio between the user and the currently tried-on clothing. ;

[0095] Next, based on the length-dimension ratio, we obtain... ;

[0096] The width of the garment's leg area was then calculated using the formula for the circumference of an ellipse. The width of clothing and trouser legs and the width of the user's ankle Therefore, the width-to-width ratio between the user and the currently tried-on garment is obtained as follows: And obtain through this proportional relationship .

[0097] This embodiment calculates the user's actual size data by comparing the size ratio between the user and the clothes being tried on, even when the user is wearing loose clothing. This allows for the recommendation of suitable clothing based on the user's body shape information.

[0098] Furthermore, this embodiment provides a step for obtaining a user's body posture information based on second data, including:

[0099] Based on the second data and the preset neural network model, multiple second human body contours of the user are obtained;

[0100] The user's third human body contour 3 is calculated based on multiple second human body contours 2;

[0101] The user's body shape information is calculated based on the third human body contour 3. The user's body shape information includes the user's shoulder width, chest circumference, waist circumference and hip circumference.

[0102] Specifically, such as Figure 6 As shown, in order to obtain more accurate size data from users, one can... , , , The data are input into a preset neural network model, which can generate the overall contour data of the user based on the local data of the user's contour, thereby obtaining multiple second human body contours 2. Then, the average value of the multiple second human body contours 2 is used as the third human body contour 3.

[0103] Specifically, the calculation method for user shoulder width is as follows: Figure 7 As shown, where The left joint of the shoulder. As the endpoint of the left shoulder, along Draw horizontal and vertical lines to the left and up respectively, and the intersection points with the third human figure contour 3 are respectively and And with The angle formed by the tangents at its two ends centered on it. It is definitely smaller than the angle formed by the other contour points around it. And because Must be located and Between, therefore, to determine The specific location only needs to be calculated separately. and The included angle between each contour point, the minimum included angle The corresponding contour points are The specific steps are as follows:

[0104] assumed express and Any contour point between them, corresponding angle The calculation steps are as follows:

[0105] Depend on Points are extracted sequentially downwards along the third human body contour 3. This pixel The coordinates of each pixel are represented as , ,…, Using the least squares method to fit the data... tangent in the upward direction ,assumed Then its slope for:

[0106]

[0107] Depend on Points are extracted sequentially upwards along the third human body contour 3. 1 pixel, obtained after tangent in the downward direction ,assumed , The specific calculation method and same;

[0108] calculate and The angle between ,Right now

[0109] Calculate according to the above steps and The angle between all contour points , , for and The endpoint of the left shoulder can be determined by counting the number of contour points between them. :

[0110]

[0111] Next, determine the right joint point of the shoulder. and along Draw horizontal and vertical lines to the right and above respectively, and the intersection points with the third human figure contour 3 will be respectively... and And the right shoulder endpoint was calculated using the same steps. Two shoulder endpoints and The upward tangent intersects at , and by , and The quadratic Bezier curve generated by these three control points has a shape that is basically consistent with the shoulder width curve, therefore the shoulder width can be calculated based on the quadratic Bezier curve. Assuming... , and The coordinates of these three control points are as follows: , and Then the quadratic Bezier curve is:

[0112]

[0113]

[0114] in Based on the above formula, the specific length of the Bezier curve can be calculated, that is, the shoulder width of the third human figure contour 3 is:

[0115]

[0116] in and They are respectively and The derivative is then used to calculate the user's actual shoulder width according to the conversion formula:

[0117]

[0118] in Indicates the user's actual height. This represents the height of the third human figure silhouette 3.

[0119] Specifically, such as Figure 8 As shown, the waistline of the human body can be further viewed as being composed of two ellipses with equal major axes and unequal minor axes. The horizontal solid line divides the waistline into two parts: the abdomen and the lower back. Because fat tends to accumulate in the abdomen, the abdomen is thicker than the lower back. The length of the horizontal solid line represents the width of the waist. The length of the vertical dotted line represents the thickness of the human waist. , , and These represent the thickness of the user's abdomen and lower abdomen, respectively.

[0120] The user's waist circumference can be obtained by fitting a hyperelliptic curve:

[0121]

[0122] in , , , The value of this value is positively correlated with the size of the opening of the abdominal curve; that is, the smaller the opening of the curve, the more pronounced the abdominal protrusion. The value should be larger.

[0123] The calculation methods for a user's bust and hip circumference are the same as those for waist circumference, both of which can be calculated by fitting an elliptic curve, and will not be explained further here.

[0124] The method provided in this embodiment can obtain the actual dimensions of various parts of the user based on the user's planar contour, so as to make clothing recommendations for the user in the future.

[0125] Furthermore, this embodiment provides a step for obtaining recommended clothing information based on popular clothing trends and user body shape information, including:

[0126] The style and design of the clothing being tried on are determined from the fitting images.

[0127] Based on the style and design of the clothes currently being tried on, the user's body shape information, and a preset database, information on recommended clothing is obtained.

[0128] The style of the clothing being tried on can be determined by recognizing its outline, and the style of the clothing can be determined by the texture of the clothing. The preset database stores information on all the popular clothing items of the season.

[0129] Specifically, the steps for identifying the outline of clothing to be tried on are as follows:

[0130] Obtain the description information of the clothing being tried on based on the fitting images, and extract the image of the clothing from the description information;

[0131] The image is filtered to obtain a denoised image;

[0132] Calculate the gradient and magnitude of each pixel in the denoised image;

[0133] For each pixel, non-maximum suppression is performed in the direction of the gradient to obtain multiple maximum points;

[0134] Multiple edge points are obtained by performing double threshold detection on multiple maxima points;

[0135] Connect multiple edge points to obtain the outline of the clothing to be tried on.

[0136] Specifically, texture features include roughness, contrast, and orientation, where roughness describes the smoothness of the texture, and the formula is:

[0137]

[0138] in The variance representing the grayscale values ​​of an image. This represents the average grayscale value of the image.

[0139] Contrast ratio describes the brightness of an image by statistically analyzing the distribution of its grayscale values. The specific formula is as follows:

[0140]

[0141] in, , The fourth moment of the image, represents the standard deviation of the image's grayscale values.

[0142] Directionality is used to measure whether a texture has obvious directionality, which can be reflected by calculating the gradient magnitude in different directions.

[0143] Specifically, the recommended clothing includes style-recommended clothing and body-fitting clothing. Style-recommended clothing refers to clothing selected from a preset database that matches the user's body shape data based on the style and design of the clothing the user is currently trying on. Body-fitting clothing refers to clothing directly selected from a preset database that matches the user's body shape data.

[0144] This embodiment determines the style and design of clothing by extracting the outline and texture features of the clothing, and recommends suitable clothing to the user based on the style and design of the currently tried-on clothing, the user's body shape information, and popular clothing in the preset database.

[0145] Furthermore, the aforementioned intelligent AR display method also includes:

[0146] Predict the user's joint positions based on the third human body contour 3 and a convolutional neural network;

[0147] By fitting the joint positions to the three-dimensional joint points, a three-dimensional model of the user is obtained.

[0148] Specifically, the Deepcut method based on convolutional neural networks can be used to predict the user's joint positions, and the joint positions can be fitted to a 3D human body model using the SMPL-X model.

[0149] The energy function of the SMPL-X model is:

[0150]

[0151] in , and These are the body shape parameters, posture parameters, and facial expression parameters of the SMPL-X model, respectively. For reprojection loss, , and For weight parameters, For simple priors used for knee and elbow flexion, Body shape as a priori The added contour constraint term represents the fitting error between the contours rendered from the SMPL-X model and the contours obtained based on Graphicality segmentation, and is defined as follows:

[0152]

[0153] in This refers to the set of contour boundary points in the third human body contour 3. This is the set of contour boundary points for each body part in the segmented image of the third human body contour 3. This represents the sum of the closest distances between the contour boundary points in the third human body contour 3 and the contour boundary points in the segmented image.

[0154] This embodiment establishes a three-dimensional human body model of the user by fitting human joints, allowing the user to see the fitting effect more intuitively. Furthermore, the energy function provided in this embodiment increases the adaptability of the model to Asian human body shapes by adding contour constraint terms, avoiding the situation of large model fitting errors.

[0155] Furthermore, embodiments of this application provide a smart AR display method, executed by a smart AR fitting mirror, the method comprising:

[0156] The camera 5 captures images of the user trying on clothes and is installed in the smart AR fitting mirror.

[0157] The fitting image is sent to the server so that the server can extract the user's first human body contour 1 based on the fitting image, and obtain the user's body shape information based on the first human body contour 1. Then, based on popular clothing and the user's body shape information, the server obtains the recommended clothing information and finally sends the recommended clothing information to the smart AR fitting mirror.

[0158] Receive information about recommended clothing items and display the recommended clothing items.

[0159] The intelligent AR display method provided in this application can accurately calculate the user's real body shape information when wearing loose clothing by comparing the size ratio between the user and the clothing being tried on. Then, a three-dimensional human body model is built based on the user's real body shape information. At the same time, clothing is recommended to the user based on the user's real body shape information and popular clothing trends, and finally the try-on effect is displayed in the intelligent AR fitting mirror.

[0160] Furthermore, such as Figure 9 As shown, this application provides an intelligent AR display system, including a server and an intelligent AR fitting mirror;

[0161] The servers include:

[0162] The first receiving module is used to receive the user's fitting images;

[0163] Recommendation module: used to extract the user's first human body outline 1 from the fitting image, then obtain the user's body shape information based on the first human body outline 1, and finally obtain recommended clothing information based on popular outfits and user body shape information;

[0164] The sending module is used to send information about recommended clothing to the smart AR fitting mirror;

[0165] The smart AR fitting mirror includes:

[0166] Camera 5 is used to capture images of the user trying on clothes;

[0167] The communication module 6 is used to send the fitting image to the server so that the server can extract the user's first human body contour 1 based on the fitting image, obtain the user's body shape information based on the first human body contour 1, and finally obtain recommended clothing information based on popular clothing and the user's body shape information.

[0168] The second receiving module 7 is used to receive recommended clothing information sent by the server;

[0169] Display screen 4 is used to show the try-on effect of recommended clothing.

[0170] The intelligent AR display method and system provided in this application accurately calculates the size data of various parts of the user's body by observing the user's human body contour in a dressed state, thereby obtaining the user's body posture information. Based on the user's body posture information and popular clothing trends, the system recommends clothing to the user. At the same time, it can generate a three-dimensional human body model of the user based on the user's body posture information and display the model's fitting effect in an intelligent AR fitting mirror, thereby enhancing the user's experience and the convenience of shopping.

[0171] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0172] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0173] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0174] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. An intelligent AR display method, characterized in that, The method, executed by the server, includes: Receive the user's fitting room image; Extract the user's first human body contour from the fitting image; The user's body posture information is obtained based on the first human body contour; Based on popular fashion trends and the user's body shape information, clothing recommendations are obtained. The information of the recommended clothing items is sent to the smart AR fitting mirror so that the smart AR fitting mirror can display the recommended clothing items.

2. The intelligent AR display method according to claim 1, characterized in that, Extracting the user's first human body contour from the fitting image includes: The fitting image is converted into an HSV color space image, and the saturation S channel in the HSV color space image is extracted. Based on the multi-directional Sobel operator, the gradient images of the saturation S channel in four directions are obtained; Based on the gradient images in the four directions and linear weighted fusion, an absolute gradient image is obtained; The absolute gradient image is binarized to obtain the first human body contour.

3. The intelligent AR display method according to claim 1, characterized in that, The user's body posture information is obtained based on the first human body contour, including: The user's first data is obtained based on the first human body contour, and the first data is the size data of the user's exposed parts. The size information of the clothing currently being tried on by the user is obtained from the fitting image; Based on the first data and the size information, the size ratio between the user and the currently tried-on clothing is obtained; The second data of the user is obtained based on the size ratio relationship, and the second data is the size data of the user's obscured part; The user's body posture information is obtained based on the second data.

4. The intelligent AR display method according to claim 3, characterized in that, The user's body posture information is obtained based on the second data, including: Based on the second data and the preset neural network model, multiple second human body contours of the user are obtained; The user's third human body contour is calculated based on the plurality of second human body contours; The user's body shape information is calculated based on the third human body contour, and the user's body shape information includes the user's shoulder width, chest circumference, waist circumference, and hip circumference.

5. The intelligent AR display method according to claim 4, characterized in that, shoulder width ,in Indicates the user's actual height. This represents the height of the third human body silhouette. , representing the shoulder width of the third human body contour. and They are respectively and The derivative of, where , , and This represents the coordinates of the two endpoints of the shoulder of the third human body contour. The coordinates represent the intersection of the tangents at the two endpoints. .

6. The intelligent AR display method according to claim 4, characterized in that, The waist circumference ,in This indicates the length of the user's waist. and These represent the widths of the user's abdomen and lower abdomen, respectively. , representing the width of the user's waist.

7. The intelligent AR display method according to claim 1, characterized in that, The information for recommending clothing based on popular fashion trends and the user's body shape information includes: The style and design of the clothing being tried on are obtained based on the fitting images; Based on the style and design of the currently tried-on clothing, the user's body shape information, and a preset database, information on recommended clothing is obtained.

8. The intelligent AR display method according to claim 4, characterized in that, Also includes: Predict the user's joint positions based on the third human body contour and the convolutional neural network; The joint positions are fitted to three-dimensional joint points to obtain the user's three-dimensional model.

9. An intelligent AR display method, characterized in that, Performed by a smart AR fitting mirror, the method includes: The smart AR fitting mirror captures images of the user trying on clothes using a camera, which is installed in the mirror. The fitting image is sent to the server, so that the server can extract the user's first human body contour based on the fitting image, obtain the user's body shape information based on the first human body contour, and then obtain recommended clothing information based on popular clothing and the user's body shape information, and finally send the recommended clothing information to the smart AR fitting mirror. Receive the information about the recommended clothing items and display the recommended clothing items.

10. An intelligent AR display system, characterized in that, Includes servers and smart AR fitting mirrors; The server includes: The first receiving module is used to receive the user's fitting images; Recommendation module: used to extract the user's first human body contour based on the fitting image, then obtain the user's body shape information based on the first human body contour, and finally obtain recommended clothing information based on popular outfits and the user's body shape information; The sending module is used to send the information of the recommended clothing to the smart AR fitting mirror; The intelligent AR fitting mirror includes: A camera is used to capture images of users trying on clothes; The communication module is used to send the fitting image to the server, so that the server can extract the user's first human body contour based on the fitting image, obtain the user's body shape information based on the first human body contour, and finally obtain recommended clothing information based on popular clothing and the user's body shape information. The second receiving module is used to receive the recommended clothing information sent by the server; A display screen is used to show the try-on effect of the recommended clothing.

Images