Remote virtual fitting method and system based on double-screen guiding and multi-size fitting

Abstract

The present application relates to the technical field of image processing, and proposes a remote virtual fitting method and system based on double-screen guidance and multi-size fitting, comprising: configuring a double-screen interactive terminal in a fitting area, and performing an image acquisition operation of a user on the double-screen interactive terminal; performing image normalization processing on the user image in a remote processing platform to obtain a target user image; establishing a loose amount mapping relationship between the key body shape girth feature of the user and the version feature vector of the target garment, generating a virtual fitting effect drawing of the user in each size; and outputting a size recommendation conclusion of the target garment to the user through the double-screen interactive terminal according to the virtual fitting effect drawing. The present application can effectively improve the fitness of the virtual fitting effect drawing and the actual body shape of the user.

Description

Technical Field

[0001] This invention belongs to the field of image processing technology and relates to a remote virtual fitting method and system based on dual-screen guidance and multi-size fitting. Background Technology

[0002] With the popularization of online clothing retail, virtual try-on technology, as a tool to help users judge clothing size and wearing effect, directly affects users' purchasing decisions with its recommendation accuracy and interactive realism.

[0003] In existing remote virtual fitting solutions, users mostly complete image acquisition by taking selfies with ordinary cameras. Due to the lack of standardized constraints on shooting posture during the acquisition process, different users have differences in body orientation, limb extension, and head height when taking pictures. In addition, the system usually uses a fixed-size clothing template to simply overlay the user's image without considering the continuous changes in the ease of the fit between different sizes. This makes it impossible to present the difference in fit of the same garment under different sizes, resulting in a low degree of match between the generated fitting effect image and the user's actual body shape. Summary of the Invention

[0004] This invention provides a remote virtual fitting method and system based on dual-screen guidance and multi-size fitting, the main purpose of which is to effectively improve the matching degree between the virtual fitting effect image and the user's actual body shape.

[0005] To achieve the above objectives, the present invention provides a remote virtual fitting method based on dual-screen guidance and multi-size fitting, comprising: A dual-screen interactive terminal is configured in the fitting area, the dual-screen interactive terminal including a first display screen and a second display screen arranged side by side; Obtain the user's basic body posture parameters and the clothing identification information of the target garment selected by the user within the fitting area; Based on the basic body posture parameters, the user's image acquisition operation is performed on the dual-screen interactive terminal to obtain the user's image; The user image is sent to a remote processing platform, where image normalization processing is performed to obtain the target user image. Based on the target user image, extract the user's key body circumference features; Based on the clothing identification information, retrieve the pattern feature vector of the target clothing; Construct a looseness mapping relationship between the pattern feature vector and the key body shape circumference features. Based on the looseness mapping relationship, overlay the pattern template of the target garment in at least two sizes onto the target user image to generate a virtual try-on effect image of the user in at least two sizes. Based on the virtual try-on effect image, the size recommendation conclusion of the target clothing is output to the user through the dual-screen interactive terminal.

[0006] Optionally, based on the basic body posture parameters, the user's image acquisition operation is performed on the dual-screen interactive terminal to obtain a user image, including: Based on the height data in the basic body posture parameters, adjust the lens focal length and framing range of the image acquisition device connected to the dual-screen interactive terminal so that the user's full-body image falls within the preset framing frame of the image acquisition device. Once the user's full-body image falls within the preset viewfinder, the user's posture guidance information is displayed on the first display screen. The second display screen shows the real-time view captured by the image acquisition device, and the real-time view is superimposed with a head alignment line and a shoulder horizontal line. The system receives a user shooting command sent by a remote control connected to the dual-screen interactive terminal, and responds to the user shooting command by controlling the image acquisition device to take a picture and obtain a user image.

[0007] Optionally, the posture guidance information includes a standard posture diagram and a framing prompt box. The standard posture diagram shows the user's target standing posture, and the framing prompt box is superimposed on the standard posture diagram in the form of a dashed frame to prompt the user about the target standing area. The head alignment line is a horizontal reference line used to constrain the user's head height position. The shoulder horizontal line is a horizontal reference line used to constrain the user's shoulder horizontal posture.

[0008] Optionally, the user image is subjected to image normalization processing in the remote processing platform to obtain the target user image, including: The user image is subjected to quality pre-inspection processing in the remote processing platform to obtain a pre-inspection qualified user image. The quality pre-inspection processing includes at least image sharpness verification, human image integrity verification and posture angle verification. The pre-inspected and qualified user images are preprocessed to obtain the target user image. The preprocessing operations include at least background removal, artifact removal, and image quality correction.

[0009] Optionally, constructing a loose mapping relationship between the pattern feature vector and the key body shape circumference features includes: Obtain the circumference values ​​corresponding to each part in the key body shape circumference features, and the pattern size corresponding to each part in each size in the pattern feature vector, wherein the parts include the chest, waist, hips and shoulders; Calculate the difference between the pattern size and the circumference value for each part of each size to obtain the initial ease difference for each part of each size; The initial slack difference is adjusted by weighting the slack weights of each part to obtain the weighted slack difference. All the weighted ease difference values ​​for each size are arranged according to the location dimension to form an ease distribution matrix, and the ease distribution matrix is ​​used as the ease mapping relationship.

[0010] Optionally, the preset ease weighting coefficients for each part are dynamically determined based on the category of the target garment: When the target garment is a top, the ease weighting coefficient corresponding to the chest is greater than the ease weighting coefficient corresponding to the waist, and the ease weighting coefficient corresponding to the waist is greater than the ease weighting coefficient corresponding to the hips. When the target garment is trousers, the ease weighting coefficient corresponding to the hips is greater than the ease weighting coefficient corresponding to the waist, and the ease weighting coefficient corresponding to the waist is greater than the ease weighting coefficient corresponding to the legs. When the target garment is a skirt, the ease weighting coefficient corresponding to the waist is greater than or equal to the ease weighting coefficient corresponding to the hips.

[0011] Optionally, based on the ease mapping relationship, the pattern templates of the target garment in at least two sizes are overlaid onto the target user image to generate a virtual try-on effect image of the user in at least two sizes, including: Based on the ease mapping relationship, the suitable size is identified from all available sizes of the target garment; Use the suitable size as the baseline size, and select at least one other size that is different from the suitable size; Based on the ease mapping relationship, the ease distribution characteristics of the baseline size and the other sizes at various locations are determined respectively; Based on the looseness distribution characteristics, non-rigid deformation processing is performed on the pattern templates corresponding to the base size and the other sizes, respectively; The templates of the baseline size and other sizes after non-rigid deformation processing are superimposed on the target user image to generate virtual try-on effect images of the user in each size.

[0012] Optionally, the non-rigid deformation treatment of the template includes: Extract the two-dimensional outline of the template; Multiple deformation control points are established on the outline of the two-dimensional plate. Calculate the displacement vector of each deformation control point based on the slack distribution characteristics corresponding to the deformation control point; Based on all the displacement vectors, the two-dimensional plate outline is interpolated and deformed to obtain the deformed plate template.

[0013] Optionally, based on the virtual try-on effect image, the user is provided with a size recommendation for the target garment via the dual-screen interactive terminal, including: Extract the visual distance data between the outline of the pattern template and the outline of the user's body surface in the virtual try-on effect image; Based on the visual spacing data, and combined with the ease values ​​corresponding to each size of the target garment, fit evaluation information and size recommendation ranking results for each size are generated. The first display screen shows virtual try-on images of the user in various sizes. The second display screen shows the ease value and fit evaluation information corresponding to each size, and the size recommendation ranking results are displayed on the first display screen or the second display screen.

[0014] To address the aforementioned problems, the present invention also provides a remote virtual fitting system based on dual-screen guidance and multi-size fitting, which executes the aforementioned remote virtual fitting method based on dual-screen guidance and multi-size fitting. The system includes: A dual-screen interaction configuration module is used to configure a dual-screen interaction terminal in the fitting area. The dual-screen interaction terminal includes a first display screen and a second display screen arranged side by side. The information acquisition module is used to acquire the basic body posture parameters of the user in the fitting area and the clothing identification information of the target clothing selected by the user; The image acquisition module is used to perform the user's image acquisition operation on the dual-screen interactive terminal based on the basic body posture parameters, and obtain the user's image; The image processing module is used to send the user image to a remote processing platform, and perform image normalization processing on the user image in the remote processing platform to obtain the target user image; The body shape feature extraction module is used to extract the key body shape circumference features of the user based on the target user image; The pattern feature extraction module is used to retrieve the pattern feature vector of the target garment based on the garment identification information; The fitting comparison module is used to construct the ease mapping relationship between the pattern feature vector and the key body shape circumference features. Based on the ease mapping relationship, the pattern template of the target garment in at least two sizes is superimposed on the target user image to generate a virtual fitting effect image of the user in at least two sizes. The size recommendation module is used to output a size recommendation conclusion for the target garment to the user through the dual-screen interactive terminal based on the virtual try-on effect image.

[0015] Compared to the problems described in the background art, this invention, by configuring a dual-screen interactive terminal in the fitting area, uses a first screen to display posture guidance information and a second screen to simultaneously display the user's real-time view. This allows the user to adjust their posture by comparing the guidance information with the real-time view, improving the first-pass yield of user image acquisition. Furthermore, this invention performs image acquisition operations based on the aforementioned basic body posture parameters, using a dual-screen guidance mechanism to constrain the user's body orientation, head height, and limb extension during shooting. This ensures consistency in posture and scale in the acquired user images, providing standardized image input for subsequent body shape feature extraction based on these images, thereby improving the match between the virtual fitting effect image and the user's actual body shape. Finally, this invention obtains the target user image by performing image normalization processing on the user image in the remote processing platform, avoiding issues arising from the original image... Differences in lighting, background, clarity, and posture angle can lead to errors in feature extraction, thus improving the accuracy and consistency of subsequent body shape feature extraction. Furthermore, by constructing a looseness mapping relationship between the pattern feature vector and the key body shape features, this invention can obtain quantitative looseness differences between the user's body shape and various sizes of patterns at different locations. This provides quantifiable fit data for generating multi-size fitting effect images, avoiding selection bias caused by the lack of objective data support when users rely on subjective experience to judge sizes. Finally, based on the virtual fitting effect image, this invention outputs the size recommendation conclusion of the target garment to the user through the dual-screen interactive terminal. This presents the differences in fit, looseness, and silhouette of each size to the user in a quantitative manner, allowing the user to obtain a clear size recommendation conclusion through the synergy of visual comparison and data verification. Therefore, this invention can effectively improve the consistency between the virtual fitting effect image and the user's actual body shape. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a remote virtual fitting method based on dual-screen guidance and multi-size fitting, provided as an embodiment of the present invention. Figure 2 This is a schematic diagram of the installation layout of a dual-screen interactive terminal for a remote virtual fitting method based on dual-screen guidance and multi-size fitting, provided in an embodiment of the present invention. Figure 3 The flowchart of dual-screen guided image acquisition is provided as an embodiment of the remote virtual fitting method based on dual-screen guidance and multi-size fitting according to an embodiment of the present invention. Figure 4A schematic diagram of a remote virtual fitting system based on dual-screen guidance and multi-size fitting provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of a computer device for implementing a remote virtual fitting method based on dual-screen guidance and multi-size fitting, as provided in an embodiment of the present invention.

[0017] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and the accompanying drawings. Detailed Implementation

[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. This application provides a remote virtual fitting method based on dual-screen guidance and multi-size fitting. The executing entity of this remote virtual fitting method based on dual-screen guidance and multi-size fitting includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the remote virtual fitting method based on dual-screen guidance and multi-size fitting can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0019] See Figure 1 The diagram shown is a flowchart illustrating a remote virtual fitting method based on dual-screen guidance and multi-size fitting, according to an embodiment of the present invention. In this embodiment, the remote virtual fitting method based on dual-screen guidance and multi-size fitting includes: S1. A dual-screen interactive terminal is configured in the fitting area, the dual-screen interactive terminal including a first display screen and a second display screen arranged side by side.

[0020] This invention utilizes a dual-screen interactive terminal within the fitting area. The first screen displays posture guidance information, while the second screen simultaneously displays the user's real-time view. This allows the user to adjust their posture by comparing the guidance information with the real-time view, thereby improving the first-pass yield of the user's image capture.

[0021] See Figure 2 The figure shows a schematic diagram of the installation layout of the dual-screen interactive terminal provided by the present invention. As shown, the dual-screen interactive terminal is installed in the fitting room, and the first display screen and the second display screen are arranged side by side on the same support structure. The fitting room is also equipped with an image acquisition device and a remote control. The image acquisition device is located above or to the side of the dual-screen interactive terminal, with the lens facing the user's standing area. The remote control is wirelessly connected to the dual-screen interactive terminal, and the user holds the remote control and stands facing the dual-screen interactive terminal.

[0022] The first display screen is any one of a liquid crystal display screen, an LED display screen, or an OLED display screen, and is used to display posture guidance information to the user. The posture guidance information includes, but is not limited to, a standard posture diagram and a framing prompt box. The framing prompt box is displayed superimposed on the standard posture diagram in the form of a dashed frame to indicate the target standing area to the user. The second display screen is also any one of a liquid crystal display screen, an LED display screen, or an OLED display screen, and its size may be the same as or different from the first display screen. It is used to display the framing image captured by the image acquisition device in real time.

[0023] In specific implementation, the dual-screen interactive terminal has the following display control logic: the first display screen and the second display screen are driven independently. The first display screen displays static guidance content, and the second display screen displays dynamic framing images. The two are updated synchronously and do not interfere with each other. When the first display screen displays static guidance content, the static guidance content includes a standard posture diagram and a framing prompt box. The standard posture diagram shows the user's target standing posture, including facing forward, arms hanging naturally, and feet shoulder-width apart. The framing prompt box is displayed superimposed on the standard posture diagram in the form of a dashed frame to indicate the user's target standing area. When the second display screen displays dynamic framing images, the dynamic framing images are video streams captured in real time by the image acquisition device, and a head alignment line and a shoulder horizontal line are superimposed on the video stream. The head alignment line is a horizontal reference line used to constrain the user's head height position, and the shoulder horizontal line is a horizontal reference line used to constrain the user's shoulder horizontal posture.

[0024] Furthermore, the fitting area of ​​this invention is also equipped with an image acquisition device and a remote control. The image acquisition device is located above or to the side of the dual-screen interactive terminal, with its lens facing the user's standing area, for acquiring user images. The image acquisition device is connected to the dual-screen interactive terminal via wired or wireless communication. The dual-screen interactive terminal responds to the trigger command of the remote control to control the image acquisition device to take a picture or retake a picture. The remote control and the dual-screen interactive terminal establish a communication connection via any one of Bluetooth, Wi-Fi, infrared, or 2.4G wireless communication protocols. The user sends corresponding control commands to the dual-screen interactive terminal via the wireless communication connection by pressing the photo or retake button on the remote control.

[0025] It should be noted that in the above dual-screen display control logic, the display content of the first display screen and the second display screen are driven by the controller built into the dual-screen interactive terminal respectively. The controller includes an independent display buffer and refresh sequence to ensure that the guidance content and the real-time view are displayed synchronously in time and presented independently in space, without overlapping or obstructing each other.

[0026] S2. Obtain the user's basic body shape parameters and the clothing identification information of the target garment selected by the user in the fitting area.

[0027] This invention obtains the user's basic body posture parameters and the clothing identification information of the target clothing selected by the user within the fitting area, thereby avoiding the problem of repeated processes or data mismatch caused by the separation of user identity and target clothing information, and ensuring the continuity and accuracy of the virtual fitting process.

[0028] The basic body parameters include the user's height, weight, gender, and age; the clothing identification information includes clothing ID, clothing name, SKU code, barcode, QR code, or RFID tag.

[0029] In an optional embodiment of the present invention, the basic body posture parameters are obtained in the following way: the dual-screen interactive terminal is equipped with a touch screen, and after the user stands in the fitting area, he or she manually inputs his or her height, weight, gender and age information through the touch screen.

[0030] In another optional embodiment of the present invention, the clothing identification information is obtained in the following manner: the user manually selects the target clothing in the clothing display interface through the touch screen of the dual-screen interactive terminal, and the dual-screen interactive terminal responds to the user's selection operation by obtaining the clothing identification information corresponding to the target clothing.

[0031] It should be noted that the acquisition method of the basic body posture parameters is not limited to manual input by the user. In other embodiments of the present invention, the basic body posture parameters can also be acquired through an identity recognition device connected to the dual-screen interactive terminal. For example, the identity recognition device is a card reader, a face recognition camera, or a fingerprint recognition module. After the user completes identity authentication by swiping a membership card, face recognition, or fingerprint recognition, the dual-screen interactive terminal retrieves the user's pre-stored basic body posture parameters from the user database of the remote processing platform, without requiring the user to input them repeatedly.

[0032] Similarly, the method of obtaining the clothing identification information is not limited to manual selection by the user. In other embodiments of the present invention, the user can also scan the barcode or QR code of the target clothing using a scanning device configured in the fitting area, or read the RFID tag of the target clothing using near-field communication. The scanning device or near-field communication reader is communicatively connected to the dual-screen interactive terminal, and the dual-screen interactive terminal obtains the clothing identification information in response to the scanning or reading result.

[0033] S3. Based on the basic body posture parameters, perform the user's image acquisition operation on the dual-screen interactive terminal to obtain the user's image.

[0034] This invention performs image acquisition operations based on the aforementioned basic body posture parameters. Through a dual-screen guidance mechanism, it constrains the user's body orientation, head height, and limb extension during shooting, ensuring that the acquired user images are consistent in posture and scale. This provides standardized image input for subsequent extraction of body circumference features based on the image, thereby improving the match between the virtual try-on effect image and the user's actual body shape.

[0035] As an embodiment of the present invention, based on the basic body posture parameters, the user's image acquisition operation is performed on the dual-screen interactive terminal to obtain a user image, including: Based on the height data in the basic body posture parameters, adjust the lens focal length and framing range of the image acquisition device connected to the dual-screen interactive terminal so that the user's full-body image falls within the preset framing frame of the image acquisition device. Once the user's full-body image falls within the preset viewfinder, the user's posture guidance information is displayed on the first display screen. The second display screen shows the real-time view captured by the image acquisition device, and the real-time view is superimposed with a head alignment line and a shoulder horizontal line. The system receives a user shooting command sent by a remote control connected to the dual-screen interactive terminal, and responds to the user shooting command by controlling the image acquisition device to take a picture and obtain a user image.

[0036] The preset viewfinder is the framing range of the image acquisition device at a preset focal length, used to constrain the integrity of the user's full-body image within the field of view of the image acquisition device; the posture guidance information includes a standard posture diagram and a framing prompt box, the framing prompt box being superimposed on the standard posture diagram in the form of a dashed frame, used to indicate the target standing area to the user; the head alignment line is a horizontal reference line used to constrain the height position of the user's head; the shoulder horizontal line is a horizontal reference line used to constrain the horizontal posture of the user's shoulders.

[0037] During the image acquisition process described above, the user adjusts their standing position and body orientation by comparing their current posture on the second display screen with the standard posture on the first display screen, until all alignment conditions are met before triggering the photo capture. (See also...) Figure 3The diagram shows the dual-screen guided image acquisition flowchart provided by this invention. As shown, the user first stands in the fitting area, facing the dual-screen interactive terminal, and views the standard posture diagram and framing prompts on the first display screen. Then, the user views the real-time framing on the second display screen, observing the deviation between their current posture and the head alignment line and shoulder horizontal line. Based on the observed deviations, the user adjusts their position, and then confirms again whether their head is aligned with the head alignment line, their shoulders with the shoulder horizontal line, and their body within the framing prompt area. If any condition is not met, the user continues to adjust their position; if all three conditions are met, the image acquisition device is triggered to take a picture via remote control. Finally, the user confirms the shooting result. If dissatisfied with the result, a retake is triggered via remote control, returning to view the real-time framing on the second display screen and readjusting the posture; if satisfied with the result, image acquisition is complete.

[0038] S4. The user image is sent to a remote processing platform, where the user image is processed for image normalization to obtain the target user image.

[0039] By sending the user's image to a remote processing platform, this invention enables centralized deployment and unified scheduling of image processing resources, reducing the hardware cost and maintenance burden of local terminals.

[0040] Specifically, the remote processing platform can be any one of a cloud server, a local server, or an edge computing device. Cloud servers are suitable for centralized processing scenarios involving multiple stores, local servers are suitable for independent processing scenarios involving a single store, and edge computing devices are suitable for scenarios with high requirements for response latency. Those skilled in the art can choose according to actual application needs.

[0041] Furthermore, by performing image normalization processing on the user image in the remote processing platform to obtain the target user image, the present invention can avoid feature extraction errors caused by differences in lighting, background, sharpness, pose angle, etc. of the original image, thereby improving the accuracy and consistency of subsequent body shape feature extraction.

[0042] As an embodiment of the present invention, the user image is subjected to image normalization processing in the remote processing platform to obtain a target user image, including: The user image is subjected to quality pre-inspection processing in the remote processing platform to obtain a pre-inspection qualified user image. The quality pre-inspection processing includes at least image sharpness verification, human image integrity verification and posture angle verification. The pre-inspected and qualified user images are preprocessed to obtain the target user image. The preprocessing operations include at least background removal, artifact removal, and image quality correction.

[0043] The image clarity verification checks whether the clarity of the user image reaches a preset clarity threshold. If it does not reach the preset clarity threshold, the user image is deemed unqualified, and a retake command is sent to the dual-screen interactive terminal. If it reaches the preset clarity threshold, the next verification step is performed. The human image integrity verification checks whether the user image contains the complete human body outline of the user. If any part of the user's head, torso, or limbs is missing or exceeds the image boundary, the user image is deemed unqualified, and a retake command is sent to the dual-screen interactive terminal. If the user image contains the complete human body outline of the user, the next verification step is performed. The posture angle verification checks whether the user's frontal orientation in the user image meets a preset angle range. If the user's frontal orientation deviates from the preset angle range, the user image is deemed unqualified, and a retake command is sent to the dual-screen interactive terminal. If the user's frontal orientation meets the preset angle range, the user image is deemed to have passed the pre-inspection, and a pre-inspection qualified user image is obtained. The background removal is used to identify and separate the background area from the user area in the pre-qualified user image, remove the background area, and retain the user area. The artifact removal is used to identify artifact areas in the pre-qualified user image caused by shooting angle or occlusion, and to repair these artifact areas. The artifact areas include at least one of the following: distorted areas at the edges of the user's clothing, gaps between the clothing and the body, and abnormal bright spots caused by light reflection. The image quality correction is used to adjust the brightness, contrast, and color balance of the pre-qualified user image so that the color and brightness of the user image meet the image quality standards for subsequent virtual try-on.

[0044] It should be noted that if any one of the above three checks fails, the user image is deemed unqualified and a retake instruction is returned; only after all three checks pass is the pre-inspection deemed qualified and the preprocessing operation begins.

[0045] In an optional embodiment of the present invention, the preset sharpness threshold is determined based on a sample image library pre-stored in the remote processing platform. The sample image library contains multiple sample images with different sharpness levels. By scoring the sharpness of the sample images and statistically analyzing the sharpness distribution, the sharpness value corresponding to the median of the score distribution is selected as the preset sharpness threshold. The preset angle range is a deviation of no more than ±15° relative to the optical axis direction of the image acquisition device. This range is determined based on the maximum permissible deviation between the horizontal line of the shoulder and the horizontal reference line of the image in a frontal standing posture, to ensure the consistency of the user's frontal orientation during subsequent body circumference feature extraction. The optical axis direction of the image acquisition device is the direction of the line connecting the center point of the image acquisition device lens to the center point of the user's standing area, and the center point of the user's standing area is the projection center of the preset viewfinder on the ground. The target resolution is determined based on the minimum pixel requirement and maximum allowable computational load of the input image in the image processing algorithm of the remote processing platform. The minimum pixel requirement is the minimum number of pixels required to ensure that the subsequent body circumference feature extraction algorithm can effectively identify the human contour in the image. The maximum allowable computational load is the upper limit of the image size corresponding to the maximum computational resource consumption that the remote processing platform can withstand when processing a single image. The preset unified format is JPEG, which has strong compatibility and high compression rate, and is suitable for image storage and transmission in multi-terminal upload scenarios. Those skilled in the art can also use any of the formats such as PNG, BMP, or WEBP as the preset unified format according to the actual application scenario.

[0046] S5. Based on the target user image, extract the user's key body circumference features.

[0047] This invention extracts key body shape features of users and expands the size matching criteria from height and weight to local circumference data such as chest circumference, waist circumference, and hip circumference, so that users with the same height and weight but different body types can obtain differentiated size recommendations. The key body shape features include at least one of chest circumference, waist circumference, hip circumference, shoulder width, arm length, leg length, and height.

[0048] As an embodiment of the present invention, based on the target user image, key body circumference features of the user are extracted, including: Identify the user's body shape outline in the target user image to obtain a body shape outline image; Based on the body contour image, the user's key nodes are located, including the neck node, shoulder node, chest node, waist node, and hip node. Based on the key nodes, the circumference dimensions corresponding to the user's chest, waist and hip circumference are calculated to obtain the user's key body shape circumference characteristics.

[0049] In an optional embodiment of the present invention, the key nodes of the user are located in the following manner: the target user image is input into a pre-trained human pose estimation model, and the human pose estimation model outputs the two-dimensional coordinate positions of each key part of the human body in the target user image; based on the two-dimensional coordinate positions, the neck node, left and right acromion nodes, chest node, left and right waist nodes, abdomen node, and left and right hip nodes are determined.

[0050] In practice, the human pose estimation model can be implemented using any of the network structures OpenPose, HRNet, or AlphaPose, all of which have the function of outputting the coordinates of human key points.

[0051] In another optional embodiment of the present invention, the perimeter of the horizontal cross section formed by the chest node and the left and right armpit points is the chest circumference, the perimeter of the horizontal cross section formed by the waist node and the left and right side waist points is the waist circumference, and the perimeter of the horizontal cross section formed by the hip node and the left and right side hip points is the hip circumference.

[0052] S6. Based on the clothing identification information, retrieve the pattern feature vector of the target clothing.

[0053] This invention retrieves the pattern feature vector of the target garment based on the garment identification information, enabling subsequent size matching to be based on the actual pattern data of the garment, thus avoiding the impact of differences in garment patterns on the fitting effect.

[0054] The pattern feature vector is a multi-dimensional numerical vector of the target garment in each size, and its components correspond to the bust size, waist size, hip size, shoulder width, garment length, sleeve length and pattern ease coefficient, respectively. The pattern ease coefficient is used to characterize the design allowance of the target garment relative to the naked body circumference in the corresponding size, and is determined by the design pattern of the target garment.

[0055] As an embodiment of the present invention, retrieving the pattern feature vector of the target garment based on the garment identification information includes: Retrieve garment pattern data that matches the garment identification information from a preset pattern database; Extract the pattern parameters of the target garment for each size from the garment pattern data. The pattern parameters include the chest circumference, waist circumference, hip circumference, shoulder width, garment length, sleeve length and pattern ease coefficient corresponding to each size. The pattern parameters of each size are arranged in a preset dimensional order to obtain the pattern feature vector of the target garment in each size.

[0056] The pattern database is preset based on the original pattern sheets for each garment style provided by the apparel supplier. The original pattern sheets contain baseline pattern data for each garment style under a base size, as well as size difference data between adjacent sizes. The apparel supplier periodically submits or updates the original pattern sheets to the remote processing platform according to the apparel brand, style, and season. The remote processing platform stores the received original pattern sheets in the pattern database. The preset dimension order refers to the fixed arrangement order of the dimensional components in the pattern feature vector, in the following order: bust, waist, hip, shoulder width, garment length, sleeve length, and ease factor. For example, the pattern feature vector for size S can be represented as [88,72,92,40,62,58,1.05], for size M as [92,76,96,41.2,64,60,1.08], and for size L as [96,80,100,42.4,66,62,1.11].

[0057] S7. Construct a looseness mapping relationship between the pattern feature vector and the key body shape circumference feature. Based on the looseness mapping relationship, overlay the pattern template of the target garment in at least two sizes onto the target user image to generate a virtual try-on effect image of the user in at least two sizes.

[0058] This invention constructs a mapping relationship between the pattern feature vector and the key body shape circumference features to obtain the quantitative difference in ease between the user's body shape and each size pattern at various parts. This provides part-by-part, quantifiable fit data for generating multi-size fitting effect diagrams, avoiding selection bias caused by the lack of objective data support when users judge the size based on subjective experience.

[0059] The ease mapping relationship refers to the correspondence between a user's body circumference and the garment's pattern size at different sizes, used to characterize the fit and comfort of the user at different body parts within the target garment at various sizes. Specifically, the ease mapping relationship uses size as an index, body part as a dimension, and ease difference as a numerical value, structurally organizing the ease difference of the user at different sizes of the same target garment according to size and body part.

[0060] As an embodiment of the present invention, constructing the loose mapping relationship between the pattern feature vector and the key body shape circumference features includes: Obtain the circumference values ​​corresponding to each part in the key body shape circumference features, and the pattern size corresponding to each part in each size in the pattern feature vector, wherein the parts include the chest, waist, hips and shoulders; Calculate the difference between the pattern size and the circumference value for each part of each size to obtain the initial ease difference for each part of each size; The initial slack difference is adjusted by weighting the slack weights of each part to obtain the weighted slack difference. All the weighted ease difference values ​​for each size are arranged according to the location dimension to form an ease distribution matrix, and the ease distribution matrix is ​​used as the ease mapping relationship.

[0061] The areas mentioned include, but are not limited to, the chest, waist, hips, and shoulders. Depending on the type of clothing, the areas may be added or removed. For example, for trousers, the areas may include at least the waist and hips; for tops, the areas may include at least the chest, waist, and shoulders.

[0062] In an optional embodiment of the present invention, the initial ease difference is calculated using the following formula: Initial ease difference = Pattern size The circumference values ​​are as follows: a positive difference indicates that there is room for movement in that area, i.e., looseness; a negative difference indicates that the size of the garment in that area is smaller than the user's actual circumference, i.e., a tight fit; and a zero difference indicates a perfect fit.

[0063] In another embodiment of the present invention, the preset ease weighting coefficients for each part are dynamically determined according to the category of the target garment: When the target garment is a top, the ease weighting coefficient corresponding to the chest is greater than the ease weighting coefficient corresponding to the waist, and the ease weighting coefficient corresponding to the waist is greater than the ease weighting coefficient corresponding to the hips. When the target garment is trousers, the ease weighting coefficient corresponding to the hips is greater than the ease weighting coefficient corresponding to the waist, and the ease weighting coefficient corresponding to the waist is greater than the ease weighting coefficient corresponding to the legs. When the target garment is a skirt, the ease weighting coefficient corresponding to the waist is greater than or equal to the ease weighting coefficient corresponding to the hips.

[0064] In this step, due to differences in wearing methods and covered areas, users perceive significantly different weights for the perceived looseness of different parts of the garment when wearing it. Specifically: When the target garment is a top, including but not limited to T-shirts, shirts, jackets, and knitwear, the looseness of these areas, which are primarily supported by the chest and shoulders, directly determines wearing comfort and visual appeal, and therefore receives a high weight coefficient. The waist, as the area where the top's hem is gathered, receives the next highest weight. The hips, due to the lower coverage of the top, especially short tops, receive the lowest weight. When the target garment is trousers, including but not limited to trousers, shorts, jeans, and dress pants, the wearing comfort of trousers is mainly determined by the hips, waist, and legs, so these three areas receive correspondingly higher weight coefficients. Among them, the hips are the most crucial fit area of ​​trousers, and their looseness directly affects the comfort of sitting and squatting, thus receiving the highest weight. When the target garment is a skirt, including but not limited to dresses, skirts, A-line skirts, and straight skirts, the weighting coefficients for these two parts are relatively high because the fit of the skirt mainly depends on the waist and hips.

[0065] For example, in the skirt category described in this invention, such as a dress, the weighting coefficients for the ease allowance of each part can be set as follows: bust 0.3, waist 0.35, hips 0.3, and shoulders 0.05, wherein the sum of the weighting coefficients for each part is 1.

[0066] It should be noted that the looseness weight coefficient can be preset according to the category of the target garment, or it can be personalized according to the user's historical wearing preference data. The specific value of the weight coefficient is not limited to the one listed in this embodiment. Those skilled in the art can set it reasonably according to the actual application scenario.

[0067] It should also be noted that the ease mapping relationship described in this invention is not limited to matrix form; it can also be stored and calculated using a multi-dimensional vector. Specifically, the weighted ease differences corresponding to each size are concatenated into a one-dimensional vector according to the body part, and this vector is stored in association with the size label. The vector form and the matrix form are mathematically equivalent and both can achieve the ease mapping purpose of this invention.

[0068] Furthermore, by superimposing the pattern templates of the target garment in at least two sizes onto the target user image according to the ease mapping relationship, the present invention generates virtual try-on effect images of the user in at least two sizes. This allows users to intuitively compare the wearing effect of the same garment in different sizes on the same interface, effectively reducing the subjectivity and uncertainty of size judgment based solely on a single static size image or text size chart in traditional solutions, and providing users with a visually comparable basis for size selection.

[0069] As an embodiment of the present invention, according to the ease mapping relationship, the pattern templates of the target garment in at least two sizes are superimposed on the target user image to generate a virtual try-on effect image of the user in at least two sizes, including: Based on the ease mapping relationship, the suitable size is identified from all available sizes of the target garment; Use the suitable size as the baseline size, and select at least one other size that is different from the suitable size; Based on the ease mapping relationship, the ease distribution characteristics of the baseline size and the other sizes at various locations are determined respectively; Based on the looseness distribution characteristics, non-rigid deformation processing is performed on the pattern templates corresponding to the base size and the other sizes, respectively; The templates of the baseline size and other sizes after non-rigid deformation processing are superimposed on the target user image to generate virtual try-on effect images of the user in each size.

[0070] In this invention, the suitable size corresponds to a user's commonly perceived fit size or recommended size. It can be calculated based on the weighted ease difference of each size across various body parts in the ease distribution matrix: for any size, the absolute values ​​of the weighted ease differences across body parts are summed to obtain the overall ease deviation value of that size. The smaller the overall ease deviation value, the lower the degree of deviation between the size and the user's overall body shape across various body parts. Therefore, the size with the smallest overall ease deviation value among all available sizes can be taken as the suitable size. The ease distribution feature refers to the set of ease difference values ​​corresponding to a specified size in the ease mapping relationship. As mentioned above, the ease mapping relationship is stored in the form of an ease distribution matrix, where rows correspond to different sizes and columns correspond to different body parts. The ease distribution feature is all the element values ​​of a certain row in this matrix.

[0071] Preferably, the other sizes include sizes one size smaller than the baseline size and / or sizes one size larger than the baseline size. When none of the available sizes are one size smaller than the baseline size, only the larger size is selected as the comparison size; when none of the available sizes are one size larger than the baseline size, only the smaller size is selected as the comparison size; when the baseline size is the only size among all available sizes, only the baseline size is used to generate the virtual try-on effect image.

[0072] In an optional embodiment, the pattern templates of the first size, the suitable size, and the second size after non-rigid deformation processing are respectively superimposed onto the target user image, including: The first size, the suitable size, and the deformed pattern template of the second size are respectively fused with the target user image to generate a try-on effect image of one size smaller, a try-on effect image of one size suitable, and a try-on effect image of one size larger. The fitting effect images of the smaller size, the fitting effect images of the fit size, and the fitting effect images of the larger size are displayed side by side on the same display interface in ascending order of size.

[0073] In this embodiment, the non-rigid deformation processing refers to stretching, shrinking or twisting the local contour of the template to different degrees according to the looseness distribution characteristics without changing the overall topology of the template.

[0074] As another embodiment of the present invention, the non-rigid deformation treatment of the template includes: Extract the two-dimensional outline of the template; Multiple deformation control points are established on the outline of the two-dimensional plate. Calculate the displacement vector of each deformation control point based on the slack distribution characteristics corresponding to the deformation control point; Based on all the displacement vectors, the two-dimensional plate outline is interpolated and deformed to obtain the deformed plate template.

[0075] The deformation control points refer to feature points used to control the local deformation of the template. Each control point is distributed on each contour segment of the two-dimensional template outline. In this embodiment, control points are selected at uniform intervals according to arc length for each contour segment, with 3 to 7 points selected for each segment. The number of control points is positively correlated with the complexity of the template. Each control point is associated with a location label, which is used to index the control point with the corresponding location in the looseness distribution feature.

[0076] Optionally, the interpolation deformation can be implemented using a linear interpolation method, including: taking the contour segment between two adjacent deformation control points as the basic unit, for any non-control point, linearly weighting the displacement of the two control points according to the proportion of the cumulative arc length of the point along the contour segment from the starting control point to the total arc length of the contour segment, and calculating the displacement of the non-control point; moving each non-control point along its normal direction by the corresponding displacement, thus obtaining the deformed template.

[0077] For example, the displacement vector of each deformation control point is calculated using the following formula. It should be noted that this calculation method is only one possible method and does not affect the implementation of the basic scheme above:

[0078] in, This represents the displacement vector of each deformation control point. This represents the weighted ease difference of size i at location m in the ease distribution characteristics, where α represents the global ease adjustment coefficient. The unit normal vector representing the j-th control point on part m, which is perpendicular to the tangent of the pattern outline and points outward from the pattern. The fabric elongation coefficient represents the part m. The normalization factor for size i is represented by the size number, j represents the control point number, and m represents the part number.

[0079] In the formula, the global ease adjustment coefficient α is automatically determined by the remote processing platform based on the ease adjustment database. Specifically, the database stores historical users' body shape characteristics and corresponding purchase size information. The remote processing platform retrieves historical records with similar body shapes to the current user, calculates the average of their actual wearing ease values, and divides this average by a preset benchmark ease standard deviation. The resulting quotient is used as the global ease adjustment coefficient α. When the number of similar historical records is insufficient, α can obtain a default value from a preset category-ease benchmark table based on the category of the target garment. The fabric stretch coefficient... The elastic modulus is determined based on the fabric composition of the target garment. Specifically, the system has a pre-set fabric parameter library that stores the elastic modulus ranges corresponding to different fabric types. For example, the elastic modulus range for pure cotton plain weave fabric is 80-120 MPa, for pure cotton knitted fabric it is 40-60 MPa, for polyester plain weave fabric it is 150-200 MPa, for spandex blended (content 5%-20%) knitted fabric it is 20-40 MPa, and for wool twill fabric it is 100-150 MPa. The fabric elongation coefficient... Calculated using the following formula: ,in Take the median elastic modulus of pure cotton plain weave fabric as 100 MPa. Take the median value of the elastic modulus range corresponding to the target garment fabric. If fabric composition information cannot be obtained... Set to 1. Grade difference normalization factor. The size difference between adjacent sizes in the feature vector of the target garment pattern is determined and can be calculated using the following formula: ,in Let be the size difference between the i-th size and the baseline size at the same location. This represents the standard size difference between adjacent sizes of the garment. It should be noted that each parameter can be acquired, stored, and updated independently. Adjusting any parameter does not affect the physical meaning or basis for the values ​​of other parameters. Furthermore, each parameter's value has a clear physical source or preset basis, and the units are consistent. When a parameter is not yet available, its default value can be used. =1, λm=1, μ(i)=1), the formula degenerates into The system can still function normally. Therefore, the displacement vector calculation formula, while ensuring dimensional consistency and physical interpretability, can still achieve systematic adaptation to differences in body shape, garment structure shape preservation, fabric characteristics, cross-brand sizing systems, and user personalized preferences, thereby effectively improving the realism of virtual try-on images and the accuracy of size recommendation results.

[0080] As another embodiment of the present invention, before using the suitable size as a reference size and selecting at least one other size different from the suitable size, the method further includes: Extract key points from the target user image, including at least the neck center point, left and right acromion points, left and right axillary points, and left and right anterior superior iliac spine points; Based on the key points, the template is registered to the target user image.

[0081] S8. Based on the virtual try-on effect image, output the size recommendation conclusion of the target clothing to the user through the dual-screen interactive terminal.

[0082] This invention, by using the virtual try-on effect image, outputs the size recommendation conclusion of the target garment to the user through the dual-screen interactive terminal. It can present the differences of each size in terms of fit, looseness and silhouette to the user in a quantitative way, so that the user can obtain a clear size recommendation conclusion through the combination of visual comparison and data verification.

[0083] As an embodiment of the present invention, based on the virtual try-on effect image, the size recommendation conclusion of the target clothing is output to the user through the dual-screen interactive terminal, including: Extract the visual distance data between the outline of the pattern template and the outline of the user's body surface in the virtual try-on effect image; Based on the visual spacing data, and combined with the ease values ​​corresponding to each size of the target garment, fit evaluation information and size recommendation ranking results for each size are generated. The first display screen shows virtual try-on images of the user in various sizes. The second display screen shows the ease value and fit evaluation information corresponding to each size, and the size recommendation ranking results are displayed on the first display screen or the second display screen.

[0084] The visual spacing data refers to the distance between the outline of the pattern template (after non-rigid deformation processing) and the user's body contour line at various points in the virtual try-on effect image. Using the user's body contour line as a reference, the value is positive when the pattern template outline is outside it, negative when it is inside it, and zero when they overlap. The fit evaluation information includes fit labels for each size at various points and a comprehensive score for each size. The size recommendation ranking result is generated by sorting each size from highest to lowest comprehensive score, including the ranking and recommendation identifier for each size, with the size with the highest comprehensive score marked as the recommended size.

[0085] In another embodiment of the present invention, the fit evaluation information is generated in the following manner: The visual spacing data of each size at each part is compared with a preset threshold range; When the visual distance data falls within the preset threshold range, fit evaluation information including body part fit is generated; When the visual spacing data is less than the lower limit of the preset threshold range, fit evaluation information including tightness of parts is generated; When the visual spacing data is greater than the upper limit of the preset threshold range, fit evaluation information including areas that are too loose is generated.

[0086] The threshold range is preset based on the category and / or part of the target garment and is pre-stored in the system memory. Different garment categories have different definitions of fit, and different parts have different sensitivities to changes in ease. Therefore, the upper and lower limits of the threshold range are pre-calibrated based on at least one of the following: the size series differences for each part of the garment as specified in the national standard "Clothing Size" (GB / T1335), anthropometric data, subjective evaluation results of comfort when different body types wear garments with different ease in clothing ergonomic experiments, and the brand's fit definition standard for a specific pattern.

[0087] For example, taking clothing as an example, the preset threshold range for the chest is -0.5cm to +1.5cm, the preset threshold range for the waist is -0.5cm to +1.0cm, the preset threshold range for the shoulders is -0.3cm to +0.8cm, and the preset threshold range for the hips is -1.0cm to +1.5cm; taking trousers as an example, the preset threshold range for the hips is -0.3cm to +0.8cm, and the preset threshold range for the waist is -0.3cm to +1.0cm. In use, the system identifies the category of the target clothing based on the clothing identification information and retrieves the threshold range corresponding to that category and body part from the system memory to generate the fit label.

[0088] See Figure 4The diagram shown is a schematic representation of a remote virtual fitting system based on dual-screen guidance and multi-size fitting, provided in an embodiment of the present invention.

[0089] The remote virtual fitting system 200 based on dual-screen guidance and multi-size fitting described in this invention can be installed in an electronic device. Depending on the functions implemented, the remote virtual fitting system includes a dual-screen interaction configuration module 201, an information acquisition module 202, an image acquisition module 203, an image processing module 204, a body shape feature extraction module 205, a pattern feature extraction module 206, a fitting comparison module 207, and a size recommendation module 208. The modules described in this invention can also be called units, referring to a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, stored in the memory of the electronic device.

[0090] In this embodiment of the invention, the functions of each module / unit are as follows: The dual-screen interaction configuration module 201 is used to configure a dual-screen interaction terminal in the fitting area. The dual-screen interaction terminal includes a first display screen and a second display screen arranged side by side. The information acquisition module 202 is used to acquire the basic body posture parameters of the user in the fitting area and the clothing identification information of the target clothing selected by the user; The image acquisition module 203 is used to perform the user's image acquisition operation on the dual-screen interactive terminal based on the basic body posture parameters, and obtain the user's image. The image processing module 204 is used to send the user image to a remote processing platform, and perform image normalization processing on the user image in the remote processing platform to obtain the target user image; The body shape feature extraction module 205 is used to extract the key body shape circumference features of the user based on the target user image; The pattern feature extraction module 206 is used to retrieve the pattern feature vector of the target garment based on the garment identification information; The fitting comparison module 207 is used to construct the ease mapping relationship between the pattern feature vector and the key body shape circumference feature, and according to the ease mapping relationship, the pattern template of the target garment in at least two sizes is superimposed on the target user image to generate a virtual fitting effect image of the user in at least two sizes. The size recommendation module 208 is used to output the size recommendation conclusion of the target clothing to the user through the dual-screen interactive terminal based on the virtual try-on effect image.

[0091] In detail, the modules in the remote virtual fitting system 200 based on dual-screen guidance and multi-size fitting described in this embodiment of the invention employ the same methods as described above. Figure 1 The method used here is the same as the remote virtual fitting method based on dual-screen guidance and multi-size fitting described above, and can produce the same technical effect, so it will not be repeated here.

[0092] In one embodiment, a computer device is provided, which may be a server or a client, and its internal structure diagram may be as follows: Figure 5 As shown. The computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with external clients via a network connection. When the computer program is executed by the processor, it implements functions or steps on the server or client side of a remote virtual fitting method based on dual-screen guidance and multi-size fitting.

[0093] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps: S1. A dual-screen interactive terminal is configured in the fitting area, the dual-screen interactive terminal including a first display screen and a second display screen arranged side by side; S2. Obtain the basic body posture parameters of the user in the fitting area and the clothing identification information of the target clothing selected by the user; S3. Based on the basic body posture parameters, perform the user's image acquisition operation on the dual-screen interactive terminal to obtain the user's image; S4. The user image is sent to a remote processing platform, where the user image is processed for image normalization to obtain the target user image. S5. Based on the target user image, extract the user's key body circumference features; S6. Based on the clothing identification information, retrieve the pattern feature vector of the target clothing; S7. Construct the ease mapping relationship between the pattern feature vector and the key body shape circumference feature. Based on the ease mapping relationship, overlay the pattern template of the target garment in at least two sizes onto the target user image to generate a virtual try-on effect image of the user in at least two sizes. S8. Based on the virtual try-on effect image, output the size recommendation conclusion of the target clothing to the user through the dual-screen interactive terminal.

[0094] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor: S1. A dual-screen interactive terminal is configured in the fitting area, the dual-screen interactive terminal including a first display screen and a second display screen arranged side by side; S2. Obtain the basic body posture parameters of the user in the fitting area and the clothing identification information of the target clothing selected by the user; S3. Based on the basic body posture parameters, perform the user's image acquisition operation on the dual-screen interactive terminal to obtain the user's image; S4. The user image is sent to a remote processing platform, where the user image is processed for image normalization to obtain the target user image. S5. Based on the target user image, extract the user's key body circumference features; S6. Based on the clothing identification information, retrieve the pattern feature vector of the target clothing; S7. Construct the ease mapping relationship between the pattern feature vector and the key body shape circumference feature. Based on the ease mapping relationship, overlay the pattern template of the target garment in at least two sizes onto the target user image to generate a virtual try-on effect image of the user in at least two sizes. S8. Based on the virtual try-on effect image, output the size recommendation conclusion of the target clothing to the user through the dual-screen interactive terminal.

[0095] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or computer device described above can be referred to the relevant descriptions on the server side and client side in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0096] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Furthermore, any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory.

[0097] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above.

[0098] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0099] Finally, it should be noted that in the above embodiments, each embodiment can be combined with each other or independent. Deleting any one of them will not affect the technical implementation of other embodiments. The above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A remote virtual fitting method based on dual-screen guidance and multi-size fitting, characterized in that, The method includes: A dual-screen interactive terminal is configured in the fitting area, the dual-screen interactive terminal including a first display screen and a second display screen arranged side by side; Obtain the user's basic body posture parameters and the clothing identification information of the target garment selected by the user within the fitting area; Based on the basic body posture parameters, the user's image acquisition operation is performed on the dual-screen interactive terminal to obtain the user's image; The user image is sent to a remote processing platform, where image normalization processing is performed to obtain the target user image. Based on the target user image, extract the user's key body circumference features; Based on the clothing identification information, retrieve the pattern feature vector of the target clothing; Construct a looseness mapping relationship between the pattern feature vector and the key body shape circumference features. Based on the looseness mapping relationship, overlay the pattern template of the target garment in at least two sizes onto the target user image to generate a virtual try-on effect image of the user in at least two sizes. Based on the virtual try-on effect image, the size recommendation conclusion of the target clothing is output to the user through the dual-screen interactive terminal.

2. The remote virtual fitting method based on dual-screen guidance and multi-size fitting as described in claim 1, characterized in that, Based on the aforementioned basic body posture parameters, the user's image acquisition operation is performed on the dual-screen interactive terminal to obtain the user's image, including: Based on the height data in the basic body posture parameters, the lens focal length and framing range of the image acquisition device connected to the dual-screen interactive terminal are adjusted so that the user's full-body image falls within the preset framing frame of the image acquisition device. Once the user's full-body image falls within the preset viewfinder, the user's posture guidance information is displayed on the first display screen. The second display screen shows the real-time view captured by the image acquisition device, and the real-time view is superimposed with a head alignment line and a shoulder horizontal line. The system receives a user shooting command sent by a remote control connected to the dual-screen interactive terminal, and responds to the user shooting command by controlling the image acquisition device to take a picture and obtain a user image.

3. The remote virtual fitting method based on dual-screen guidance and multi-size fitting as described in claim 2, characterized in that, The posture guidance information includes a standard posture diagram and a framing prompt box. The standard posture diagram shows the user's target standing posture, and the framing prompt box is displayed superimposed on the standard posture diagram in the form of a dashed box to indicate the target standing area to the user. The head alignment line is a horizontal reference line used to constrain the user's head height position. The shoulder horizontal line is a horizontal reference line used to constrain the user's shoulder horizontal posture.

4. The remote virtual fitting method based on dual-screen guidance and multi-size fitting as described in claim 1, characterized in that, The user image is subjected to image normalization processing in the remote processing platform to obtain the target user image, including: The user image is subjected to quality pre-inspection processing in the remote processing platform to obtain a pre-inspection qualified user image. The quality pre-inspection processing includes at least image sharpness verification, human image integrity verification and posture angle verification. The pre-inspected and qualified user images are preprocessed to obtain the target user image. The preprocessing operations include at least background removal, artifact removal, and image quality correction.

5. The remote virtual fitting method based on dual-screen guidance and multi-size fitting as described in claim 1, characterized in that, Constructing a loose mapping relationship between the pattern feature vector and the key body shape circumference features includes: Obtain the circumference values ​​corresponding to each part in the key body shape circumference features, and the pattern size corresponding to each part in each size in the pattern feature vector, wherein the parts include the chest, waist, hips and shoulders; Calculate the difference between the pattern size and the circumference value for each part of each size to obtain the initial ease difference for each part of each size; The initial slack difference is adjusted by weighting the slack weights of each part to obtain the weighted slack difference. All the weighted ease difference values ​​for each size are arranged according to the location dimension to form an ease distribution matrix, and the ease distribution matrix is ​​used as the ease mapping relationship.

6. The remote virtual fitting method based on dual-screen guidance and multi-size fitting as described in claim 5, characterized in that, The preset ease weighting coefficients for each part are dynamically determined based on the category of the target garment: When the target garment is a top, the ease weighting coefficient corresponding to the chest is greater than the ease weighting coefficient corresponding to the waist, and the ease weighting coefficient corresponding to the waist is greater than the ease weighting coefficient corresponding to the hips. When the target garment is trousers, the ease weighting coefficient corresponding to the hips is greater than the ease weighting coefficient corresponding to the waist, and the ease weighting coefficient corresponding to the waist is greater than the ease weighting coefficient corresponding to the legs. When the target garment is a skirt, the ease weighting coefficient corresponding to the waist is greater than or equal to the ease weighting coefficient corresponding to the hips.

7. The remote virtual fitting method based on dual-screen guidance and multi-size fitting as described in claim 1, characterized in that, Based on the ease mapping relationship, the pattern templates of the target garment in at least two sizes are overlaid onto the target user image to generate a virtual try-on effect image of the user in at least two sizes, including: Based on the ease mapping relationship, the suitable size is identified from all available sizes of the target garment; Use the suitable size as the baseline size, and select at least one other size that is different from the suitable size; Based on the ease mapping relationship, the ease distribution characteristics of the baseline size and the other sizes at various locations are determined respectively; Based on the looseness distribution characteristics, non-rigid deformation processing is performed on the pattern templates corresponding to the base size and the other sizes, respectively; The templates of the baseline size and other sizes after non-rigid deformation processing are superimposed on the target user image to generate virtual try-on effect images of the user in each size.

8. The remote virtual fitting method based on dual-screen guidance and multi-size fitting as described in claim 7, characterized in that, The non-rigid deformation processing of the template includes: Extract the two-dimensional outline of the template; Multiple deformation control points are established on the outline of the two-dimensional plate. Calculate the displacement vector of each deformation control point based on the slack distribution characteristics corresponding to the deformation control point; Based on all the displacement vectors, the two-dimensional plate outline is interpolated and deformed to obtain the deformed plate template.

9. The remote virtual fitting method based on dual-screen guidance and multi-size fitting as described in claim 1, characterized in that, Based on the virtual try-on effect image, the dual-screen interactive terminal outputs a size recommendation conclusion for the target clothing to the user, including: Extract the visual distance data between the outline of the pattern template and the outline of the user's body surface in the virtual try-on effect image; Based on the visual spacing data, and combined with the ease values ​​corresponding to each size of the target garment, fit evaluation information and size recommendation ranking results for each size are generated. The first display screen shows virtual try-on images of the user in various sizes. The second display screen shows the ease value and fit evaluation information corresponding to each size, and the size recommendation ranking results are displayed on the first display screen or the second display screen.

10. A remote virtual fitting system based on dual-screen guidance and multi-size fitting, employing the remote virtual fitting method based on dual-screen guidance and multi-size fitting as described in any one of claims 1-9, characterized in that, The system includes: A dual-screen interaction configuration module is used to configure a dual-screen interaction terminal in the fitting area. The dual-screen interaction terminal includes a first display screen and a second display screen arranged side by side. The information acquisition module is used to acquire the basic body posture parameters of the user in the fitting area and the clothing identification information of the target clothing selected by the user; The image acquisition module is used to perform the user's image acquisition operation on the dual-screen interactive terminal based on the basic body posture parameters, and obtain the user's image; The image processing module is used to send the user image to a remote processing platform, and perform image normalization processing on the user image in the remote processing platform to obtain the target user image; The body shape feature extraction module is used to extract the key body shape circumference features of the user based on the target user image; The pattern feature extraction module is used to retrieve the pattern feature vector of the target garment based on the garment identification information; The fitting comparison module is used to construct the ease mapping relationship between the pattern feature vector and the key body shape circumference features. Based on the ease mapping relationship, the pattern template of the target garment in at least two sizes is superimposed on the target user image to generate a virtual fitting effect image of the user in at least two sizes. The size recommendation module is used to output a size recommendation conclusion for the target garment to the user through the dual-screen interactive terminal based on the virtual try-on effect image.

Images