The paper aims to transfer the item image of a given clothing product to a corresponding area of the user image. Existing classical methods suffer from unconstrained deformation…
Abstract
Purpose
The paper aims to transfer the item image of a given clothing product to a corresponding area of the user image. Existing classical methods suffer from unconstrained deformation of clothing and occlusion caused by hair or poses, which leads to loss of details in the try-on results. In this paper, the authors present a details-oriented virtual try-on network (DO-VTON), which allows synthesizing high-fidelity try-on images with preserved characteristics of target clothing.
Design/methodology/approach
The proposed try-on network consists of three modules. The fashion parsing module (FPM) is designed to generate the parsing map of a reference person image. The geometric matching module (GMM) warps the input clothing and matches it with the torso area of the reference person guided by the parsing map. The try-on module (TOM) generates the final try-on image. In both FPM and TOM, attention mechanism is introduced to obtain sufficient features, which enhances the performance of characteristics preservation. In GMM, a two-stage coarse-to-fine training strategy with a grid regularization loss (GR loss) is employed to optimize the clothing warping.
Findings
In this paper, the authors propose a three-stage image-based virtual try-on network, DO-VTON, that aims to generate realistic try-on images with extensive characteristics preserved.
Research limitations/implications
The authors’ proposed algorithm can provide a promising tool for image based virtual try-on.
Practical implications
The authors’ proposed method is a technology for consumers to purchase favored clothes online and to reduce the return rate in e-commerce.
Originality/value
Therefore, the authors’ proposed algorithm can provide a promising tool for image based virtual try-on.
Details
Keywords
PengPeng Hu, Duan Li, Ge Wu, Taku Komura, Dongliang Zhang and Yueqi Zhong
Currently, a common method of reconstructing mannequin is based on the body measurements or body features, which only preserve the body size lacking of the accurate body geometric…
Abstract
Purpose
Currently, a common method of reconstructing mannequin is based on the body measurements or body features, which only preserve the body size lacking of the accurate body geometric shape information. However, the same human body measurement does not equal to the same body shape. This may result in an unfit garment for the target human body. The purpose of this paper is to propose a novel scanning-based pipeline to reconstruct the personalized mannequin, which preserves both body size and body shape information.
Design/methodology/approach
The authors first capture the body of a subject via 3D scanning, and a statistical body model is fit to the scanned data. This results in a skinned articulated model of the subject. The scanned body is then adjusted to be pose-symmetric via linear blending skinning. The mannequin part is then extracted. Finally, a slice-based method is proposed to generate a shape-symmetric 3D mannequin.
Findings
A personalized 3D mannequin can be reconstructed from the scanned body. Compared to conventional methods, the method can preserve both the size and shape of the original scanned body. The reconstructed mannequin can be imported directly into the apparel CAD software. The proposed method provides a step for digitizing the apparel manufacturing.
Originality/value
Compared to the conventional methods, the main advantage of the authors’ system is that the authors can preserve both size and geometry of the original scanned body. The main contributions of this paper are as follows: decompose the process of the mannequin reconstruction into pose symmetry and shape symmetry; propose a novel scanning-based pipeline to reconstruct a 3D personalized mannequin; and present a slice-based method for the symmetrization of the 3D mesh.
Details
Keywords
Yueqi Zhong, Duan Li, Ge Wu and PengPeng Hu
The automatic body measurement is the key of tailoring, mass customization and fit/ease evaluation. The major challenges include finding the landmarks and extracting the sizes…
Abstract
Purpose
The automatic body measurement is the key of tailoring, mass customization and fit/ease evaluation. The major challenges include finding the landmarks and extracting the sizes accurately. The purpose of this paper is to propose a new method of body measurement based on the loop structure.
Design/methodology/approach
The scanned human model is sliced equally to layers consist of various shapes of loops. The semantic feature analysis has been regarded as a problem of finding the points of interest (POI) and the loop of interest (LOI) according to the types of loop connections. Methods for determining the basic landmarks have been detailed.
Findings
The experimental results validate that the proposed methods can be used to locate the landmarks and to extract sizes on markless human scans robustly and efficiently.
Originality/value
With the method, the body measurement can be quickly performed with average errors around 0.5 cm. The results of segmentation, landmarking and body measurements also validate the robustness and efficiency of the proposed methods.
Details
Keywords
Ge Wu, Duan Li, Yueqi Zhong and PengPeng Hu
The calibration is a key but cumbersome process for 3D body scanning using multiple depth cameras. The purpose of this paper is to simplify the calibration process by introducing…
Abstract
Purpose
The calibration is a key but cumbersome process for 3D body scanning using multiple depth cameras. The purpose of this paper is to simplify the calibration process by introducing a new method to calibrate the extrinsic parameters of multiple depth cameras simultaneously.
Design/methodology/approach
An improved method is introduced to enhance the accuracy based on the virtual checkerboards. Laplace coordinates are employed for a point-to-point adjustment to increase the accuracy of scanned data. A system with eight depth cameras is developed for full-body scanning, and the performance of this system is verified by actual results.
Findings
The agreement of measurements between scanned human bodies and the real subjects demonstrates the accuracy of the proposed method. The entire calibration process is automatic.
Originality/value
A complete algorithm for a full human body scanning system is introduced in this paper. This is the first publically study on the refinement and the point-by-point adjustment based on the virtual checkerboards toward the scanning accuracy enhancement.
Details
Keywords
PengPeng Hu, Taku Komura, Duan Li, Ge Wu and Yueqi Zhong
The purpose of this paper is to present a novel framework of reconstructing the 3D textile model with synthesized texture.
Abstract
Purpose
The purpose of this paper is to present a novel framework of reconstructing the 3D textile model with synthesized texture.
Design/methodology/approach
First, a pipeline of 3D textile reconstruction based on KinectFusion is proposed to obtain a better 3D model. Second, “DeepTextures” method is applied to generate new textures for various three-dimensional textile models.
Findings
Experimental results show that the proposed method can conveniently reconstruct a three-dimensional textile model with synthesized texture.
Originality/value
A novel pipeline is designed to obtain 3D high-quality textile models based on KinectFusion. The accuracy and robustness of KinectFusion are improved via a turntable. To the best of the authors’ knowledge, this is the first paper to explore the synthesized textile texture for the 3D textile model. This is not only simply mapping the texture onto the 3D model, but also exploring the application of artificial intelligence in the field of textile.
Details
Keywords
Hongyan Liu, Yueqi Zhong and Shanyuan Wang
Reusability of geometric model is of essential importance in garment simulation since in reality, the same garment dressed on different human bodies will exhibit different draping…
Abstract
Purpose
Reusability of geometric model is of essential importance in garment simulation since in reality, the same garment dressed on different human bodies will exhibit different draping results. Due to the multiplicity of a garment which can be dressed on various human bodies, it is necessary to reuse a given geometrical garment model in the cyberspace, the purpose of this paper is to explore this. This requires the technique of converting a geometrical model to a physically‐based model to predict the dressing result.
Design/methodology/approach
In this paper, a 3D garment model obtained from range data scanning is reconstructed as a mass‐spring system driven by the first order backward Euler integrator. The penetration of dressing onto the given avatars is compensated through an AABB accelerated collision detection and response scheme.
Findings
This paper implements efficient techniques resolving intersection for cloth simulation. Two stages are selected for handling the collision. The first stage is trying to prevent collision from happening in the first stage, and the second stage is to deal with actual intersections. This two‐stage approach eliminates almost all collision in the first stage, leaving only a few collisions for the second stage. Results obtained with this approach, in the case of dressing garments on different avatars are presented and validate the conversion method from geometric model to deformable one.
Originality/value
The experimental results validate this method as a useful and effective approach to exhibit draping behavior on various human body shapes accordingly.
Details
Keywords
Hongyan Liu, Yueqi Zhong and Shanyuan Wang
The purpose of this paper is to investigate the draping behavior of a scanned garment model in irregular and regular mesh topology. 3D garment with high fidelity can be obtained…
Abstract
Purpose
The purpose of this paper is to investigate the draping behavior of a scanned garment model in irregular and regular mesh topology. 3D garment with high fidelity can be obtained via range data scanning. The original output of a body scanner is an unorganized points cloud. In this paper, the geometrical surface of 3D garment is reconstructed through a series of treatments.
Design/methodology/approach
The primary target of this work is to investigate the dynamic behavior of the corresponding physical model transferred from different mesh topologies. A mass‐spring model is constructed for both regular and irregular meshes. The performance under various integration methods is evaluated.
Findings
Experimental results reveal the procedure of regularization is suitable for the integrators that are sensitive to the physically‐based simulation of scanned garments.
Originality/value
The geometrical surface of 3D garment is reconstructed through a series of treatments solving the problem of points cloud data for high definition 3D data.
Details
Keywords
Yueqi Zhong and Bugao Xu
This paper presents methods and algorithms to automatically segment and measure the human body.
Abstract
Purpose
This paper presents methods and algorithms to automatically segment and measure the human body.
Design/methodology/approach
In the segmentation procedure, two different methods are designed to find the crotch point for the situation of non‐contacted thigh and contacted thigh, respectively. Three different methods: minimum distance algorithm, minimum inclination angle algorithm, and directional neighbor identification algorithm are introduced to search the branching points or triangle. In the body measurement procedure, a pre‐sorted circling method is designed for circumference measurement, and the basic principle of landmark acquisition has been discussed. These techniques are validated via testing over different type of scanned model.
Findings
The results of automatic segmentation and body measurement have verified that our methods are efficient and versatile in processing different type of scanned body.
Research limitations/implications
The accurate and automatic locating of wrist, ankle and knees contour can be more difficult than it appears to be.
Practical implications
The main usage of scanned body in our research is for 3D garment try‐on.
Originality/value
This paper introduces the methods for crotch identification, and the methods including minimum distance algorithm, minimum inclination angle algorithm, and directional neighbor identification algorithm for human body segmentation. It also explains the fundamental measuring techniques, and outlines the results of using these techniques in segmentation and measurement.
Details
Keywords
Zhaoli Wang, Yueqi Zhong and Shanyan Wang
The purpose of this paper is to show how shape analysis and quantitative characterization of fiber cross sections, with the aid of image analysis techniques, provide a quick…
Abstract
Purpose
The purpose of this paper is to show how shape analysis and quantitative characterization of fiber cross sections, with the aid of image analysis techniques, provide a quick, powerful approach to automated profiled fiber identification.
Design/methodology/approach
In this paper, an effective method of cross‐sectional shape characterization for profiled fiber identification is reported with extraction of the distance fluctuation curve of fiber cross‐sectional boundary to the centroid. By calculating their cross‐correlations using signal processing techniques, the authors tackle the problem of calibrating the starting points of fiber objects orientated arbitrarily in image successfully, which are difficult to deal with by means of image processing, to finish the normalization of distance fluctuation curves. For two fiber cross‐sections, the similarity degree of their boundary fluctuation curves normalized can effectively reflect the similarity degree of themselves.
Findings
Based on this, the method presented extracts the curves of all fiber cross‐sections in one sample, compares the similarity degrees between each other, and creates clusters to identify profiled fiber.
Originality/value
Experimental results validate that this curve can effectively characterize profiled fiber cross‐sectional contour for profiled fiber identification and the normalization method is feasible.
Details
Keywords
Hongyan Liu, Yueqi Zhong and Shanyuan Wang
In order to obtain visually appealing simulation results with smooth cloth surfaces, high resolution meshes are required. Since cloth simulation with high resolution meshes is…
Abstract
Purpose
In order to obtain visually appealing simulation results with smooth cloth surfaces, high resolution meshes are required. Since cloth simulation with high resolution meshes is very time consuming, subdivision of a reusable model is preferred. The purpose of this paper is to adopt an approach for the subdivision of a reusable model, using different subdivision methods.
Design/methodology/approach
In order to obtain visually pleasing reusable garment model, the authors subdivide the model after the conversion to reusable garment model. Two meshes are employed, of which the coarse mesh is used to convert to deformable model while the subdivided mesh, i.e. the refined mesh, is for pleasing modeling results. The modified loop subdivision and modified butterfly subdivision scheme is adopted for the optimization of determining which is more suitable for the reusable model. The authors adopt an algorithm for resolving the collisions happened after the subdivision.
Findings
A 3D reusable garment model obtained from previous work was tested. The experimental results validate this method as a useful and effective approach to resolve collisions after the subdivision process.
Originality/value
The paper presents an approach for the subdivision of a reusable model, using different subdivision methods.