Design And Implementation Of Key Preprocessing Algorithms For 3D Virtual Fitting

With the continuous development of the Internet technology and hardware tech-nology,the emerging technology in computer graphics has been rapidly promoted and widely applied.The trend of 020 electronic commerce promotes the popularity of on-line shopping.Purchasing clothes online not only has a diversity of choosing clothes,but also can help customers save time and effort.It's also convenient for customers to compare prices in online stores.In recent years,researches on virtual fitting have been performed mostly based on the data of garment pieces through CAD(Computer Aided Design).However,three-dimensional garment models are more intuitive,integrated and easy to access and manage.In the process of fitting,it's often necessary to match some designated key points of garment model and human model,which makes the process not flexible enough.In addition,it's also an important concern for customers whether the garment is suitable or not.If virtual fitting can provide the results of size judgments,it will save much time customers spend on choosing sizes.This paper is intended to design and propose methods of automatic alignment,size judgment and penetration recovery in the process of virtual fitting,which applies to 3D garment models and 3D human models.Two models are basically aligned through three-dimensional geometric calculation and postural alteration of virtual humans.By comparing the volumes of garment model and the intercepted human model after align-ment,the results of size judgment are obtained.After the basic alignment,it's still possible that the garment penetrates into the human model,which is called penetration phenomena in this paper.Then,we detect the penetration phenomena and locate the points of the garment model penetrating the human model.Based on the penetration detection,we use mesh deformation to adjust the penetration points by imitating the adjustment process of dragging clothes.With the above steps,the garment model and human model obtained can be directly used for physical-based simulation.Experimen-tal results show that the method we propose can achieve good effects of alignment.We can give the results of size judgments for given garment models and human models.Besides,the penetrating phenomenon caused by the alignment process is recovered by mesh deformation,which can enhance the real experience for consumers during virtual fitting.The methods proposed in this paper can be applied to the actual process of vir-tual fitting,and have good versatility and practicality with good prospects of research and application.