Research On The Key Technology Of Virtual Fitting System Based On Somatosensory Interaction

In recent ten years, with the rapid development of virtual reality technology and sensor technology, based on somatosensory interaction and enhance the reality of virtual fitting system has been widely studied. Compared to the traditional stores and shop fitting model, the virtual fitting method has obvious advantages. For shoppers, it brings a relaxed and immersive virtual fitting way, and a novel, interesting new shopping experience. What’s more, shoppers get twice the result with half the effort of the fitting effect. In the circle of friends, the mutual exchange of virtual fitting photo and the virtual fitting experience even cited a fashion trend. For businesses, virtual fitting occupied small space, saving store logistics and human services, greatly reduce operating costs. The existing problems of virtual fitting system are mainly concentrated in three areas. Firstly, the clothing stereoscopic effect is not ideal. It is difficult to change the zoom coefficient automatically according to the user’s body characteristics. Secondly, only the appearance virtual clothing is displayed. Meanwhile, it is very important for the experience of user to show the characteristics of fabric, texture and draping. Last but not least, the interactive immersion experience of virtual fitting is very poor. The stippling operation of virtual fitting is easy to be tired. And the voice control mode is natural and simple. The above three aspects makes the experience effect of current virtual fitting system is far from ideal.After analyzing and comparing the domestic and foreign virtual fitting technology. This paper researchs the the implementation methods of key technologies of virtual fitting system. By using 3ds Max and Clo3 d software programming, we realize the human body model and the garment modeling. At the same time, a human gesture input based deformation algorithm is applied for deformation processing. Those processing methods make the physical characteristics of the virtual clothing showed more truly. Thus users can obtain a more realistic experience. Image acquisition device is also used to capture real image, and the clothing image transfer algorithm is also adopted to make the virtual clothes overlap the real image. With feature extraction technology, we obtain the joint position of the human body. At last, not only the scaling of virtual clothing is calculated, but also the virtual clothes is tightly combined with the real image.Besides, in this paper, we use the speech recognition technology based on Kinect speech recognition instead of the gesture control way. And the technologies of noise reduction, automatic gain control and an echo eliminating technology is adopted to improve the accuracy of virtual fitting.The enhanced virtual fitting system by our key technologies makes the garments deformation effect more realistic. Finally, the user’s experience of service(EoS) is optimized greatly.