A modular approach for cloth modelling in virtual fitting room

摘要

The application is an augmented virtual reality application where the user is recognized at close range and can virtually try clothes. The unique point about this application is the fact that there'll be image processing and computer vision used to determine size of shirt and waist and so on of a person using the Kinect V2 perceptual computing kit rather than just allowing the user to try on virtual clothes. These clothes can be showcased online among the friends or can be directly used for printing and manufacturing the shirt using CO2 laser from a robotic arm which is out of scope of this paper. The modelling of cloth which is supposed to be augmented on the virtual human body is being designed with competent geometric algorithms rather than just relying on Gaming or Motion Control Frameworks. The results are tested on latest Kinect V2 sensor which provides user a compelling experience to opt for. The paper initially overviews the major technical components for complete virtual try-on system, followed by elucidating several key challenges such as calibration of Kinect and estimation of measurements for individual subjects like outfits, etc. Eventually we discuss key details of the implementation using Kinect V2. Quality of these steps is the key to achieve seamless try-on experience for users.