A virtual try-on system in augmented reality using RGB-D cameras for footwear personalization

Abstract

This paper presents a system for design evaluation of footwear using commercial depth-sensing technologies. In a mixed reality environment, the system allows users to virtually try on 3D shoe models in a live video stream. A two-stage object tracking algorithm was developed to correctly align shoe models to moving feet during the try-on process. Color markers on the user's foot enabled markerless tracking. Tracking was driven by an iterative closest point (ICP) algorithm that superimposed the captured depth data and predefined reference foot models. Test data showed that the two-stage approach resulted in increased positional accuracy compared with tracking using only surface registration. Trimming the reference model using the instant view angle increased the computational efficiency of the ICP algorithm. The proposed virtual try-on function is an effective tool for realizing human-centered design. This study also demonstrated a new application of RGB-D cameras to product design.