3D Scanning of Clothing Using a RGB-D Sensor with Application in a Virtual Dressing Room

This paper presents an approach for creating digital clothing with application in a virtual dressing room. The clothes are made digital by scanning real clothes using a RGB-D sensor. While creating digital clothing using specialized programs, e.g. Marvelous Designer 2, is a time consuming process, the 3D scanning process is relatively fast. The model is acquired via registration of different views while rotating on a platform; a complete rotation is performed in approximately 60 s, however, we have achieved satisfactory results even with rotations of 15 s. The surface is textured using the RGB images acquired during the scanning process, and the color of the occluded parts is approximated via K-Nearest Neighbors (KNN). We show early results of creating digital clothing of high quality by scanning real clothes.