3D scanning is the process of capturing, processing, and reconstruction of real-world objects within static and dynamic scenes to characterize shape and other appearance information. One type of object of particular interest in computer vision and graphics is the representation of the full human body. As technology is advancing, 3D body models are being used in fields such as healthcare, physical fitness, and entertainment. In the clothing industry, 3D body models have found applications in body measurements, posture analysis, and are often used for virtual try-ons. Traditional scanning methods require complex multi-view setups, specialized equipment, and/or high degrees of processing power. For widespread consumer end-use, there is a need for a platform that can conduct 3D scanning remotely, with a high degree of accuracy, using a portable device.
The focus of this research thrust is on 3D body scanning and extracting body measurements using depth cameras readily available on smartphone devices for virtual try-on applications. Today, consumers use often-unreliable sizing charts for online apparel purchases and return orders are often due to size-related challenges. The cost of return on the e-commerce industry is high with unsalvageable items discarded in landfills raising environmental concerns. Remote measurement tools that incorporate 3D scanning technology can help to optimize recommendations for the consumer, reducing waste and the associated environmental impacts.
The objective of this research is to develop techniques, algorithms and software for accurate and robust 3D body reconstruction using data from a single smartphone depth camera. Our approach incorporates the following key tasks:
- Automatic identification of key points in the body or a body part
- Optimization of deformations for fusion of depth input data
- Extraction of body measurements from the 3D model
