Research Progress in Non-Intrusive Three-Dimensional Reconstruction of Transparent Rigid Bodies

Chifai Pun, Rui Chen, Changping Hu, Jing Xu

Abstract:

The three-dimensional (3D) reconstruction technology has considerable application prospects in industrial fields
such as autopilot and metauniverse. The 3D reconstruction of transparent rigid bodies exhibits some challenges owing to
its complex imaging laws. This review examines the 3D reconstruction technology of transparent rigid bodies, focuses on
the methods based on optimized visual hull and deep learning, discusses the principle and different optimization methods
based on visual hull reconstruction, describes the RGB-D depth completion and multiview reconstruction based on deep
learning, presents the current dataset of transparent objects, and finally discusses the possible development direction of the
3D reconstruction technology.