Structured light-based 3D reconstruction in the radiation environment based on multi-frame iterative filtering and image feature-based point cloud registration

Abstract

In this paper, a novel 3D reconstruction method is proposed for radiation environments based on structured light technology. Firstly, throughout the principles of structured light-based 3D reconstruction, we see that the primary sources of error in radiation environments are the impacts of radiation noise and point-cloud registration errors. Then, considering the random nature of radiation noise, this paper proposes a multiframe iterative image-filtering algorithm that effectively and quickly removes radiation noise while preserving the phase information of the structured light. Additionally, a point-cloud initial registration method based on image features is introduced, which reduces the time required for initial matching and ensures accuracy. 3D reconstruction experiments were conducted at radiation dose rates of 60 Gy/h and 100 Gy/h. The results demonstrate that the proposed method maintains high modeling accuracy even in high-radiation environments, validating the reliability of the approach.