Novel approach for fast structured light framework using deep learning

The research team led by Professor Jae-Sang Hyun from the Department of Mechanical Engineering (with first author Won-Hoe Kim, integrated course) conducted joint research with Purdue University and has developed an 'Novel approach for fast structured light framework using deep learning.' Their research findings were published in the prestigious journal 'Image and Vision Computing,' a leading publication in the field of computer vision. Unlike existing structured light methods that combine multiple images for reconstruction, this study utilizes deep learning to enable 3D reconstruction using only two images, maximizing speed. Additionally, training data necessary for framework development was generated and validated through a virtual environment, and an algorithm was developed to minimize errors, allowing for accurate and rapid 3D reconstruction.

The link: doi.org/10.1016/j.imavis.2024.105204