- > 연구성과 > 학술발표
공지사항
논문명 |
Single RGB Image to 3D Point Cloud Translation Based on Generative Adversarial Networks |
구분 |
국내발표 |
저자 |
Phuong Minh Chu, Seoungjae Cho, Kaisi Huang, Kyungeun Cho |
국내/국외 |
|
학술회의명 |
The International Conference on Big data, IoT, and Cloud Computing |
개최국가 |
대한민국 |
주관기관 |
|
게재일 |
2018.08 |
Three-dimensional (3D) point clouds are important for many applications such as 3D scene reconstruction. We usually obtain point clouds using laser scanners, which are expensive. This study proposes a novel approach to generate a 3D point cloud from a single red–green–blue (RGB) image via two steps. First, we use a generative adversarial network model to generate a depth image from a single RGB image. Second, a point cloud is created from the depth image based on an estimation method. The experiment results verify that the proposed generation method can provide a high-quality 3D point cloud from a single RGB image.
|
|