http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
라이다 기반의 포인트 클라우드를 이용한 ADOP 성능 개선
김혜지(Hye-Ji Kim),안지수(Ji-Su An),이정언(Jeong-Eon Lee),최소미(So-Mi Choi),한현수(Hyeon-Soo Han),황성수(Sung-Soo Hwang) 대한전자공학회 2023 대한전자공학회 학술대회 Vol.2023 No.6
Replicating real-world spaces in virtual environments is becoming increasingly important. However, the current methods for constructing virtual spaces have the drawback of increased costs and time as the quality of the space improves. This paper addresses this issue by introducing ADOP, a neural rendering technique that utilizes point cloud as an additional input data. The paper also proposes a technique to generate more compact and precise point clouds by fusing LiDAR and stereo cameras, including color and normal vector information. The resulting point cloud is then used as input for ADOP to enhance its rendering performance. Experiments have confirmed that using point clouds obtained with LiDAR results in improved rendering performance in featureless indoor spaces.