A learning based approach is proposed to calibrate the geometric distortion of a Time-of-Flight (ToF) camera. Our method is flexible as it requires only a ToF camera and a standard camera calibration chessboard. We treat the noise model of a ToF camera as a black box, and employ random forest to automatically learn the underlying unique noise model. The geometric property of the point-cloud can be effectively restored by the learned distortion model. The method can be used in a range of computer vision applications including e.g. hand pose estimation.
CITATION STYLE
Xu, C., & Li, C. (2018). A flexible method for time-of-flight camera calibration using random forest. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 11010 LNCS, pp. 207–218). Springer Verlag. https://doi.org/10.1007/978-3-030-04375-9_18
Mendeley helps you to discover research relevant for your work.