Abstract
Time-of-flight (TOF) camera images can be combined with data from other device sensors for enhanced depth perception and mapping capabilities. Effective fusion of data from TOF cameras with data from other types of sensors requires accurate calibration that can mitigate the lower range and resolution of TOF cameras. This disclosure describes techniques to perform efficient targetless recalibration to estimate the extrinsic transformation between a TOF camera and an IMU on a mobile device. The approach involves analyzing individual dTOF frames within a sequence of TOF and IMU data to identify three planar surfaces in the image, e.g., via iterative application of the RANSAC algorithm. The corner point at which the planes intersect can be determined by minimizing the unit vector consistency cost and corner centrality cost. The analysis can additionally extract the orientation of wall lines within each frame. Corner points and wall lines across multiple frames can be examined collectively to verify and recalibrate the 6DOF extrinsic. The techniques can be implemented to perform targetless calibration of TOF cameras within any device without needing image association. The techniques can be employed indoors in environments with low illumination or low textural detail.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Ji, Dinghuang; Jian, Yong-Dian; and Zhou, Verse, "Targetless Extrinsics Calibration of Time-of-Flight (TOF) Cameras on Mobile Devices", Technical Disclosure Commons, (November 20, 2025)
https://www.tdcommons.org/dpubs_series/8913