Sensor to segment calibration for magnetic and inertial sensor based motion capture systems

This study aims to propose a novel sensor to segment calibration method for magnetic and inertial measurement unit (M-IMU) based human motion capture systems. The calibration procedure is designed to be easily performed by the subject, which can calibrate the whole body segments with only three predefined postures. Firstly, the orientation measurements for each body segment are collected during calibration process. Secondly, a segmentation procedure based on a moving average window algorithm and double-threshold technique is introduced to recognize and segment the calibration postures automatically. To suppress the shaking during the calibration process, an intrinsic average algorithm is presented to smooth the orientation measurements. Finally, an iteration hand-eye calibration approach is utilized to retrieve the sensor to segment calibration matrices. The error due to the ungraded calibration postures can be reduced observably, and we also define an indicator to evaluate the quality of the performed calibration postures. The calibration procedure has been validated on several subjects using a wearable M-IMU based motion capture system. Experiment results show that the proposed calibration method can suppress the shaking disturbances effectively, and has good repeatability.