Abstract
We present a novel calibration method for a multi-view laser Doppler speed sensing (MLDSS) system. In contrast with the traditional method where only the laser geometry is independently calibrated, the proposed method simultaneously optimizes all the laser parameters and directly associates the parameters with a motion sensing model. By jointly considering the consistency among laser Doppler velocimetry, the laser geometry and a visual marker tracking system, the proposed calibration method further boosts the accuracy of MLDSS. We analyzed the factors influencing the precision, and quantitatively evaluated the efficiency of the proposed method on several data sets.
Highlights
Multiview laser Doppler speed sensing (MLDSS) [1] uses several one-dimensional speed measurements to recover the six-degree-of-freedom (6-DOF) motion of an arbitrary rigid body
It is important to mention that the precision of multi-view laser Doppler speed sensing (MLDSS) strongly relies on the sensing accuracy of the laser Doppler velocimeter (LDV), the system-layout parameters, and calibration precision
Knowing that matrix A can be uniquely determined by parameter set P, the 6-DOF speed of the target measured by the MLDSS system at frame k could be calculated by Xk ( P) = [ A( P)]−1 bk
Summary
Multiview laser Doppler speed sensing (MLDSS) [1] uses several one-dimensional speed measurements to recover the six-degree-of-freedom (6-DOF) motion of an arbitrary rigid body. Conventional contactless 6-DOF motion-sensing techniques, including computer-vision approaches, usually rely on the target structure [4,5] and texture [6,7], making it difficult to directly reapply them when the scenario changes. A similar calibration principle can be commonly found in the literature for the calibration of galvanoscopic laser systems using various types of sensors [13,14,15] It makes sense and achieves relatively good accuracy, but it has to be noted that LDV measurements and the system layout are not involved in the process. It is important to incorporate a system measurement model into the geometrical calibration, as implied in References [18,19] Based on this idea, we present a novel calibration technique for MLDSS. We validated the proposed method on several different datasets, and an obvious improvement in accuracy was confirmed
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have