Calibration method for event-based structured light systems using frequency information

Abstract

Event-based structured light (ESL) systems combine the high temporal resolution and dynamic range of event cameras with the fast, low-latency projection of Digital Light Processing (DLP) projectors, offering significant potential for 3D reconstruction in challenging environments. However, both event cameras and DLP projectors do not directly output images, which poses challenges for the calibration of ESL systems and limits their application. This paper proposes a novel calibration method for ESL systems using frequency information, which relies solely on event stream data without requiring image output from the event camera. We first introduce a frequency-based reconstruction algorithm that uses DLP projectors to project strobe-stripe light fields and reconstructs stripe images by processing the event stream using frequency information. This method addresses the challenges of event camera imaging and establishes the mapping relationship between DLP and camera pixels. Next, we use pixel correlation to back-project the event camera’s “imaging map” onto the DLP image plane, solving the DLP imaging issue. Finally, we calibrate the ESL system using the “imaging maps” from both the event camera and DLP. Experimental results show that our method achieves subpixel-level accuracy of 0.078 and 0.069 pixels in calibrating ESL system and event camera, respectively. Moreover, the ESL system calibrated by our method demonstrates a low root mean square error of 0.223 mm in 3D reconstruction at a distance of 0.5 meters, validating the accuracy of the proposed calibration technique.