Calibration of a Time-of-Flight Camera Using Probability and Reflection Analysis

Abstract

This article introduces a general calibration approach for typical time-of-flight (ToF) depth cameras. The 3-D range detection needs in various industry areas, such as machine vision and robotics, are increasing. Recent advancements in microelectronics have enabled the development of more powerful ToF cameras, which provide high-quality 3-D information with high frame rates. ToF camera outputs contain various error types, most of which could be classified as systematic. To get accurate 3-D information, a calibration procedure that compensates for errors in range measurements is needed. In this work, we propose a novel calibration approach for the Azure Kinect Developer Kit (DK) depth camera that employs a probability-based model involving reflection analysis. A planar checkerboard pattern and a planar pattern containing gray-valued areas are used to determine depth camera calibration parameters. Evaluations are done for each pixel, and a corresponding lookup table of camera parameters is acquired. The results of the presented method at various distances are assessed, and an improvement in the accuracy of the 3-D measurements of the ToF camera is achieved.