A comparative study of calibration methods for Kinect-style cameras

Kinect-style (or Depth) cameras use both an RGB and a depth sensor that acquire color and per-pixel depth data (depth-map), respectively. Due to their affordable price and rich data they provide, depth cameras are being extensively used on research in assistive environments. Most of the robotic and computer-vision systems that use these Kinect-style cameras require an accurate knowledge of the camera-calibration parameters. Traditional calibration methods, e.g., those that use a checker-board pattern, cannot be straight-forwardly used to calibrate the Kinect-style cameras since the depth sensor can not distinguish patterns. Several calibration methods have emerged that try to calibrate depth cameras. In this paper, we present a comparative study of some of the most important Kinect-sytle calibration algorithms. Our work includes an implementation of these methods along with a comparison of their performance in both simulation and real-world experiments.