A Comparison of Depth Sensors for 3D Object Surface Reconstruction

Abstract

The ability of depth cameras like Kinect to cap-ture a scene’s depth information in three-dimensions, along with 2D color RGB images, in real-time makes marker-less human motion capture a potential option for applications such as rehabilitation, robotics, education, etc. Various depth sensor technologies are commercially available, and selecting the ap-propriate depth sensor is highly dependent on the desired appli-cation. This research compared Kinect V2, Astra Pro, and Re-alSense D415 depth sensing technologies for object surface reconstruction within an interior daily living environment. Intel RealSense D415 was resistant to interference with multiple sen-sors and point cloud data at 1m range was more accurate than Kinect V2 and Astra Pro.