Comparing Depth Estimation of Azure Kinect and RealSense D435i Cameras

Azure Kinect and Intel RealSense D435i are modern depth cameras based on two different depth estimation approaches: Azure Kinect uses Time-of-Flight, and RealSense uses stereo vision technologies . In this paper, we present a systematic comparison of Azure Kinect and Intel RealSense D435i cameras, focusing on the accuracy of depth estimation under various influencing factors. We analyze the influence of factors such as sensor temperature, camera distance, ambient illumination and temperature, and target color on the depth estimation accuracy and noise sensitivity. In addition to comparing the spatial and temporal distribution of per-pixel mean distance, noise, and offset distribution, we also analyze the noises and outliers on depth images and apply appropriate statistical treatment to mitigate them. Our findings offer insights into selecting suitable depth cameras for specific applications, highlighting the strengths and limitations of each camera under different operating conditions.