An indoor performance comparison of time-of-flight depth cameras

The use of depth sensors for precision animal management has grown in recent years, with the ability to capture animals in many different lighting conditions. Time-of-flight (TOF) cameras are a technology that has demonstrated good performance for capturing images in indoor conditions However, with the continuous turnover in commercially available cameras, there is a need for an objective comparison between cameras. Cameras with different sensors and optics can greatly vary in terms of accuracy and repeatability. The objective of this study is to test three time-of-flight cameras for distance and object dimension accuracy: Pico Flexx, Kinect v.2, and Pico Zense. Images from each camera were collected while viewing three different sizes of foam boards in four different positions within the images, with distances ranging from 1 to 3 m. Dimensions of the boards collected from the images include maximum length (row and column), area, and distance from the cameras. Statistical measures were used to compare the data obtained from the images in relation to the distance, size and position parameters. The results show that the cameras achieved similar accuracy. Pico Flexx presented the best results for distance and Pico Zense the best results for dimension measurements (area, row, and column). Length measurements were significantly affected by the position of the object within the image for both Pico Flexx and Kinect v.2. Board sizes did not significantly influence dimension measurents collected from the images.