Error control in the set-up of stereo camera systems for 3d animal tracking

TL;DR

This paper analyzes how the setup and calibration of stereo camera systems impact the accuracy of 3D animal tracking, which is crucial for studying collective behavior.

Contribution

It provides a detailed analysis of errors in 3D reconstruction and how they depend on camera setup and calibration precision.

Findings

Error sources significantly affect 3D reconstruction accuracy

Calibration precision is critical for reliable 3D trajectories

Optimal camera setup improves tracking reliability

Abstract

Three-dimensional tracking of animal systems is the key to the comprehension of collective behavior. Experimental data collected via a stereo camera system allow the reconstruction of the 3d trajectories of each individual in the group. Trajectories can then be used to compute some quantities of interest to better understand collective motion, such as velocities, distances between individuals and correlation functions. The reliability of the retrieved trajectories is strictly related to the accuracy of the 3d reconstruction. In this paper, we perform a careful analysis of the most significant errors affecting 3d reconstruction, showing how the accuracy depends on the camera system set-up and on the precision of the calibration parameters.