An image analysis system to compute the predominant direction of motion in a foucault pendulum

In this document, we propose a machine vision system to detect the predominant direction of motion of a Foucault pendulum. Given a certain configuration where the camera has a top view of the pendulum's bob in motion, the system builds an adaptive model of the background. From it, the bob's center of mass is computed. Then, an ellipse model is fitted to the trajectory. Finally, the noise in the observed predominant direction of motion is filtered out to get a robust estimate of its value. The system has proved to be quite reliable on a simple version of the Foucault pendulum where it was tested.