In our data, typical video frames contained measurement noise, resulting from floating food particles, light/shadow specks, reflections on the water’s surface, and dirt on the bottom of the chamber. Figure 3a provides an example frame from our camera setup. Our processing begins by attempting to remove much of this clutter. We first apply a standard image segmentation technique [Nob79] followed by binary separation of the video to foreground/background pixels, in order to separate the background from noise and fish blobs (Figure 3 b)
