Tracking the Rotation of Lab Animals via Computer Vision

In this work, rotation information of lab animals are extracted using test videos obtained with a computer vision system that includes a QVGA color camera, PC, cage and the developed software. Grayscale morphological operations are performed to remove small artefacts and then the lab animal region is extracted for a given image frame. Euclidean distance and Radon transforms are performed to compute the rotation direction afterwards. The information of rotation and number of turns are obtained taking the predefined state model into account. It is observed that the developed system gives successful results for different kinds of scenes.