Rat Paw Tracking for Detailed Motion Analysis

This work is part of a research project studying the learning of fine motor skills in rats. A system for tracking a rat paw using a system of high-speed cameras is presented along with preliminary analysis of the correlation between paw movement and neural data. The tracking method is generative, modeling the rat paw as a set of linked ellipsoids. To find the most probable paw pose a number of hypothetical parameter values are explored. For each set of values the paw model is projected in the different cameras and compared to the actual measured images. The pose that is most consistent with the intensity and edge information in the different views is chosen. By efficient utilization of integral images, this evaluation can be performed very quickly, creating a tractable and completely automatic method with good performance