Markerless motion-capture for point-light displays

A markerless motion capture technique is described for creating point-light displays. The technique has a number of advantages over other methods for creating point-light displays, it does not require specialized equipment and can be used with human and animal actions. The method uses public domain and open source software to record the x-y coordinates of the visible joints in each frame of a video taped action. The coordinate files allow individual presentation of each element for psychophysical research on biological motion. A corpus of over 90 actions collected with this method is described. 3In the area of event perception Johansson’s (1973) point-light display, in which human action is reduced to a few moving points of light, have generated particular interest. In such displays there may be no static form information, yet complex forms will be seen when the points move. For example, if the major joints of a human (e.g., head, shoulders, elbows, wrists, hips, knees, and ankles) are visible as discrete points, the motions of these thirteen dots are sufficient to see a human engaged in an action. Indeed, the complex kinematic patterns of motion, generated by biological organisms, provide a wealth of information about the organism. Observers can use the point-light motion to identify the gende