Image Processing for Undulatory Locomotion

Sean Waters


The swimming patterns of undulatory swimmers can be modeled by tracking changes in their medial axis. In the past, the process for extracting this axis from video consisted of time consuming manual analysis. Our research proposes to optimize collection of medial axis data via a mathematical model of the image. By identifying the outline of the swimmer in a video frame, we identify the medial axis through a series of computations based on the concept of a medial axis as rst described by Blum. The algorithm depends upon two parameters, the rate at which outline points are chosen for inclusion in the computation, and the size of the contour ratio associated with each candidate medial axis point. Through selective alterations of these parameters the algorithm can be applied exibly in a wide array of video. To demonstrate the utility of these techniques in the modeling of swimming mechanics, we apply it to video clips of sh of several species in a variety of situations.


Image Processing; Undulatory Swimmers; Blum's Medial Axis

Full Text: PDF


  • There are currently no refbacks.