TRACKING DEFORMABLE OBJECTS IN THE PLANE USING AN ACTIVE CONTOUR MODEL

In this paper, we consider the problems of segmenting a noisy intensity image and tracking a nonrigid object in the plane. In evaluating these problems, we explore a new technique based on an active contour model commonly called a ''snake.'' We have applied this technique to cell locomotion and tracking studies. The snake permits us to simultaneously solve, in constrained cases, both the segmentation and tracking problems. We present a detailed analysis of the snake model, emphasizing its limitations and shortcomings, and propose improvements to the original description of the model. We discuss in detail how the various parameters and forces of the snake model can be selected. We also consider problems of convergence of the optimization scheme. In particular, we propose an improved terminating criterion for the optimization scheme on the basis of topographic features of the graph of the intensity image (or potential surface). We also address the problem of how to derive useful potential surfaces on which the snake can crawl. Hierarchical filtering methods, as well as a continuation method based on a discrete scale-space representation, are discussed in this context. Results for both segmentation and tracking are presented. Possible failures of the method are also discussed.