Boundary extraction algorithm based on particle motion in a vector image field

This paper proposes a novel concept of a vector model for representing images. This opens up the possibility of performing vector operations on an image. A Gaussian weighted image moment vector operator is derived and used to generate an edge vector field. The vector field in turn is used to drive a particle. The particle trajectory in a force field is simulated to obtain a close object boundary. Results of implementing the algorithm in single and multiple object images under noisy condition are presented. The technique is simple, robust, fast, and achieves subpixel accuracy.