EVERYTHING YOU ALWAYS WANTED TO KNOW ABOUT SNAKES (BUT WERE AFRAID TO ASK)

Active contour models – known colloquially as snakes – are energy-minimising curves that deform to fit image features. Snakes lock on to nearby minima in the potential energy generated by processing an image. (This energy is minimised by iterative gradient descent according to forces derived using variational calculus and Euler-Lagrange Theory.) In addition, internal (smoothing) forces produce tension and stiffness that constrain the behaviour of the models; external forces may be specified by a supervising process or a human user. As is characteristic of gradient descent, the energy minimisation process is unfortunately prone to oscillation unless precautions – typically the use of small time steps – are taken. Active contour models provide a unified solution to several image processing problems such as the detection of light and dark lines, edges, and terminations; they can also be used in stereo matching, and for segmenting spatial and temporal image sequences. Snakes have often been used in medical research applications; for example, in reconstructing threedimensional features from planar slices of volume data such as NMR or CT images. In addition, many motion tracking systems use snakes to model moving objects. The main limitations of the models are (i) that they usually only incorporate edge information (ignoring other image characteristics) possibly combined with some prior expectation of shape; and (ii) that they must be initialised close to the feature of interest if they are to avoid being trapped by other local minima.