where E_elastic and E_stiffness represents the energy functions with reference of the snake itself and E_external represents energy functions due to the image properties and can be modified depending on image requirements [1].
Advantages of Snakes
Snakes are useful as an edge detector because they incorporate possible interactions between high level ‘intelligence’ processing models and the algorithm itself, since
(1) the user can both supply an initial good estimate of a boundary to the snake and apply the algorithm to get a ‘lock-on’ on the boundary,
2) the user can specify weightings on the different energy functions or even add a specific external energy function that would then shift the snakes to search for other features.
Disadvantages of Snakes
However, snakes have several disadvantages.
(1) Snakes are mainly a local operator, meaning that their search for edges is largely constrained by the neighborhood size in which they are to search for the next local minima in energy. This brings about two problems, first that the snake would not be capable of traversing areas of uniform intensity since there is no edge information available to pull the snake accordingly. Also, snakes would tend to ‘lock on’ to closest edges it would find, possibly distorting the boundary significantly, especially if the image is noisy. These two problems places a strict constraint on the snakes where they have to be placed relatively close to the object boundary itself, lest they ‘lock-on’ to erroneous edges and distort the final boundary or worse, completely collapse upon itself as it attempts valiantly to search for edges that are too far from the original region they were placed in.
(2) The proper functioning of the snake was critically dependent on the relative weighting on each of the energy terms. This means that rigorous trial and error might be required before a ‘good’ weighting is obtained for a particular set of images – something that does not seem practical for a general application.