 |
 |
   |
|
|
|
Dynamic Programming
Other related methods make it possible to obtain interesting results. Dynamic programming requires to have a limited space of research, a zone in which is located the object to detect, but it has the double advantage of restricting the effect of noise, and of being undisturbed by occlusions.
An optimal path is searched between two ends by means of the costs of all the paths between these ends. Paths must be contained in a space of research, in order to avoid a combinatory explosion. The selection of a point is conditioned by the cost of the path which led to him.
The theory was introduced in 1957 by BELLMAN and widely developed by VITERBI and applied to many fields: ways planning, voice recognition, etc. The dynamic programming constitutes one of the most promising methods for lines tracking.
The main default is the heaviness of the algorithm, but by choosing criteria easy to calculate, the process is carried out at a reasonable speed.
