Simulated Annealing and Genetic Algorithms for Shape Detection

The paper presents three heuristical methods for the solution of the shape detection problem. This problem arises in a large number of applications and is therefore of large interest to develop eeective methods for its solution. We formulate the shape detection problem as a combinatorial optimization problem and use methods based on simulated annealing and genetic algorithms for its solution. To speed up the detection process we develop a parallelization strategy for both methods. As a third method we present a constructive heuristic based on ltering noise. Experimental results show that the simulated anneal-ing and the constructive method give very good results, computed in short time.