MATCHING STRUCTURAL SHAPE DESCRIPTIONS USING GENETIC ALGORITHMS

This paper presents a genetic algorithm for solving the problem of str uctural shape matching. Both sequential and parallel versions of the a lgorithm have been presented. The genetic operators-reproduction, cros sover and mutation-have been constructed for this specific problem. A new variation of the crossover operator, called the color crossover, i s presented. This operator has resulted in significant improvement in runtime and algorithm efficiency. Parallelization has been achieved us ing an ''island'' model, with several subpopulations and occasional mi gration. A complete framework for an object recognition system using t his genetic algorithm has been presented. Encouraging experimental res ults have been obtained. (C) 1997 Pattern Recognition Society. Publish ed by Elsevier Science Ltd.