A New Approach for Polygonal Approximation of Shape Contours using Genetic Algorithm

Polygonal approximation is an important issue in pattern recognition and image processing. A new approach for polygonal approximation of shape contours based on genetic algorithm (GA) is proposed. The goal of feature preservation is proposed definitely at first time. Two combined objective measures are also proposed, which are simple to be calculated and do good work on different types of shape contours. Moreover, special policies for initialization and genetic operators are proposed to improve the quality of optimization. The experimental results confirmed that the proposed combined measures for feature preservation are better than ISSE (integral summed square error) which is most frequently used by other approaches for optimization of polygonal approximation.