Improved Genetic Algorithm And Its Application In Image Processing

In recent years, the genetic algorithm, a global random search method, has attracted a great deal of attention and has an important effect on many large projects. In this thesis, an overview of the basic theories and the recent development is given first. Then, IGA (Improved Genetic Algorithm), a new genetic algorithm, is proposed and applied in primitives extraction. Added five new steps, IGA efficiently solves the problem of global convergence and improves the convergence speed .The new steps are hybrid selection operator, competition among brothers , greedy operation, filter operation, dynamic supply new individuals. The first three steps solve the problem of global convergence and accelerate the convergence process. Besides retaining the variety of the population, the last two steps also avoid converging in local optimum. The experimental results show that IGA has great advantage of global convergence property and high convergence speed over many existing genetic algorithms. For shape analysis, the idea that primitives extraction can be converted to an optimization problem, which is finding the maximum match of the primitive represented by parameters with the edge image, is proposed. With the application of IGA, the experimental results show that this method could find the primitives correctly.