Multispectral remote sensing image classification based on simulated annealing clonal selection algorithm

Clonal selection algorithm(CLONALG) has successfully performed pattern recognition and optimization tasks. However. it is difficult to apply CLONALG to remote sensing image classificafion because of characteristics of huge volume data. Therefore, in this paper, we propose a novel classification algorithm: simulated annealing clonal selection algorithm (SACSA) to perform multispectral remote sensing image classification task. By applying simulated annealing, the proposed algorithm avoids the premature convergence problem in CLONALG which is used to explain the basic features of all adaptive immune response to all antigenic stimulus, and call file requirement of diversity in the population of antibody. SACSA involves two stages. Firstly to enhance the globe convergence and accelerate lite computation, SACSA applies simulated annealing to optimize the population of antibody. During each move, the algorithm has some probability of changing its current configuration to a worse one that enables SACSA to jump out of local maxima or minima. Secondly, the classification task employs the property of clonal selection in SACSA. The clonal selection proposes a description of the way that the immune systems copes with file pathogens to mount all adaptive immune response. Classification results are evaluated by applying three known algorithm: parallelpiped, minimum distance and maximum likelihood to LandSat image. It is demonstrated that SACSA is superior to file three traditional algorithms, and is a dynamic clustering algorithm with global optimization.