Research on adaptive target detection based on improved genetic algorithm from infrared images

In this paper, a novel optimized genetic algorithm based on morphology for target detection from infrared images is proposed. In our improved algorithm, a new fitness measurement method based on target characteristics value is introduced to meet specific target detection needs. Male and female parent dynamic clustering methods are put forward to make crossover operator more reasonable. Besides, multi-population parallel evolution and optimum individual transplant strategy are adopted to merge optimum individual keeping and gene keeping effectively. Crossover probability and mutation probability are adjusted adaptively according to population diversity and more reasonable target characteristics variable is designed according to the features of infrared images. Morphology filter based on genetic optimization for infrared target detection is given to recognize structural information and target background information. Experimental results demonstrate that the convergence speed can be controlled and local search ability is increased as well by using trained structural elements based on improved genetic optimum. In addition, the efficiency and accuracy is boosted evidently and noise can be restrained to a great extent.