基于遗传算法的影像分级决策融合

在传统的影像单级融合技术基础上提出了基于遗传算法的分级影像决策融合方法，该方法适用于影像之间的冗余度、互补程度和融合顺序均未知的情况。算法采用模糊逻辑和遗传算法确定了影像间的冗余度和互补程度，并由此得出了近似的影像最优分级决策融合方式，最后通过实验比较了分级融合、单级融合以及源影像的目视效果和数学统计结果。

The Stepped Strategy for Image Merging Based on GA

In general, two or more images merging of multi-sensors is to merge a variety of image information of different sensors or different wave band, and then gain the determinant image of the target status and signature. In the process of image merging, a lot of information is complementary, but a great deal of information may still be redundant, we can reduce the redundancy degree and enhance the reliability of images through the merging of redundant information. On the basis of traditional method of image single-step merging, the paper puts forward a stepped strategy for image merging based on Genetic Algorithms, it is fit for the situations that the redundancy degree, complementary degree and the order of merging of images are all unknown. The algorithm uses fuzzy logic and Genetic Algorithms to make sure the redundancy degree and complementary degree among images, and determine the approximate optimal image merging strategy. At last we compare the visual effect and mathematic statistics of stepped merging with single-step merging and source image through experiments.