一种基于SGNN网络的模糊图像融合算法

首先对现有的M-L算法提出了优化顺序和优化次数的改进;接着提出了对生成的SGNN网络先剪枝再一次优化的综合处理方法;然后对原有的基于SGNN和模糊理论的图像融合提出了两点改进:在原有的图像像素聚类后,加入了综合处理环节,使得图像像素聚类的效果更好;针对由于不同传感器获得的图像灰度特性的不一致导致的聚类后各类类中心的灰度值差别很大,甚至分类数目都不一致的问题,提出了改进的融合方法.仿真证明了所提的模糊融合方法的优越性.

Fuzzy image fusion algorithm based on SGNN

The optimized order and number for current M-L optimization algorithms are presented;and the composite processing method of clipping first and optimizing again for the generated SGNN network is proposed.Then,after clustering the original image element,the proposed composite processing method is added to get better element clustering.Finally,aiming at the large grayscale differences of every class's center and the different class number because of the different grayscale characteristics of images acquired from different sensors,a modified fusion method is proposed.Simulation result verifies the superiority of the proposed fuzzy fusion algorithm.