摘要
高精度配准是图像融合的前提。随着膜计算的深入研究,利用膜计算分布式的、并行的特点,我们先前已提出了两种在P系统框架下多模态图像配准算法,即GA-MCIR和DE-MCIR。首先简单阐述两种已提出的算法。然后,将卫星图像、红外与可见光多模态图像作用在所提的两种算法之上,并进行比较;对于卫星图像配准实验,DE-MCIR的平均互信息值为1.4306,标准差为0.00341;对于红外与可见光等多模态真实图像,DE-MCIR的平均互信息值为0.0125,标准差为0.00187。最后,比较实验的结果表明,相比于GA-MCIR,DE-MCIR具有更高的配准精度和鲁棒性。
Higher-precision registration is a precondition for image fusion.Based on the distributed and parallel characteristics of the membrane computing,we have proposed two kinds of multimodality image registration algorithms named as GA-MCIR and DE-MCIR in the framework of P system.Firstly,simply elaborates the two algorithms.Then,the two algorithms are compared for the satellite image,infrared im-age and visible multimodality images.For the satellite images registration experiment,the average mutual information value of DE-MCIR is 1.4306,the standard deviation is 0.00341;For the infrared and visible images,the average mutual information value of DE-MCIR is0.0125,the standard deviation is 0.00187.Finally,the experiment results reveal that DE-MCIR algorithm shows better registration accuracy and robustness than GA-MCIR.
基金
四川省教育厅重点项目(No.14ZA0118)
关键词
膜计算
P系统
多模态图像配准
Membrane Computing
P System
Multimodality Image Registration