结合NSST与优化脉冲发放皮层模型的红外与可见光图像融合  

Infrared and visible light images fusion combining with nonsubsampled shearlet transform and bee colony optimization parameters of spiking cortical model

作　　者：侯瑞超 周冬明[1] 聂仁灿[1] 刘栋 HOU Ruichao;ZHOU Dongming;NIE Rencan;LIU Dong(Information College,Yunnan University,Kunming Yunnan 650504,China)

机构地区：[1]云南大学信息学院,昆明650504

出　　处：《计算机应用》2018年第A01期202-207,239,共7页journal of Computer Applications

基　　金：国家自然科学基金资助项目(61365001;61463052)

摘　　要：针对传统红外与可见光图像融合时边缘不清晰、对比度不高、细节丢失等问题,结合非下采样剪切波变换(NSST)具有多尺度,最具稀疏表达特性与脉冲发放皮层(SCM)具有耦合、脉冲同步激发等优点,提出一种基于NSST与多目标人工蜂群优化SCM参数的图像融合方法。首先,通过NSST分解红外与可见光图像获得高频与低频子带系数,然后低频子带系数采用基于边缘信息指导SCM融合策略,高频子带系数采用改进的空间频率作为优化SCM的激励进行融合,最后经过NSST逆变换得到最终图像。实验结果表明,该方法相比其他融合算法不仅在主观评价上有一定程度的改善而且在客观评价指标互信息(MI)与边缘保留度(QAB/F)上有明显的提高。Aiming at edge blun＇ing, low contrast, loss of details problems in traditional infrared and visible light image fusion, combined with NonSubsampled Shearlet Transform （NSST） with multi-scale transformation and the most sparse expression characteristics, and Spiking Cortical Model （SCM） with the advantages of coupling and pulse synchronization, a fusion method for infrared and visible light images based on NSST and multi-objective artificial bee colony optimizing the parameters of SCM was proposed in this paper. Firstly, NSST was used to decompose infrared and visible light images to obtain high frequency and low frequency subband coefficients, then low frequency coefficients were fused by the SCM, which was guided by the edges of images. For the high frequency subband coefficients, a modified spatial frequency was adopted as the input to motivate the SCM. Finally the fused image was reconstructed by inverse NSST. Compared with other fusion algorithms, the experimental results show that the proposed method has a certain degree of improvement both on subjective evaluation and objective evaluation including Mutual Information （MI） and edge information preservation value （QAB/F）.

关 键 词：非下采样剪切波变换 脉冲发放皮层模型 多目标人工蜂群 图像融合 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]