红外光强与偏振图像多类拟态变元组合融合

Combination fusion of multi-types mimic variables of infrared intensity and polarization image

现有红外光强与偏振图像融合算法不能根据图像差异特征变化动态调整融合算法,造成图像部分差异特征融合效果不理想甚至失效。根据拟态仿生学思想,借鉴拟态章鱼的多拟态过程,提出一种红外光强与偏振图像多类拟态变元组合融合方法。首先,分析拟态章鱼的多拟态过程,剖析其多拟态原因;其次,寻找多拟态过程与图像融合过程之间的对应关系,并确定图像融合过程的多类变元类型;最后,建立面向图像融合的多类变元组合关系并利用该关系进行图像融合。实验结果表明:所得融合图像的信息熵、标准差、边缘强度、平均梯度、清晰度方面平均提升1.16%、7.25%、3.00%、0.31%、10.18%。该方法的建立可以使融合算法内变元组内变元选择和组合根据原始图像差异特征变化而进行动态调整,从而得到具有针对性的融合算法。

The existed infrared intensity and polarization image fusion algorithm could not dynamically adjust fusion algorithm according to the change of image difference characteristics,which results in that the partial differences feature fusion effect was not ideal or even failure.According to thought of mimicry bionics and learning from Mimicry Octopus ′ s multi-mimicry process, a fusion method of combination fusion of multi-types mimic variable of infrared intensity and polarization images was proposed. Firstly,Mimicry Ooctopus′ s multi-mimicry process,the reason of multi-mimicry were analyzed. Secondly, the correspondence between multi-mimicry process and image fusion process was found, and multiple types variables of the image fusion process were determined.Finally, a reversible variable composition relation for image fusion was established and image fusion was carried out that used the combined relation. The experimental results showed that the information entropy, standard deviation, edge intensity, average gradient and sharpness of the obtained fusion image were obviously improved 1.16%, 7.25%, 3.00%,0.31%, 10.18% in average. The establishment of this method could make the fusion algorithm within the variable selection and combination according to the original image difference characteristics of the dynamic adjustment, so as to obtain a targeted fusion algorithm.