摘要
偏振成像可以得到目标的多维偏振信息,各偏振参量图像之间具有很强的相关性,它们之间不仅包含大量的共有信息,还拥有各自的特有信息。针对如何从复杂背景中凸显目标,提高目标的辨识度这一问题,提出一种基于改进神经网络模型的偏振图像融合算法,将传统PCNN模型扩展到双通道,采用像素的平均梯度(AG)作为PCNN的链接系数,空间频率(SF)作为模型的输入项,对强度图与偏振度图进行融合处理,最后将实验结果与其他常用方法进行对比,客观评价指标表明,该算法融合效果更佳,且融合图像的边缘细节特征也更为清晰明显。
Polarization imaging can obtain multi-dimensional polarization information of the target.The polarization parameter images have a strong correlation.It contains not only a lot of common information,but also its own unique information.For the prob⁃lem of how to highlight the target from a complex background and improve target recognition.A polarization image fusion algorithm based on an improved neural network model is proposed to extend the traditional PCNN model to dual channels,The average gradi⁃ent(AG)of pixels is used as the link coefficient of the PCNN,and the spatial frequency(SF)is used as the input of the model.Then,the intensity image and the polarization image are fused,and the experimental results are compared with other commonly used methods.Objective evaluation indicators show that this fusion algorithm has better effect,and the edge details of the fusion im⁃age are also more clear and obvious.
作者
姜兆祯
韩裕生
任帅军
张延厚
JIANG Zhaozhen;HAN Yusheng;REN Shuaijun;ZHANG Yanhou(Department of Information Engineering,Army Academy of Artillery and Air Defense Forces,Hefei 230031;Anhui Province Key Laboratory of Polarized Imaging Detection Technology,Hefei 230031)
出处
《舰船电子工程》
2021年第3期33-36,175,共5页
Ship Electronic Engineering
