偏振图像融合技术综述

Review of Polarization Image Fusion Technology

偏振图像融合技术旨在充分利用偏振信息弥补单一强度图像在一定场景条件下无法提供充足信息的劣势,广泛应用于目标探测识别以及计算机视觉领域。文章从偏振信息获取设备及方法出发,介绍了基于传统方法和基于深度学习方法的融合算法思路,以及其在典型场景和遥感中的研究现状。其中,传统偏振融合算法部分着重陈述了多尺度分析方法、表示学习方法、伪彩色融合方法以及综合方法等几种方法的原理。最后,文章对偏振融合的发展做出展望,为后续研究工作提供了参考。

Polarization image fusion technology aims to make full use of polarization information to make up for the deficiency that a single intensity image cannot provide sufficient information under certain scene conditions. It is widely used in target detection and recognition, computer vision fields. This paper proceeds from the equipment and method of polarization information acquisition, then introduces the algorithm ideas based on traditional and deep learning method. At the same time, the research status of these algorithms′ usages in typical scene and remote sensing were also given. Among the traditional polarization fusion algorithm part, the principle of several methods are emphatically stated, such as Multi-scale analysis method, representation learning method, false-color fusion method and synthesis method. Finally, this paper predicts several research directions of polarization fusion, which guide to the subsequent research work.