项目名称: 联合基于学习的超分辨率技术和多传感器超分辨率技术在红外图像复原中的研究
项目编号: No.61271330
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 无线电电子学、电信技术
项目作者: 吴炜
作者单位: 四川大学
项目金额: 76万元
中文摘要: 红外图像在民用和军用领域都得到了极为广泛的应用。受光学系统等因素的性能的影响,红外图像的分辨率通常都较低。如何提高红外图像的分辨率成了一个迫切需要解决的问题。超分辨率技术是从单幅或者多幅低分辨率图像中复原出高分辨率图像的技术,它是一种具有广阔前景的方法。现阶段,国内外对红外图像分辨率技术的研究还处于初级探索阶段,亟待更深入地研究。基于学习的超分辨率和多传感器超分辨率技术具有各自的优势,它们依据的不同的信息对红外图像进行超分辨率复原,联合考虑它们的技术优势,完全能够取得更好的复原效果。本课题拟以增强红外图像的分辨率为背景,对联合基于学习的技术和多传感器技术进行研究,拟解决的关键问题包括:1)基于学习的红外图像超分辨率算法中的关键技术问题;2)多传感器红外图像超分辨率算法中的关键技术问题;3)联合基于学习的和多传感器的红外图像超分辨率复原问题;4)通过正则化框架在超分辨率复原同时抑制噪声问题。
中文关键词: 基于学习的超分辨率;多传感器图像超分辨率;基于稀疏表示的超分辨率;图像配准;稀疏表示
英文摘要: Infrared imaging has been an extremely wide range of applications in both civilian and military fields. The resolution of the infrared images is usually relatively low due to the performance of optical system. How to improve the resolution of infrared image has become an urgent problem to infrared imaging. Deriving a high-resolution (HR) image from the low-resolution (LR) one or a sequence of LR images provides a prospective solution to this problem, which is known as the super-resolution (SR) imaging technique. Currently, there still have many problems, which are needed to study for infrared image SR. Learning-based SR and multi-sensor SR has its own advantages; each of them is based on different information to enhance the resolution of infrared image By making full use of their advantages, a better result can be achieved. This project intends to enhance the resolution of infrared image by combining learning based SR and multi-sensor SR. The key issues include:1) the key issues in learning based SR for infrared image; 2) the key issues in multisensor SR for infrared image; 3) the issues of combining learning based SR and multi-sensor SR for infrared image restoration; 4) the issues of simultaneous SR and denoising in a framework.
英文关键词: Learning based super resolution;multi-sensor image resolution;sparse representation based super-resolution;;image registration;sparse representation