超分辨率重建算法在热红外图像中的应用  

Application of Super Resolution Reconstruction Algorithm in Thermal Infrared Image

作　　者：曹嘉豪 李春来[1] 金健[1] 姬弘桢[1] 张旭东[1] 

机构地区：[1]中国科学院上海技术物理研究所,上海200083

出　　处：《红外》2016年第5期27-32,共6页Infrared

基　　金：国家高技术研究发展计划("863"计划)项目(2012AA121102)

摘　　要：高空间分辨率的热红外图像能够提供更多关于目标场景的细节信息,因而在计算机视觉、医学和遥感等诸多领域有着广泛的应用需求。由于通过提升热红外相机硬件性能的方式往往需要付出高昂的代价,因此我们选择通过超分辨率重建的方式来提高热红外图像的空间分辨率。本文所重建的热红外图像来源于舟山航拍试验,所用热红外相机由自主设计搭载。分别采用凸集投影法和迭代反投影法对热红外序列图像进行了超分辨重建。实验结果表明,这两种算法均能有效提高热红外图像的空间分辨率。Since high spatial resolution thermal infrared images can provide more details of target scenes,they are widely used in the fields of computer vision,medicine and remote sensing etc.Because it is usually more expensive to improve the performance of a thermal infrared camera,the super-resolution reconstruction algorithm is selected to improve the spatial resolution of thermal infrared images.The thermal infrared images to be reconstructed are obtained in the flight test over Zhoushan.The thermal infrared camera used is designed and equipped independently.The thermal infrared images are reconstructed by using the iterative back-projecting algorithm and the Projection onto Convex Sets Approach algorithm respectively.The experimental results show that both algorithms can improve the spatial resolution of thermal infrared images effectively.

关 键 词：超分辨率重建 热红外图像 凸集投影法 迭代反投影法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]