基于改进POCS算法的太赫兹图像超分辨率重建  

Super-resolution reconstruction for terahertz images based on improved POCS algorithm

作　　者：雷茂 郭锋[1] 秦明伟[1] LEI Mao;GUO Feng;QIN Mingwei(School of Information Engineering,Southwest University of Science and Technology,Mianyang 621010,China)

机构地区：[1]西南科技大学信息工程学院,四川绵阳621010

出　　处：《传感器与微系统》2022年第3期122-125,共4页Transducer and Microsystem Technologies

基　　金：国家重点研发计划资助项目(2016YFF0104003)。

摘　　要：在太赫兹成像系统中,因衍射极限而导致图像具有较低的空间分辨率。为了提高图像的分辨率,提出了一种改进的凸集投影(POCS)算法。采用新的边缘指导插值(NEDI)算法获取初始图像;采用对中心像素在边缘的模板系数加权的方式优化点扩散函数,并且加权值能随边缘与水平方向夹角的改变而自适应调整。利用太赫兹扫描成像系统生成0.5 THz图像进行算法验证。实验结果表明:该算法重建后图像峰值信噪比较传统POCS算法提升1.02 dB,验证了改进算法的有效性。In terahertz(THz)imaging system,the image has a lower spatial resolution due to the diffraction limitation.In order to improve the resolution of the image,an improved projection onto convex set(POCS)algorithm is proposed.The algorithm uses a new edge-directed interpolation(NEDI)algorithm to obtain the initial image;it uses the method of weighting the template coefficients of the central pixel at the edge to optimize the point spread function,and the weighted value can be adjusted adaptively as the angle between the edge and the horizontal direction changes.The THz scan imaging system is used to generate 0.5 THz image for algorithm verification.Experimental results show that the peak signal-to-noise of the image after the reconstruction of the algorithm is improved by 1.02 dB compared with the traditional POCS algorithm,which verifies the effectiveness of the improved algorithm.

关 键 词：太赫兹图像 超分辨率重建 边缘指导插值 凸集投影 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]