欢迎访问《兵工学报》官方网站,今天是 分享到:

兵工学报 ›› 2014, Vol. 35 ›› Issue (5): 654-661.doi: 10.3969/j.issn.1000-1093.2014.05.012

• 论文 • 上一篇    下一篇

一种基于多元探测器阵列的分块图像压缩传感算法

张智诠1, 丁晟1,2, 金伟其3   

  1. (1.装甲兵工程学院 控制工程系, 北京 100072; 2.73021部队, 浙江 杭州 310023;3.北京理工大学 光电成像技术与系统教育部重点实验室, 北京 100081)
  • 收稿日期:2013-08-21 修回日期:2013-08-21 上线日期:2014-06-23
  • 作者简介:张智诠(1959—),男,教授,博士生导师

A Block Compressed Sensing Method Based on Multi-element Detector Array

ZHANG Zhi-quan1, DING Sheng1,2, JIN Wei-qi3   

  1. (1. Department of Control Engineering, Academy of Armored Force Engineering, Beijing 100072, China;2. No.73021 Unit of PLA,Hangzhou 310023, Zhejiang, China;3. Ministry of Education Key Laboratory of Photo Electronic Imaging Technology and System,Beijing Institute of Technology, Beijing 100081, China)
  • Received:2013-08-21 Revised:2013-08-21 Online:2014-06-23

摘要: 针对压缩传感理论应用于单元探测成像时图像重构时间随图像尺寸增大而迅速增大的问题,提出了一种利用数字微镜和多元探测器进行编码测量,利用最小全变分法进行图像重构的分块图像压缩传感算法,并对重构分块图像进行灰度拉伸,提高了图像的峰值信噪比和结构相似度。仿真实验结果表明:算法具有计算时间短、重构图像质量高的特点,对于16图像分块,至少可缩短40%重构时间;通过分块图像灰度拉伸,重构图像的峰值信噪比和平均结构相似度较不拉伸时分别提高70%和11%. 文中算法为高分辨压缩成像的应用研究提供了一种有效的技术参考。针对压缩传感理论应用于单元探测成像时图像重构时间随图像尺寸增大而迅速增大的问题,提出了一种利用数字微镜和多元探测器进行编码测量,利用最小全变分法进行图像重构的分块图像压缩传感算法,并对重构分块图像进行灰度拉伸,提高了图像的峰值信噪比和结构相似度。仿真实验结果表明:算法具有计算时间短、重构图像质量高的特点,对于16图像分块,至少可缩短40%重构时间;通过分块图像灰度拉伸,重构图像的峰值信噪比和平均结构相似度较不拉伸时分别提高70%和11%. 文中算法为高分辨压缩成像的应用研究提供了一种有效的技术参考。

关键词: 信息处理技术, 压缩传感, 图像稀疏与重构, 多元探测器, 平方约束最小全变分法, 等式约束最小全变分法

Abstract: A block compressed sensing method is proposed considering that the calculation time of image reconstruction increases rapidly with the image size when compressive sensing (CS) theory is applied in single-pixel imaging. In the method, a digital micro-mirror device (DMD) and a multi-element detector array are used for encoding and measuring, and a total variation minimization method is used for image reconstruction, where the peak signal-to-noise ratio (PSNR) and the structural similarity (SSIM) of recovered images are improved by using gray stretch technique. Simulation results show that the method has the characteristics of shorter calculation time and higher recovered image quality. For 16 image blocks, the calculation time can be shorten by 40% at least. The PSNR and mean SSIM values of recovered images can be improved by 70% and 11%, respectively, through gray stretching of image blocks.

Key words: information processing, compressive sensing, sparse representation and reconstruction, multi-element detector array, total variation minimization with quadratic constraints, total variation minimization with equality constraints

中图分类号: