单次曝光频域振幅编码压缩成像  

Single-Exposure Frequency-Domain Amplitude Encoding Compressive Imaging

作　　者：张成[1,2] 程鸿[1] 张芬[1] 沈川[1] 韦穗[1] 

机构地区：[1]安徽大学计算智能与信号处理教育部重点实验室,安徽合肥230039 [2]安徽省现代成像与显示技术重点实验室,安徽合肥230039

出　　处：《电子学报》2014年第7期1262-1267,共6页Acta Electronica Sinica

基　　金：NSFC-广东联合基金(No.U1201255);国家自然科学基金(No.61201396;No.61201227;No.61301296;No.61377006);高等学校博士学科点专项科研基金(No.20113401130001);安徽省自然科学基金(No.1208085QF114);安徽大学博士科研启动经费(No.33190218);安徽大学青年基金(No.KJQN1120)

摘　　要：超分辨率被认为是光学成像和图像处理的"圣杯"之一.有别于传统的多幅亚像素图像配准融合实现超分辨率的方法面临的配准误差以及高成本问题,得益于大多数图像普遍具有的稀疏表示特性,本文将压缩传感理论引入超分辨率成像,提出一种新的单次曝光频域振幅编码压缩成像方法.利用4-f傅里叶光学架构实现图像信息的频域0/1振幅随机调制,然后可以使用低分辨率CCD器件实现积分下采样记录对应的测量值,最后利用优化方法从少量的测量值中重建原高分辨率图像.模拟实验验证了本文提出的方法可以有效地实现二维图像信息的获取与重构.此外,本文的方法可以有效地处理大尺寸图像的压缩成像问题,具有重要的应用前景.Super resolution （SR） is being considered as one of the ＂holy grails＂ of optical imaging and image processing. Different from the registration error and costly problem faced in multiple subpixel image registration fusion method to achieve super- resolution, this paper introduces the compressive sensing theory into super-resolution imaging, which benefit from the general sparse representation of most nature images, and proposes a novel single-exposure frequency-domain amplitude encoding compressive imaging method.Exploiting the 4-f Fourier optics architecture for modulating the image information by the 0/1 amplitude randomly in the frequency domain, low-resolution CCD device can then be used to records the corresponding measured values by integral downsampling and finally apply optimization methods to reconstruct the original high-resolution images from small number of mea- sured values. Simulation experiments demonstrate that the 2D image information can be effectively acquired and reconstruction from the measured data by our proposed method. In addition, our method can effectively deal with large-scale image compressive imaging problem and thus has an important application prospects.

关 键 词：压缩成像 图像超分辨率 频域振幅编码 4-f光学架构 单次曝光 

分 类 号：TN911.7[电子电信—通信与信息系统]