【作者】 林波；

【导师】 韩国强；

【作者基本信息】 福州大学 ， 测试计量技术及仪器， 2018， 硕士

【摘要】 压缩感知理论是一种新兴的采样理论,它突破了奈奎斯特采样定理的限制,能从较少的采样数据中,精确重构出原始信号。将压缩感知理论应用于成像技术中,可以有效地减少设备的成像时间,提高成像质量。本文主要以原子力显微镜(Atomic Force Microscope,AFM)和核磁共振成像仪(Magnetic Resonance Imaging,MRI)为研究对象,深入研究了基于压缩感知的AFM和MRI成像技术。AFM是纳米尺度下的高精密测试仪器,它可以对多种材质的样品进行三维表面形貌成像。然而受到AFM的机械结构和测量原理的限制,其成像范围小,成像速度较慢。高速AFM是其发展的必然趋势。压缩感知技术可以有效地减少AFM的测量点数,大幅减少AFM成像时间,减少探针针尖与样品表面间的相互作用力,避免样品的损伤和探针的磨损。本论文着重研究了基于压缩感知的AFM成像技术,分析了采样率和样品形貌的不同对AFM重构图像质量的影响。并在此基础上,分析了压缩感知扫描模式即测量矩阵(随机扫描模式、SRM扫描模式、螺旋扫描模式、光栅扫描模式和方形扫描模式)对测量结果的影响,分析了不同的重构算法(11-ls算法、GPSR算法、TVAL3算法和IHT算法)对AFM图像重构的影响。为了实现超高分辨率的压缩感知AFM成像,本论文提出了基于分块压缩感知的AFM成像技术。通过引入重叠分块的思想,减弱甚至消除因分块处理而造成的块效应。本论文对分块压缩感知中分块大小和重叠率对AFM重构图像质量的影响进行了详细的分析,并将该方法与传统的压缩感知成像技术进行实验对比。基于分块压缩感知的AFM成像技术可以在保证成像质量的基础上,有效地增加AFM成像范围。核磁共振成像仪是临床医学中一种重要的医学影像设备。MRI由于其成像特点,成像时间过于漫长,会对患者造成一定的不便和危害。为了消除这一点,利用压缩感知技术对K空间数据进行降采样可以有效地缩短核磁共振成像时间而不降低其成像质量。本论文利用三种采样模板(一维下采样模板、仿射线下采样模板和二维下采样模板)进行了降采样模拟实验,并利用三种重构算法(TVAL3算法、Sparse-MRI算法和GPSR算法)对降采样数据进行重构。压缩感知实现了快速核磁共振成像。更多还原

【Abstract】 Compressive sensing is an emerging sampling theory.It can break through the limitation of the Nyquist sampling theorem,and accurately reconstruct the original signal with minimal sampling data.The application of compressive sensing theory to imaging technology can effectively reduce the imaging time and improve the imaging quality.In this paper,we focus on the research of compressive sensing imaging technology with atomic force microscope(AFM)and magnetic resonance imaging ceremony(MRI).AFM is a high-precision test instrument on the nanometer scale.It can perform three-dimensional surface topography imaging for a variety of materials.However,limited by the mechanical structure and measurement principle of the AFM,the imaging range is small and the imaging speed is slow.High speed AFM is the inevitable trend of its development.Compressive sensing technology can effectively reduce the number of AFM measurement points,significantly reduce the AFM imaging time,reduce the interaction between the probe tip and the sample,and avoid sample damage and probe wear.In this paper,we focus on the study of AFM imaging technology based on compressive sensing,and analyze the effects of different sample rates and sample morphologies on the quality of the AFM reconstructed image.On this basis,we analyze the effects of different scan patterns(random scan pattern,SRM scan pattern,spiral scan pattern,raster scan pattern and square-shape scan pattern)on the measurement results,and analyze the effects of different reconstruction algorithms(11-1s algorithm,GPSR algorithm,TVAL3 algorithm and IHT algorithm)on the AFM image reconstruction.In order to achieve ultra-high resolution compressive sensing AFM imaging,we propose AFM imaging technology based on block compressive sensing in this paper.By introducing the idea of overlapping blocks,the block effect caused by block processing is reduced or even eliminated.We analyze the effect of block size and overlap rate on the reconstruction of AFM images and compare this method with the traditional compressive sensing imaging technology.AFM imaging technology based on block compressive sensing can effectively increase the AFM imaging range.MRI is an important medical imaging device in clinical medicine.Because of the imaging characteristics of MRI,its imaging time is too long,which will cause some inconvenience and harmful to the patient.Using compressive sensing technology to downsample K-space data of magnetic resonance can effectively shorten the imaging time without degrading the imaging quality.In this paper,three undersampling patterns(cartesian sampling pattern,pseudo radial sampling pattern and 2D sampling pattern)were used to perform downsampling simulation experiments,and three kinds of reconstruction algorithms(TVAL3 algorithm,Spa rse-MRI algorithm,and GPSR algorithm)were used to reconstruct the downsampling data.Compressive sensing achieves fast MRI.更多还原