【作者】 朱宏权；

【导师】 王奎禄； 魏义祥；

【作者基本信息】 清华大学 ， 核科学与技术， 2008， 博士

【摘要】 γ射线针孔照相系统是用于诊断γ射线源或X射线源图像的重要手段,它主要由厚针孔、闪烁体、微通道板像增强器和CCD相机等器件组成。由于受到各器件空间分辨能力的限制和噪声的影响,系统获得的图像的质量产生了退化,难以如实地反映源区射线强度的空间分布。因此,图像复原技术的研究是提高系统图像诊断水平的重要课题。本文通过分析γ射线针孔照相系统各组成器件的点扩散函数和噪声特性,建立了系统的图像退化模型,根据系统的探测量子效率分析了图像退化的主要影响因素,在此基础上提出了相应的图像去噪和图像复原方法,并通过对钴源和杆箍缩二极管的针孔照相实验结果的复原处理,验证了图像复原的效果。论文主要成果如下:1.采用理论计算或实验测量的方法得到厚针孔、闪烁体、像增强器和CCD相机的点扩散函数,计算得到其调制传递函数,以此来评价针孔照相系统的空间分辨性能。?2.根据探测量子效率的级联理论和γ射线针孔照相系统的量子传递过程,计算了系统的探测量子效率。计算结果表明,厚针孔的透过效率和调制传递函数是引起系统图像退化的主要因素。3.通过对γ射线针孔照相系统噪声特性的分析,将图像噪声分解为伽马噪声和高斯噪声,在此基础上提出的自适应中值替换法和自适应小波滤波法取得了良好的图像去噪效果。4.基于Pixon图像模型的多分辨率特性和Richardson-Lucy迭代法原理,提出了一种空间移不变系统的图像复原方法,用于闪烁体、像增强器和CCD相机的图像复原。模拟结果表明,该方法具有很好的收敛性和复原效果。5.针对厚针孔成像过程的空间移变性,采用Richardson迭代法进行图像复原处理,取得了满意的复原效果,并提出迭代次数的确定方法。更多还原

【Abstract】 γ-ray pinhole imaging system is a major approach to diagnose the radioactive source. It mainly comprises four parts: thick pinhole, LSO scintillator, Microchannel-plate image intensifier and CCD camera. The acquired image is degraded due to the limited spatial resolution and random noise of the constituent instruments. In order to improve the image quality and give a better understanding of the radioactive source, this dissertation aims at systematical study on the image restoration of thick pinhole imaging system. The image denoising methods and deblurring methods were respectively proposed according to the properties of point spread function(PSF) and noise. The methods were verified by the application to the pinhole imaging experiment results of 60Co source and rod-pinch-diode. The main contributions of the dissertation can be outlined as follows.1. The PSFs of thick pinhole, LSO scintillator, microchannel-plate image intensifier and CCD camera were obtained through simulation or experiment.2. The cascade model of the detective quantum efficiency was established forγ-ray pinhole imaging system. The model shows that the detective efficiency and modulation transfer function of the thick pinhole are two main factors of the image degradation.3. The image noise is described by gamma noise and Gaussian noise. An adaptive algorithm is proposed to substitute gamma noise with the median of the neighborhood, which achieves satisfactory noise removal. The Gaussian noise is greatly reduced with the locally-adaptive wavelet filter.4. Pixon-based Richardson-Lucy method is proposed to restore the image blurred by LSO scintillator, image intensifier and CCD camera. This algorithm outperforms Richardson-Lucy method on the image improvement and convergency.5. Richardson iterative method is adopted to restore the image degraded by thick pinhole. The stopping rule of the iteration is determined by the simulation results.更多还原