【作者】 杨博雄；

【导师】 傅辉清；

【作者基本信息】 中国地震局地球物理研究所 ， 固体地球物理学， 2005， 博士

【摘要】 电荷耦合器件CCD(Charge Coupled Devices)是20世纪70年代初发展起来的新型半导体集成光电器件,它是美国贝尔电话实验室的W. S. Boyle和G. E. Smith于1970年首先提出的。近30年来,依靠已经成熟的MOS集成电路工艺,CCD器件及其应用技术得以迅速发展。CCD器件按其感光单元的排列方式分为线阵CCD和面阵CCD两类,但无论是线阵CCD还是面阵CCD,由于其固有的物理特性、工作机理等原因以及芯片结构、制作工艺等的限制,CCD像素精度不能做得很高。为了保证CCD像元具有足够的感光面积和防止相邻像元之间的互相串扰,所有CCD器件的像元面积以及像元间距都不能做得太小,目前一般在微米级别,这就限制了CCD在高精度测量领域中的应用。 CCD图像测最系统主要由照明系统、被测物体、光学成像系统、信号处理电路和计算机组成,因此影响CCD测量精度的因素主要有:CCD像元的制造误差,CCD对光照度的分辨率,A/D转换的精度,照明系统的不稳定性,CCD驱动电路的附加噪声,成像系统调焦误差,成像系统的误差(如球差、像差等),外界环境的影响(如温度、振动等)以及图像处理水平和软件算法优劣等。为了提升CCD测量水平,可以选用高分辨率的CCD器件和采样频率比较高的图像采集卡,或进行光学系统的放大,或者采用特殊的光源进行照明等。但这些方法的使用一般会受到某种程度的限制和影响,如CCD像素制造精度不能做得很高;采用光学系统放大时像的质量会下降;光源的稳定性均匀性对CCD成像质量的影响,而通过CCD图像处理和软件算法来提高CCD测量精度是一种简单方便而又行之有效的方法。为了提高CCD应用系统的测量精度,满足CCD在高精度测量领域的应用,人们提出CCD细分,同时开始与此相关的CCD细分技术与应用的研究。 CCD细分技术在图像处理领域中应用较多,但在高精度的测量测控领域中还需要进一步的研究和发展,CCD细分技术在高精度计量测试领域中的研究和应用是本论文的主要研究方向。本文从CCD基本结构和工作原理的介绍开始。对影响CCD测量精度的各个方面进行了分析,对CCD测量中的照射光源和光学处理系统,CCD视频信号处理与数据采集、CCD工作信号的噪声分析与处理、CCD边缘检测与像点定位等几个方面进行了详细的研究与探讨。同时,还建立了一套高精度微动平台和定位检测装置,通过实验手段来研究和比较各种CCD细分方法和技术,对不同CCD细分算法以及亚像素定位算法等进行了较为详尽的理论研究和实验检测,提出了一系列新的细分方法与技术,并将CCD细分技术应用到高精度CCD测量产品的开发中去。 在本论文研究中,将CCD细分基本分为边缘检测细分和像点定位细分两种研究思路,对这两种思更多还原

【Abstract】 Charge Coupled Devices(CCD) is the new type semiconductor integrated photoelectric device developed at the beginning of the seventies of the 20th century. It was put forward at the first of 1970 by W.S.Boyle and GE.Smith from the American Bell telephone laboratory. In the past thirties years, CCD device and its application technology have been developed rapidly with the developed MOS integrated circuit craft. CCD. device falls into linear CCD and matrix CCD according to its sensitization unit permutation way. No matter linear CCD or matrix CCD, the precision of CCD image cell cannot be made very high for the reasons such as its inherent physical characteristic and working mechanism ,etc, or the restrictions as the structure of the chip and manufacture technology ,etc. In order to ensure CCD image cell having adequate sensitization matrix and prevent the commutative disturbing of close image cell, the image cell area and the space of all the CCD devices cannot be so small, in one micron of ranks generally at present. So CCD is restricted to the application of high accuracy measure field.CCD image measure system is composed of lighting system, testee, optic imaging system, signal processing circuit and computer. So, the factors influencing CCD measure accuracy is: the manufacture error of CCD image cell, CCD resolution ratio to the illumination, the accuracy of A/D transforming, the unstability of lighting system, additional noise of CCD driver circuit, focusing error of the imaging system(for instance the sphere error, difference, etc.), the influence of external environment(such as temperature, oscillate, etc.), the image processing level and the goodness or badness of software algorithms, etc. In order to improving the CCD measure level, the high resolution CCD device and high sampling frequency image collecting board is adopted, or the optic enlarging system carried on and the lighting with special lighting source used. But the adoption of these method can be commonly restricted and influenced to some certain degree, such as the precision of CCD image cell not so higher, the imaging quality declined after adopting optic enlarging system, CCD imaging quality affected by the stability and equality oflighting source. But it is simple and effective to improve CCD measure precision by using CCD image processing and soft algorithms. The CCD subdivision subject is put forward and the relative research of CCD subdivision technology and application started at the same time in order to improve the measure precision of CCD application system and meet the demand of CCD application on high precision measure field.With the beginning of introduction on CCD basic structure and working principle in the paper, the various aspects that influence CCD measure precision is analyzed, lighting source and optic processing system, CCD video signal processing and data collection, the noise analysis and processing of CCD working signal, CCD edge detection and image spots location, etc. researched and discussed in detail. Meanwhile, a set of high accuracy micrometer platform and position detection installation is build up. Various CCD subdivision method and technology can be researched and compared by the means of experiment. The academic research and experimental detection of the different CCD subdivision algorithms and subpixel position algorithms is carried on in detail, a series of new subdivision method and technology provided, and the CCD subdivision technology put into the application of high precision measure product development.CCD subdivision can be basically divided into two kinds of research scheme on edge detection subdivision and image spots location subdivision in the paper. Edge is the most principle characteristic of CCD image, the analysis and discussion about two schemes is the main research content of CCD subdivision technology. In the research of CCD edge detection subdivision, the method of dynamic probability > modulation and demodulation> spatial fitting functionn polynominal interpolation, etc. are analyzed theoretically, and the mechanism of these method is inferred on the mathematical aspect, and the advantage and insufficiency summarized. The fuzzy imaging method is introduced fundamentally and checked up experimentally.CCD image spots position subdivision method is introduced between matrix CCD and linear CCD. The matrix CCD subdivision position algorithm is the main method of centroid> power-added centroidx Gauss curve area fittings parabola curve area fitting. The linear CCD subdivision location algorithm have the one-dimension linearity compensation interpolation method, centroid interpolation method, ending voltage method, proportional centrality method, centre of gravity method, step-method, etc. These algorithms is to utilize the high relativity of gray degree value between the image elements for object image spots, and construct mathematics model that reflects accurately the object area image elements gray degree value,更多还原