【作者】 李钦；

【导师】 胡洁；

【作者基本信息】 上海交通大学 ， 机械工程， 2016， 硕士

【摘要】 数字病理切片扫描成像系统是临床病理检测、新药研发、生命科学探索、生物工程与植物生态监测等应用领域在微观层次研究的重要硬件基础。目前商用设备由于对焦原理采用传统被动对焦方式,导致实现病理切片扫描成像过程中,载物台及Z轴对焦装置需多次启停切换对焦及成像位置,且对焦精度较差,无法满足现代医学、生物学大批量切片的需求。本研究基于上述问题,提出了基于倾斜像面的预对焦连续线扫成像技术和宏微双驱动技术,研究了以上技术在实现切片扫描成像过程中“边扫描边成像”的效果的关键技术。本论文详述了基于倾斜像面的预对焦及连续线扫描成像系统的系统构成和工作原理,对核心的高分辨率大像场成像光路和大量程高分辨率自动对焦光路进行了设计和参数优化,并使用Zemax仿真软件验证了两个主要镜筒透镜的分辨率等重要参考指标。为实现系统预对焦的功能,本论文提出了基于人眼视觉特性的小波分解清晰度评价函数,实验表明该方法的综合评价指标相对于常见图像清晰度评价函数有明显优势;设计了基于一致性矢量的多窗口选择算法,可以避免对焦窗口定位于切片污染物处,干扰对焦的准确性。本论文还分别对由步进电机驱动,齿轮减速器、丝杠螺母副两级传动的宏动系统和压电陶瓷驱动的微动系统进行了数学建模和系统辨识,采用基于BP网络PID的控制方法控制微动系统,并针对宏微双驱动系统的宏动系统进行了PI补偿,从而提高了宏微驱动系统的控制精度。文章最后介绍了倾斜像面的预对焦及连续线扫描成像系统的硬件和软件设计,对上位机软件功能模块进行了设计和编写。更多还原

【Abstract】 Digital pathological scanner is the necessary hardware equipment,used for the microscopic research on pathological diagnosis,drug development,biological sciences,bioengineering and ecological environment supervising.At present,since commercial equipments are usually rely on the image-based passive focus principle and the servo motors as the power source,during the scanning process,the X-Y working stage and Z-axis autofocusing stage needs corresponding starts and stops to switch the field of view or change the defocusing value.And as a consequence,the consuming time and focusing accuracy can’t satisfy the scanning requirements from the modern medicine and biology.This research concentrates on a new solution which mainly combines the bevel image plane based pre-focusing and continuous linear scanning imaging technology with the macro-micro dual-drive technology.This paper dilates the system construction and the operating principle of the bevel plane based pre-focusing and scanning system in detail,and then designs the imaging light path with high resolution and wild image field and the autofocusing light path with high-resolution in wild range,then optimizes the lens parameters using Zemax.To realize the pre-autofocusing function,a image definition evaluating function combining the characters of the human vision and the wavelet decomposition method,which is demonstrated much better than traditional ones;on the other hands,it also proposes a consistency vector based multi-windows selection method,which can avoid locating the window locating in the pollution of the section.Then this research also models and identifies the macro-moving system composed of a servo motor,gear reducer and screw-nut pairs,and the BP-PID controller is used to solve the micro-moving system’s control problem,followed by a PI compensation to the macro-moving system.At the end,this paper describes the hardware design and the upper-computer software developing process.更多还原