高速扫描三维测量系统宽动态自适应成像方法分析-analysis of wide dynamic adaptive imaging method for high-speed scanning three-dimensional measurement system.docx

摘 要光学视觉三维测量方法虽已被广泛研究，但对非漫反射物体表面的三维测量 仍然是最具挑战性的问题，原因在于高亮或镜面反射导致采集的图像无法提取结 构光特征信息，这一难题限制了三维视觉测量系统在工业现场的大规模应用。测 量速度、测量精度和解析度是三维视觉测量系统需要平衡的矛盾体；为了使激光 扫描三维视觉测量达到高速度、高精度和高解析度增强环境适应性的目的。本论 文提出基于高动态(High Dynamic Range, HDR)成像的三维视觉测量系统及测量 模型，在光学空间实现高动态成像系统，扩展成像系统的动态范围，把高动态图 像作为三维测量图像处理软件的输入，用于特征提取进行三维坐标的解算，提高 测量精度和解析度，从根源上解决因采集的图像质量影响视觉测量精度问题。论文研究的主要内容有：(1)分析了成像系统的动态范围对图像的质量影响，研究在光学空间扩展成 像动态范围的原理及方法，提出用光学衰减器扩展三维视觉测量成像系统的动态 范围。(2)选择高解析度反射式硅基液晶(Liquid Crystal on Silicon, LCoS)作为光学 衰减器，设计基于 LCoS 和图像传感器(Charge Coupled Device, CCD)相结合的像 素级硬件亮度调整器，实现三维视觉高动态成像测量。(3)简化摄像机直接线性变换模型实现 LCoS 像素和 CCD 像素之亮度像素级 调节算法。研究了 LCoS 亮度衰减值和 CCD 图像亮度值的函数映射关系及其标 定方法。(4)提出图像亮度过饱和区域(Saturation Region Adjustment, SRA)模板代替 LCoS 模板，避免在 HDR 图像恢复重构过程中进行多次数学变换；SRA 模板和 CCD 图像直接实现 HDR 图像的恢复重构。(5)提出采用卡尔曼滤波和 PID 模糊控制器相结合的方法进行高动态成像自 适应控制，初步实现高动态成像采集图像亮度自适应调节。关键词： 高速扫描 视觉测量 结构光 高动态图像 自适应成像ABSTRACTStructured light 3D laser scanning has been studied over the past several decades and is widely used in many industrial fields for dimensional analysis, real-time inspection, component quality control and solid modelling. However, the measurement of a 3D profile of a high dynamic range (HDR) brightness surface such as a partially highlighted object or a partial specular reflection remains one of the most challenging problems. This difficulty has limited the adoption of such scanner systems. The measuring speed, accuracy and resolution are a contradiction cube for the traditional 3D vision measurement method. In order to improve the speed, accuacy and resolution of the 3D vision measurement system to enhance environmental adaptability, a novel measuring priciple and mathematical model with high dynamic range imaging have been proposed in this paper. The dynamic range of imaging subsystem is expended by combination of optical attenuator (LCoS) and solid-state image sensor (CCD or CMOS) in optical space. The brightness adaptive imaging was implemented by an auto-adjust algorithm making the brightness to change in response to t