【作者】 孙宇臣；

【导师】 葛宝臻；

【作者基本信息】 天津大学 ， 光学工程， 2005， 博士

【摘要】 三维彩色数字化技术不仅能够得到真实物体的数字化模型,而且可以方便有效地对模型实现编辑、修改、显示、测量、人机交互和特征提取等操作。基于光学三角形原理的激光三维彩色数字化技术集光、机、电、算于一体,实现真实物体到三维彩色数字化模型的转化,具有无接触、速度快、精度高、智能化等优点,在产品设计与制造、虚拟现实、逆向工程、人体工程学等领域具有重要的应用价值和广阔的发展前景。本文在分析国内外研究现状的基础上,围绕三维彩色数字化技术,分别在理论和实验上对三维彩色数字化系统的设计、系统标定和数据预处理等关键技术进行了研究,详细给出了三维彩色数字化理论、系统标定理论和数据预处理算法,确定了传感器的结构参数,设计和加工了细丝标定靶和原理性实验系统。以此为基础,利用标定靶对系统进行了标定实验,并对服装模型和头像模型进行了三维彩色数字化实验,对获取数据按所述理论和算法进行了处理,得到了预期的实验结果,验证了所提出理论和算法的正确性。本工作的创新点主要如下:1.提出采用相互独立的多轴同步扫描系统和彩色传感器系统实现物体三维彩色数字化的方案,三维系统和彩色系统相互独立,提高了系统的灵活性。2.提出三维彩色传感器的线性分区和BP神经网络组合的标定方法,简化了标定过程,实现了三维彩色数字化模型空间与真实物体空间的准确映射。3.设计了能够实现空间共面和非共面标定点获取的细丝标定靶,可以同时实现多个传感器的标定、三维坐标信息与彩色信息的匹配和多传感器数据的自动匹配。4.提出采用直接处理彩色传感器获取的二维彩色图像实现三维彩色数据颜色边界提取的方案,简化了边界提取过程,快速准确地获得了颜色边界约束条件,避免了在数据简化时彩色信息的失真;并提出采用分组筛选法实现数据中杂散点的删除。5.设计了三维彩色数字化原理性实验系统,进行了服装模型和头像模型的三维彩色数字化实验,并对数据进行了处理,实验结果验证了所提出理论和算法的正确性。更多还原

【Abstract】 3D and color digitization technique can not only obtain the digitized models ofreal objects, but also carry out the modification, edit, display, measurement,interaction and feature extraction of the models easily and efficiently. Based onoptical triangulation principle, laser 3D and color digitization technique integratesoptics, mechanics, electronics and algorithms, which fulfill the transformation fromreal objects to 3D and color digitized models. It is a technique that has many specialcharacteristics such as non-contact, high efficiency, good precision and highintelligence, which enable the technique to have important application value andextensive development prospect in many fields such as product design andmanufacture, virtual reality, reverse engineering, human engineering and so on.Based on the detailed analysis of present research status on 3D and colordigitization technique, theoretical and experimental studies have been carried out onsystem design, system calibration and data pre-processing. 3D and color digitizationtheory, system calibration theory and data pre-processing algorithms are presented indetail. Structure parameters of the sensor are calculated, and a filament target and aprinciple experiment system are designed and set up. Then, calibration experimentsand measurements of a clothing model and a head model have been made, andexpected results are obtained after data processing procedure, which verify thecorrectness of the proposed theories and algorithms.Here are the main innovative ideas of this paper:1. A scheme is proposed to fulfill the 3D and color digitization of object withmulti-axis-synchro-scanning system and color sensor system, the two systems workindependently, which enhance the flexibility of the system.2. A calibration method integrating linear partition method and BP neuralnetwork method is proposed to calibrate the 3D and color digitization system, whichsimplifies the calibration process, and accurate mapping relationship between spaceof 3D and color digitized models and space of real objects is obtained.3. A filament target is designed to meet the demand of obtaining coplanar ornon-coplanar calibration points, which can fulfill the calibration of the sensors,matching between 3D information and color information, data auto-matching ofsensors simultaneously.4. A scheme is proposed to fulfill the extraction of color edge of 3D color databy processing the 2D color images obtained by color sensors directly, whichsimplifies the edge extraction process and color edge constraint condition isobtained quickly and accurately, color information distortion is avoided during thedata simplification procedure;also, an algorithm based on grouping and deleting isproposed to delete the stray points in the data set.5. A principle experiment system is designed to fulfill the 3D and colordigitization of real objects. 3D and color data of a clothing model and a head modelare obtained by the system and data processing results verify the correctness of theproposed theories and algorithms.更多还原