三维激光扫描仪中图像处理快速算法设计

论文编号:TX064        论文字数:32041,页数:54

摘要：三维激光彩色扫描仪中的三维空间信息是从二维图像序列中得到的。因此，必须在扫描图像中准确检测出激光条纹的位置，这样才能根据该位置计算出空间信息。但是由于物体表面形状、反射特性比较复杂，而传统的确定结构光条纹中心位置的方法极值法、门限法、重心法、高斯拟合法等都存在不足，这就使得结构光条纹中心位置的准确检测成为影响三维信息获取精度的关键问题之一。文章提出了利用可变方向模板检测结构光条纹中心的方法，实验结果表明该方法能够准确地获取光条纹中心。又讨论了常用的各种光刀提取法的基础上，提出了一种光刀中心自适应阈值提取法。该方法结合了极值法、阈值法和重心法，避开了它们各自的缺点，消除了噪声干扰，从而实现了光刀的精确定位。该方法具有稳定性好、精度高、速度快等特点。同时又提出了基于自适应阈值的结构光条纹中心的方向模板快速检测方法。该算法是把自适应阈值法与可变方向模板法相结合，充分发挥了两者的优点。实验表明这种方法具有稳定性好、精度高、计算简单、数据存储量小、实现速度快的特点，而且还有抗白噪声能力和修补断线的能力，能较好地保留结构光条纹的细节信息。另外，在本文中，运用Matlab软件编程环境，设计并实现了快速图像处理算法。完成了对图像的背景分割，人头的正侧轮廓线的提取，图像中心线的彩色信息的提取，激光标志线的提取。
关键字：图像处理，方向模板，Matlab，背景分割，激光标志线

ABSTRACT: The 3D space information in the 3D color scanner is obtained form the 2D image list. So, the light stripe’s location should be detected correctly. By this, we can obtain the space information, however, because of the body sharp and reflex characters are more complicated, and the traditional way which used for detecting the center of structured light have some weakness. So the accuracy of structured light methods hinges on the ability to locate the “center” of the 3D information. New method presented in this paper starts by convolving a given image with masks in a selected number of directions. Results that demonstrate the validity of the method are presented. Discusses several common methods used to acquire light-knife center, and analyzes their defects .A new adaptive threshold method is proposed ,which combines the extrem method with the threshold and the barycenter method .The new method eliminates the disadvantages of the previous methods such as the noisy and realizes the precise orientation .The method is characterized by good stability ,high precision and high speed ,etc. Meanwhile, this thesis also proposes an adaptive threshold method based on alterable direction template for fast detecting the center of structured light stripe. The combination of the adaptive threshold method and the alterable direction template one made it superior and gave full play to the two methods. Experiments show that the method possessed the advantages of good stability, high precision, easy computing, less data throughput and fast implementation, etc. It can eliminate white noise and repair broken line and remain detail information of structured light stripe. Furthermore, we also propose the application of  Matlab software environment,  design and realize the fast algorithm for image processing. And we accomplish the background division of the image,the distill of the obverse profile, the distill of the colour information of image centerline and the extraction of the central line of laser stripe.
Keywords: Image Processing, Direction Template, Matlab, Background division， The central line of laser stripe

目   录
摘  要   
Abstract  
第1章 绪 论   1
1.1图像处理概述   1
1.2数字图像处理基本概念   2
1.3数字图像处理特点  2
1.4数字图像处理的研究范畴   3
1.5数字图像的处理方法   3
1.6数字图像处理的应用   4
第2章 3DLCS中空间信息的获取  6
2.1引言  6
2.2 三维信息获取技术  7
2.2.1三维获取技术综合  7
2.2.2三维信息获取方法  8
2.3 三维表面颜色信息的获取   12
第3章 课题总体方案设计与方案论证  15
3.1 课题的主要内容与要求   15
3.2 课题指导思想  15 
3.3 课题总体方案设计  16
3.3.1基于简单蓝背景下的目标分割   16
3.3.2激光标志线的提取   18
第4章 算法实现与实验结果分析   26
4.1 快速图像处理的主要算法 26
4.1.1需要完成的功能   26
4.1.2设计工具的介绍   26
4.1.3各个模块的介绍  27
4.1.3.1 蓝色背景下的简单图像分割模块   27
4.1.3.2 激光标志线的提取模块   29
4.1.3.3 提取图像垂直中心线处的彩色信息模块   33
结论与总结  34
参考文献   36
致谢   37
附录A 图象背景分割的程序清单   38
附录B 提取激光标志线的程序清单   41
附录C 提取中心彩色信息线的程序清单   44
附录D 英文翻译   45