【作者】 陈克勇；

【导师】 王辅明；

【作者基本信息】 厦门大学 ， 微电子与固体电子学， 2007， 硕士

【摘要】 电子计算机科学、图像处理、模式识别理论和技术的迅速发展给工业自动化中的机器视觉技术带来了良好的实现条件。近年来作为机器视觉范畴之一的自动光学定位技术日益得到重视,已经成为高技术领域的一项重要研究课题。本文主要研究工业自动光学定位系统的实现,从硬件和软件两个方面详细介绍了自动光学定位系统的设计开发过程和其他一些相关技术。系统硬件部分是根据数字图像信号处理的特点,设计一个基于CMOS传感器采集图像、复杂可编程逻辑门器件(CPLD)控制采集、数据传输和图像存储、高速数字信号处理器(DSP)实现图像识别算法的定位点图像采集和处理系统。其中详细介绍了硬件系统的结构、各块芯片的原理、芯片之间的接口设计、通信和控制以及图像信号在该系统中的流程。该系统作为一种低成本、高精度、便于携带的实时自动光学定位技术的实现平台,只要修改CPLD和DSP的程序便可实现多种功能的算法处理,并将处理结果输出。软件部分是针对定位点图像的特点,开发一种新的定位点识别算法,它基于二值图像的三个几何性质:图形的质心和其几何中心重合;可以通过图形面积来求其半径;定位点图形能被以算法计算的中心为圆心、计算半径为半径的圆囊括。算法由适合图形图像处理的MATLAB语言实现,实验结果分析显示该算法对定位点图像的处理速度快而且定位精度高。根据算法思路采用C语言修改和优化后的程序通过了DSP集成开发环境CCS下的模拟。更多还原

【Abstract】 The machine vision has been carried out by the great development of computer science, image processing technologies and pattern recognition theories. The automatic optical positioning anchor technology which is a part of the machine vision has been regarded as a significant content and becomes an important research quest in high technological field.How to realize the system of industrial automatic optical positioning technology is main work in this article. This system is introduced by two parts: hardware and software, and some other correlative technologies are also related.In the part of hardware, a professional system for gathering and dealing with positioning points based on characters of the digital signal has been designed. It is made up of a CMOS image sensor, complex programmable logic device (CPLD) and digital signal processor (DSP) which takes role as image gathering, transmission controlling and image processing respectively. The configuration of hardware system, principles, interface designing, communication and control of each chip are introduced in this article and the flow of image data in the system is also illuminated here. This system is used as a platform for realizing automatic real-time optical positioning technology because of its low-cost, high precision and easy to carry. It can deal with many identification methods by modifying programs in CPLD and DSP.In the part of software, a new method for identifying positioning points is proposed according to characters of binary images. It is based on the three geometric properties of the circle: the centroid overlaps with the geometric center, radius can be calculated from the area and area can be enclosed with the calculated radius. The arithmetic is coded by MATLAB which is proper to image processing. Experimental results show that the geometry method gives quite accurate location of the center of circles using very short time. A modified and optimized C program according to the MATLAB arithmetic can be transported to DSP after simulating in CCS which is the integrated development environment of DSP.更多还原