【作者】 井建辉；

【导师】 张振东；

【作者基本信息】 河北工业大学 ， 控制理论与控制工程， 2003， 硕士

【摘要】 计算机视觉是计算机科学和人工智能的一个重要分支。随着科学技术的发展，其应用研究受到越来越广泛的重视。在工业领域，机器视觉正成为一种提高生产效率和保证产品质量的关键技术。对于未来的机器人系统，视觉伺服将是重要且通用的控制模式。本课题的应用背景是机器人施釉系统。其研究成果不仅对当地陶瓷生产的现代化进程有直接推动作用，而且可以促进视觉伺服的应用研究。 首先，本文结合生产的要求，采用基于位置的动态look-and-move结构设计实现视觉伺服系统。在数字图像的获取方面，需要解决图像设备的选择，图像数字化和照明等问题。本文以该视觉系统为平台还进行了以下一些方面的研究。 摄像机定标：摄像机定标是大多数计算机视觉任务的重要前提。设计应用灵活的定标方法具有重要的意义。本文对现有定标技术做了分析和比较，并采用简便而适用的定标方法进行系统定标。 图像的低层次处理：在视觉系统中，没有对图像低层的前期处理，就不可能完成高层的后期理解。考虑计算机运算速度和低成本的要求，结合系统需要，提出更新更快的算法，研究原始图像的滤波、分割等问题。 特征提取：首先通过对图像的预处理把摄像机获取的图像转换为便于特征提取的图像。选取合适的图像预处理方法，有利于提高系统对反光、阴影等各种噪声的抗干扰能力，从而提高系统对复杂环境的适应性。本文对二值图像采用数学形态学处理，并在实验中得出特征提取的最优策略，获取物体的定位信息。 设计完成视觉伺服任务：通过研究实验，使用VC、Intel的OpenCV等开发工具完成软件系统，实现施釉机器人利用视觉信息进行定位。更多还原

【Abstract】 The computer vision is an important branch of computer science and artificial intelligence. Along with developing of science and technology, the application study of computer vision has been paid to more and more great attention. In the industry domain, machine vision is becoming one kind of critical technology of raise production efficiency and guaranty product quality. As for the coming robot system, visual servoing will be the important and general control model. The application background of this subject is the robot system for spraying ceramic glaze. The research not only can push forward the modernize course of local ceramic production, but also can promote the application study on visual servoing.Firstly, this paper combines the requirement of production, and design the visual servo system with dynamic position-based look-and-move structure. And at the digital image acquiring aspect, it is need to solve questions such as the selection of image equipment, image digitize and illumination etc. Some researches as below have been carried out with the vision system.Camera calibration: Camera calibration is an important prerequisite for most computational vision tasks. Designing a flexible calibration method is a matter of important meaning. This paper has done analyse and compares to the now available calibration technique, and adopts a simple and applicable calibration method for the system.Low-level image processing: If does not do the earlier low-level image processing, we can not accomplish the image understanding in the later period in a vision system. Thinking over the computer calculation speed and the requirement of low cost, and combines the system needs, puts forward the newer and quicker algorithm for studies the questions such as filtering and segmentation of original image.Feature extraction: Firstly, do the image pre-processing to make the video image being easy to feature extraction. Choosing the suitable image pre-processing method is helpful for raising the antijamming ability to various noises such as reflection of light and shadow etc., thus raise the adaptability to the complicated environment of system. In this paper, mathematics morphology is adopted to deal with the binary image, the optimal method of feature extraction and the location information of object was found in experiments.Accomplish the visual servoing task: Using development tools such as VC and OpenCV of Intel etc to accomplish the software system by study and experiment to make the spraying robot realize the position and orientation by vision information.更多还原