摘要对一个自主式机器人来说，其自主性与智能性在很大程度上依赖于对内外环 境的感知。而视觉系统可以为机器人获取大量的外界信息，所以机器视觉领域现 已成为国内外专家们的研究热点之一。视觉系统所利用的视觉传感器主要分为两 种，一种是采集 2。5D 点云图像数据的 3D 传感器，另一种是工业相机或摄像头。 第二种方式在经济成本上具有巨大优势，受到广大研究人员的重点研究与开发， 已经在各领域得到了广泛的应用。84804

本文以微软公司的 Microsoft Visual Studio 2012（简称 VS2012）为集成开发 环境（IDE），依据开源视觉库 OpenCV2。4。9，对机器视觉系统中的视觉识别算法 进行了研究。针对图像预处理中的噪声的消除、目标检测、识别与跟踪中的检测、 跟踪等具体步骤的各种算法进行对比分析，选择其中更为合适的算法，尽可能减 少算法运行所需要的时间，在实现对运动目标的识别与跟踪的情况下，保证了系 统的实时性。本文在最后阶段对整个机器视觉识别算法做了大量实验，获得了大 量实验现象与数据，并对这些实验现象与数据进行了全面分析，证明了本算法的 实时性与有效性。

毕业论文关键词：机器视觉；视觉识别；OpenCV；图像处理

Abstract For an autonomous robot, its autonomy and intelligence is largely dependent on the perception of the external environment。 The vision system can obtain a lot of info- rmation about the outside world for the robot, the machine vision experts at home and abroad has become one of the hotspots。 Vision systems use vision sensors are pided into two types, one is to collect 2。5D 3D point cloud image sensor data, and the other one is industrial cameras or camera。 The second approach has great advantages in ec- onomic costs, focusing on research and development by the majority of researchers, it has been widely used in various fields。

In this paper, Microsoft's Microsoft Visual Studio 2012 (referred to VS2012) for the integrated development environment (IDE), based on open source vision library OpenCV2。4。9, machine vision systems in visual recognition algorithm is studied。 Pe- rformed for a variety of image pre-processing algorithms in noise cancellation, target detection, recognition and tracking in the detection, tracking and other specific steps comparative analysis, choose a more suitable algorithm to minimize the time required to run the algorithm, in the case of movement to achieve the goal of identifying and tracking to ensure that the system in real time。 In this paper, in the final stage of the entire machine vision recognition algorithm experiment a lot, we get a lot of experim- ental phenomena and data, and these experimental phenomena and a comprehensive analysis of data, real-time and proved the effectiveness of the algorithm。

Keywords: Machine vision; visual recognition; OpenCV; image processing

目 录

第一章绪论 5

1。1 自主式机器人… 5

1。2 机器人视觉… 5

1。2。1 机器人视觉研究意义与背景… 5

1。2。2 机器人视觉国内外研究现状… 7

1。2。3 机器人视觉实际应用分析… 8

1。2。4 机器人视觉系统的组成… 9

1。3 课题内容及章节安排… 10

第二章 开发平台的搭建 11

2。1 OpenCV 2。4。9 的介绍…