节点文献

基于机器视觉的图像采集与处理系统研究

【作者】 张杰

【导师】 李灿平;

【作者基本信息】 成都理工大学 , 计算机应用技术, 2006, 硕士

【摘要】 机器视觉广泛应用于许多领域。本文介绍了一种基于机器视觉的集采集、传输和分析于一体的图像处理系统。并阐述了该系统的图像采集模块和机器视觉分析模块之间连接通信的原理、接口电路和软件处理系统的设计。实践证明这种小型化、价格低廉的图像采集处理系统在工程实际中有重要的应用价值。 本文从机器视觉系统的组成结构出发,介绍了图像采集、图像传输、低层视觉、中层视觉到高层视觉处理的机制。根据机器视觉系统的组成,首先给出了采用单片机和闪存设计的图像采集系统的设计方案。接着为了传输图像采用了基于TCP/IP的传输技术,设计了图像的传输系统,作为机器视觉处理和图像采集系统的连接模块。 在机器视觉软件处理的部分,作为基础的中低层视觉处理,涵盖了多种数字图像处理技术,包括图像的空间变换、直方图处理、滤波和边缘检测等,这些图像的处理对于高层视觉是非常重要的。作为高层视觉应用的重要方向,本文将机器视觉技术引入到快速人脸检测和人脸识别的研究中,论述了采用类Haar特征的分级分类器快速目标检测算法。介绍了一种较为新颖的具有旋转特征的类Haar特征集,以及快速计算特征值的方法。并且使用基于隐马尔可夫模型的人脸识别算法,在人脸定位的基础上进行人脸的识别和匹配。最终设计完成了一个人脸检测识别系统,并且进行了应用测试。 最后,对本文的研究和设计成果进行了总结。

【Abstract】 Machine vision technology is widely applied in many fields. In this paper, a system for acquisition, transferring and analysis of image based on machine vision is described. The communication and interface between image collection system and machine vision - based analysis system, the control and transferring trough firmware, the realization of image collection and realization of application by software was discussed. Proved by experiment, the system has great value in engineering.A typical machine vision system include: image collection system, transfer system, early vision system, mid-level vision system and high-level vision system. The image collection system is designed based on MCU and flash memory. The transfer system on TCP/IP is a connection between image collection system and machine vision system.Early vision and mid-vision is very important, include grayscale processing, linear filter and edge detection etc. In the end, this paper combines machine vision technology into a research for rapid face detection and recognition, and describes rapid object detection algorithm using a boosted cascade of haar-like features and HMM. A set of rotated haar-like features is introduced. These features can also be calculated efficiently. The complete face detection and recognition system are available.Finally, the research achievements of the paper are summarized.

  • 【分类号】TP274.2
  • 【被引频次】7
  • 【下载频次】1062
节点文献中: 

本文链接的文献网络图示:

本文的引文网络