中文摘要：

计算机视觉技术的发展是许多自动化生产系统的前提与保证，机械零件的自动识别和分拣是生产自动化和智能化的重要手段之一。然而，目前提出的大多数方法都围绕零件的类别检测进行，并无零件方位朝向自动分拣输出这一应用研究。针对这一问题，设计了一套基于计算机视觉的零件方位朝向自动分拣系统，系统分为视频采集与图像转换系统、图像处理系统和分拣输出执行机构三部分；集图像采集、处理和输出于一体，实现各类零件方位朝向的自动识别和实时分类输出。实验结果表明，所提出的方法是可行和有效的。

英文摘要：

The development of computer vision technology is a prerequisite and guarantee of many automated production systems. Automatic identification and sorting of mechanical parts is one of the important means of production automation and intelligence. However, most methods proposed currently focus on category detection of mechanical parts. There is no applied research on automatic sorting and output of orientation of mechanical parts. To solve this problem, a set of automatic sorting system of parts orientation based on computer vision is designed. The design consists of three parts： video capture and image conversion system, image processing system, sorting output executing agency. It integrates image capture, handling and output and achieves automatic identification and real-time classification and outpour of orientation of various parts. The results indicate the method proposed is feasible and effective.