本文描述一个识别工业物体位置和取向的先进计算机视觉系统。工业自动化的一个主要问题就变成:这类系统怎样能自动适应新的物体、这些物体表面特性的变化、照明条件和摄象机位置的变更。这种自适应过程目前很耗费时间,并需要专家的帮助(而专家的人数稀少且费用高昂)。在本文的新方法中,类似专家系统那样采用人工智能手段后,部分的专家知识已经自动地进入了系统。 This article describes an advanced computer vision system that identifies the location and orientation of industrial objects. One of the major problems in industrial automation has become: how these systems adapt automatically to new objects, changes in the surface properties of these objects, lighting conditions, and camera position changes. This process of adaptation is currently time-consuming and requires the assistance of experts (while the number of experts is scarce and expensive). In the new method of this article, part of the expert knowledge has automatically entered the system after the introduction of artificial intelligence, similar to expert systems.