用多距离特征识别三维物体的表面形状

本文讨论一种识别物体的方法。三维空间中的物体表面可以由部分球面、柱面和平面来近似,并且这些面是能够被识别的。处理方法是通过分割图象成为区域(Region)且并行处理这些区域来实现。文内叙述一种技术。这种技术应用具有激光扫描器的摄象机系统获取图象。测量三维物体相应表面点的座标。然后用相应区域中的三点决定一个参考平面,进而计算区域中的特征点到参考平面的距离。这些多距离特征可被用来识别相应区域中的物体表面是球面、柱面还是平面。这些面的方向,可用具有两个斜角的表面法线方向来描述。这种技术在自动装配线上和工业监视检验中是很有用的。可用其找出另件在三维空间中的位置和方向,计算检验另件表面形状外形尺寸等等。

Recognition for 3-D Surface Shape Using Multiple Distance Features

In this paper an approach to object recognition is discussed in which surfaces of objecta in 3-D space are approximated by pieces of spheres, cylinders and planes. This processing is realized by partitioning an image into regions which are processed in parallel. This technique uses a recorded image by a camera with a laser scanner to measure coordinates of 3-D surface points. Three of the measured points in a region are chosed to determine a plane and distances from feature points in the region to the plane are calculated. The multiple distance features can be used to recognize whether the surface in a region is a piece of a sphere, a piece of a cylinder, or a piece of a plane. The surface orientations can be discribed by surface normal which hag two oblique angles. This technique is useful for many purposes such as finding the three-dimensional positions and orientations of parts and calculating their surface shape properties which provide informations for industrial inspection.