Compact —a surface representation scheme

作者:

Highlights:

摘要

This paper describes the continuation of work on extracting simple surfaces from a 3D line-segment representation of stereo image data. The surface parameters form a basis of a scene representation which is suitable for navigation by a mobile robot and also for recognition and manipulation of simple objects by a robot arm. Following earlier work on planes, methods and algorithms for recovering spherical, cylindrical and conical surfaces have been developed. The essence of the approach is testing of small groups of 3D line segments for compatibility with a particular surface type. Maximal sets of segments supporting different surfaces are then identified. In this a novel use of the dual space representation is made.

论文关键词:robot vision,robot navigation,object recognition,surface representation

论文评审过程:Available online 14 August 2003.

论文官网地址:https://doi.org/10.1016/0262-8856(89)90005-X