ARTIFICIAL INTELLIGENCE 95 A Context Sensitive Line Finder for Recognition of Polyhedra Yoshialfi Shirai Department of Information Science, Electrotechnicai Laboratory, Chiyoda.ku, Tokyo, Japan Rec_ .nmended by T. Winograd ABSTRACT A program to recognize polyhedra by a context sensitive line finder is presented. The program is based on the strategy of recognizing objects step by step, at each time making use of the previous results. At each stage, the most obvious and simple assumption is made and the assumption is tested. To find a line segment, a range of search is proposed. Once a line segment is found, more of the. line is determined by tracking along it. Whenever a new fact is found, the prggram tries to reinterpret the scene taking the obtained information into consideration. Results of the experenent usin~ an image dissector are satisfactory for scenes containing a few blocks and wedges. Some limitations of the present program and proposals for future develop- ments are described. 1. Introduction We do not know how to make a program to recognize objects visually as well as a human being. One of the shortcomings of many computer programs is, as M;.nsky has pointed out [l, 2], their hierarchical structure. A human may recr ,tize objects in the context of the environment. The environment may be recognized based on his a priori know!edge. The hun~an recognition pro- cedure is well programmed so that the simple obvious parts are recogni=ed first and the recognition proceeds to the more: complicated details b~sed on :he previous results. The work in this paper studies an example of a context sensitive line finder to recognize polyhedra with an image dissector. Most previous works begin by trying to find feature points in an entire scene and make a complete line drawing. If there is not enough difference of light intensity between two adjacent faces, it is not easy to find the edge between them. Some better ways are proposed to overcome this difficulty [3, 4]. It is, however, very difficult to The research reported herein was done at the Intelligence Laboratory of M.I.T. Artificial Intelligence 4 (1973), 95-119 Copyright © 1973 by North-Hol{and Publishing Company