| Along with the popularization of Computer-Aided Design (CAD), it has been becoming increasingly necessary and desirable for a computer to recognize engineering drawings and diagrams. Methods exist for inputting and recognizing such engineering drawings and diagrams. This is primarily because they are drawn to conform to specific standards. In contrast, architectural drawings are not prepared in accordance to existing standards. Hence, the problem of reading, recognizing, and extracting morphological information from them automatically remains unsolved. It is this problem that this study focuses on.;The research undertaken by this author has three distinct but interrelated objectives. The first objective is to design, implement, and test a computer-based framework which allows its user to extract automatically the geometric and/or architectural structures of a two-dimensional plan. The second objective entails designing, implementing, and testing a computer-based framework which may be employed to compare the geometric and/or architectural structures of individual plans or classes of such plans. The third objective is to integrate the two aforementioned frameworks. Computer vision techniques are used to investigate, analyze, and compare plans of buildings from a morphological standpoint. Such techniques can contribute toward detecting differences or similarities between individual plans. Their ability to search for, combine, and compare morphological information is both parsimonious and effective.;Predicated on the assumption that designers derive knowledge from past solutions to form-making problems, this study focuses on the methods by which the morphological information which is contained in building plans can be extracted automatically and entered in a knowledge base. Conceptually, this is part of a larger project which entails investigating how knowledge can be incorporated in a CAD system in a manner which aids and supports the form-making process. Conceivably, the approach of this work is, wholly or partially, applicable to the problem of extracting useful information from graphic representations used in a variety of disciplines (e.g., engineering). |
