Springer, 2000. — 196 p.This book is concerned with the theory and practice of computer interpretation of images of line drawings. At present, many companies, agencies and individuals throughout the world regularly invest significant financial and human resource in the input of information into computers. Much of the information involved is presented in the form of line drawings. Manual input of those drawings to, for example, Computer-aided Design (CAD) or Geographical Information System (GIS) is laborious and time-consuming task. Line drawings are often large, containing many narrow, closely spaced line segments, each of which must be separately identified the computer-based representation. Manual input generally involves tracing over the drawing with an appropriate pointing device, an operation that produces at best variable and often erroneous results. The availability of automatic and/or easily usable, interactive line drawing input systems would make this important process both easier and cheaper. The technological infrastructure needed to support the development of such systems is already in place. Equipment capable of efficiently scanning a document to produce a digital image is now widely available, as is software capable of basic image manipulation, storage and display. The extraction from the resulting image of the representations required by target applications such as CAD and GISs is, however, a complex problem that remains an active area of research and development. Line drawings arise in a wide variety of disciplines and situations. To provide focus for discussion, we will deal primarily with the transformation of images of engineering drawings and maps into formats required by mechanical/manufacturing engineers and cartographers respectively. Many of the issues discussed and techniques described are, however, relevant to the interpretation of other types of line drawing. One might wish to consider, for example, images of floor plans, electrical schematics or data flow diagrams. As in these other domains, machine interpretation of maps and engineering drawings involves recognition of quite abstract graphical objects. Engineering drawings might be described in terms of, e.g. the outlines of physical objects, cross-hatched areas, centre lines and dimension sets. Entities of similar complexity (contour lines, roads, pipelines and cables, etc.) must be extracted from images of maps. Identification of these high level drawing constructs requires considerable knowledge of the type of drawing concerned; it is rare for someone involved in manual drawing input to simply copy the document without giving at least some thought to its contents. Current commercial document input systems, however, only begin to exploit prior knowledge of drawing type and typically produce lower level representations than are really needed. Research results must be exploited, and further research is required, if useful interpretations are to be delivered at acceptable levels of speed and accuracy.The Line Drawing Interpretation Problem Components of a Line Drawing Interpretation System Document Image Acquisition Binarisation Binary Image Processing and the Raster to Vector Transformation Analysis of Connected Components Vectorisation Interpreting Images of Maps Recognising Cartographic Objects Recovering Engineering Drawing Entities from Vector Data Knowledge-Directed Interpretation of Engineering Drawings Current Issues and Future Developments
Чтобы скачать этот файл зарегистрируйтесь и/или войдите на сайт используя форму сверху.