Engineering Drawings Input And Intelligent Recognition Technology

The EDLRS(Engineering Drawing Inputting and Recognizing System)which automatically converts image drawings into graphics drawings is an important problem solved urgently in the research field of CAD/CAM nowadays. In this paper, on the bases of analyzing typical EDIRS and recognizing algorithm, a new system for interpretation of images of paper-based line drawings was designed. On the bases of analyzing feature run and character of line, a new approach to extract the whole line by tracing black blocks was proposed. The line intersection was decided by vector calculating according to information of the whole line and the mark of line station. The line elements and intervals were analyzed to get line types. It was adequate to generate a topological graph of the drawing in the course of recognizing with line segments and their interconnections. In this graph, the result was adjusted according to relative knowledge. Since a typical engineering drawing contains both strings and graphics; an algorithm has been developed to locate and separate text string of various font size, style, and orientation. This algorithm based on three features: concentrated strokes, shape and size, and isolation. Expansion and erosion algorithm is first used to strengthen those features, and other algorithms like sliding frame algorithm are used to find the locations of characters based on the characteristic features. Then, it is easy to delete them with the known size and shape. - Dimension, hatching and filling patterns were also identified. Furthermore, a communication program between the system and AutoCAD was designed, which takes advantage of ARX (AutoCAD Runtime eXtension) developing environment. The results demonstrate the effective of these methods.