A topology-based component extractor for understanding electronic circuit diagrams

Abstract

An automatic understanding system using the techniques of image processing, pattern recognition, and artificial intelligence has been developed for electronic circuit diagrams. Part of the system is presented to extract three categories of essential components: circuit symbols, characters, and connection lines. Each essential component consists of a set of picture segments which are appropriately detected by a segment tracking algorithm. A heuristic piecewise linear approximation algorithm is proposed to approximate picture segments for primitive recognition. On the basis of topological context, a one-pass manner called the relational best search method applies a depth first search technique uniting a set of specified rules during the traversal of a circuit diagram. This method combines the constituents of each circuit symbol or character into a cluster. All the clusters together with the remaining components are extracted and grouped into the three categories as soon as the traversal is finished. A variety of electronic circuit diagrams have been used for testing the component extractor. So far, the present extractor has shown favorable results.