Online parsing of visual languages using adjacency grammars

Visual computing environments continue to grow in importance, yet fast, general parsing algorithms for visual languages remain elusive. In this paper, we present an incremental parsing algorithm for a broad class of visual languages which do not contain overlapping elements. Our algorithm is based on the concept of adjacency grammars, where adjacencies are defined so as to encompass both spatial and logical constraints. Our approach combines bottom-up and top-down methods to support incremental parsing of visual input, allowing for measurably efficient online parsing of diagram-like visual languages, with observed linear run-times for large visual sentences.