Parsing graphic function sequences
K. Kohima, Brad A. Myers · 2002
With the rapid progress of desktop publishing, documents containing both texts and pictures can be easily composed on a computer display. In terms of the internal representation, pictures included in visual documents are usually represented as a sequence of graphic function calls. The paper describes an analysis technique for pictures which are expressed by means of graphic function sequences. This technique has a wide application area such as compilation of visual languages, information retrieval from visual documents, and so forth. The authors introduce graphic functional grammars in order to specify syntactic structures of pictures. Graphic functional grammars are based on definite clause grammar, which has been proposed for describing the syntax of natural languages. A vocabulary consists of not just symbols but graphic functions. A power of describing sentence structures are also enhanced by allowing to add constraints to each production rule. Constraints are written as arithmetic or logical equations and inequations among formal parameters of graphic functions. An efficient parsing algorithm for graphic functional grammars is presented. A novel feature of the parsing algorithm is its order-free property. It can accept any input order of graphic functions as far as they represent a syntactically correct picture. A way of handling graphic functions with side-effect is also discussed.>