The objective of this paper is to introduce a novel framework for forms processing that provides seamless processing of two different kinds of formats: physical formats, whose fields have rigidly defined positions and sizes, and topological formats, in which variations in the positions and sizes of fields are acceptable as long the topological relations between pairs of fields are preserved. A line-shared adjacent (LSA) cell relation and an LSA format are introduced to define topological formats, and are then enhanced to describe physical information as enhanced LSA (e-LSA) and an enhanced line oriented (e-LO) format. The e-LSA format has good flexibility for defining nor only physical and topological formats but also hybrid formats, on which our framework is based. It has characteristics of both physical and topological formats, and enables the framework to handle the two kinds of information seamlessly. The framework consists of four modules: a format generator; a format converter; a format class manager; and a form processor which perform all the processes for field detection on which our research focuses. In this paper the way in which these modules collaborate is illustrated with some examples, which confirms the effectiveness of our framework.
展开▼
