Classification Method Study for Automatic Form Class Identification

Pierre Héroux (1) , Sébastien Diana (1, 2) , Arnaud Ribert (1) , Eric Trupin (1)

1 PSI - Perception, Systèmes, Information
2 DPCI SA

In this paper we present three classifiers used in automatic forms class identification. A first category of classifier includes the k-Nearest Neighbours (kNN) and the Multi-Layer Perceptron (MLP) classifiers. A second category corresponds to a new structural classifier based on tree comparison. On one hand, a low level information based on a pyramidal decomposition of the document image is used by the kNN and the MLP classifiers. On the other hand, a high level information represents the form content with a hierarchical structure used by the new structural classifier. Experimental results are presented. Some strategies of classifier co-operation are proposed.

Domaines: Traitement du texte et du document

Pierre Héroux, Sébastien Diana, Arnaud Ribert, Eric Trupin. Classification Method Study for Automatic Form Class Identification. International Conference on Pattern Recognition, 1998, Brisbane, Australia. pp.926-928, ⟨10.1109/ICPR.1998.711385⟩. ⟨hal-00601502⟩