Use of positional information in sequence alignment for multiple classifier combination

Abstract

There are problems in pattern recognition where the output of a system is a sequence of classes rather than a single class. A well-known example is handwritten sentence recognition. In order to make those problems amenable to classifier combination techniques, an algorithm for sequence alignment must be provided. The present paper describes such an algorithm. The algorithm extends an earlier method by including information about the location of each pattern in a sequence. The proposed approach is evaluated in the context of a system for handwritten sentence recognition. It is demonstrated through experiments that by the use of positional information the computationally expensive process of multiple sequence alignment can be significantly sped up without loosing recognition accuracy.