Non-commutative Logic for Hand-Written Character Modeling

Abstract

We have proposed a structural approach for on-line handwritten character recognition. Models of characters are writer-dependent. They are codified with the help of graphic primitives and represented in a data base. The originality of our approach comes from the ability for the system, to explain and justify its own choice, and to deal with all different writing systems, such as the Latin alphabet, or the Chinese or Japanese scrip for example, providing that an appropriate data base has been built up. For this reason, our recognizer can be very helpful for learners of “exotic” scripts. In this paper, we propose to analyse the recognition process in an appropriate logical framework, given by non-commutative Logic. We first point out the class of sequents which allows us to describe accurately the recognition process in terms of proofs, then, we will give some results about the complexity of the recognition problem depending on the expressive power of the representation language.KeywordsLinear LogicCharacter RecognitionDistance ComputingProofsTopicsFoundations and Complexity of Symbolic ComputationLogic and Symbolic Computing