Recognition of benzene structure from handwritten chemical expression with radial basis function neural network and rule-based approach

Abstract

The chemical symbols and structures are basic building blocks of chemical expressions and reactions. Benzene symbol is widely used in aromatic chemical reactions. In this paper, we attempt to build a system which can recognise a benzene symbol from the handwritten chemical expressions, reactions or statements (HCERS). In this work a classifier has been developed. It identifies the parts of an image, which can possibly represent a benzene symbol from HCERS. On the next stage, i.e., recognition, the correct benzene structure is recognised from the identified parts of images. Two approaches, first rule based and second radial basis function neural network (RBFNN) based, have been proposed for the classifier. The scanned image of the handwritten chemical reaction, expressions or statements is input to our system. The output shows the presence of valid benzene ring structure or otherwise in the scanned image.